# Adobe Experience Manager
Source: https://docs.tryprofound.com/agent-analytics/adobe_experience_manager

This documentation explains how to set up Adobe Experience Manager as a Cloud Service (AEMaaCS) Log Forwarding to deliver CDN logs to the Profound Agent Analytics platform.

## Overview

The integration uses Adobe Experience Manager as a Cloud Service's self-serve HTTPS Log Forwarding to deliver CDN logs to our Agent Analytics API. AEM's CDN is Fastly-backed and streams logs continuously over HTTPS to the destination configured in your program. For more information about AEM Log Forwarding, [visit the Adobe documentation](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/developing/log-forwarding).

## Prerequisites

* An AEM as a Cloud Service program with self-serve Log Forwarding enabled (sandbox-tier programs are not supported by Adobe and cannot forward logs)

* Cloud Manager access to commit to the program's Git repository and run a Config Pipeline

* A Profound Log Ingestion Token for Agent Analytics

## Configuration

**1. Step 1**

In Cloud Manager, add your Profound Log Ingestion Token as a secret variable on the environment(s) you want to forward logs from.

Navigate to **Configuration → Environment Variables → Add** and configure:

* **Name** — `PROFOUND_LOG_INGESTION_TOKEN`

* **Type** — `Secret`

* **Service Applied** — `All`

* **Value** — Your provisioned Log Ingestion Token for Profound's Agent Analytics platform (`bot_PROFOUND_LOG_INGESTION_TOKEN`)

**2. Step 2**

Add a `logForwarding.yaml` file under a top-level `config/` directory in your program's Git repository. This is the folder structure required by AEM's Config Pipeline.

```yaml config/logForwarding.yaml theme={null}
kind: "LogForwarding"
version: "1"
metadata:
  envTypes: ["prod"]
data:
  https:
    default:
      enabled: true
      url: "https://artemis.api.tryprofound.com/v1/logs/aem"
      authHeaderName: "X-API-Key"
      authHeaderValue: "${{PROFOUND_LOG_INGESTION_TOKEN}}"
    aem:
      enabled: false
```

`aem.enabled: false` disables forwarding of AEM application, Apache, and Dispatcher log families: only CDN logs are sent to Profound.

`metadata.envTypes` scopes the configuration to specific environment types. Set it to `["prod"]` to forward only production CDN logs, or include `"stage"` and `"dev"` to forward those too. Omit the `metadata` block to apply to every environment the Config Pipeline targets.

**3. Step 3 (optional)**

AEM CDN logs do not include a referer field by default. To populate Profound's `referer` field, add a CDN request transformation that copies the incoming `Referer` header into a custom log property.

Add the following to your program's CDN configuration file (for example, `config/cdn.yaml`):

```yaml config/cdn.yaml theme={null}
kind: "CDN"
version: "1"
data:
  requestTransformations:
    rules:
      - name: profound-log-referer
        when: "*"
        actions:
          - type: set
            logProperty: referer
            value:
              reqHeader: referer
```

> [!NOTE]
>
> The `logProperty` name must be exactly lowercase `referer`. Variants like `Referer` or `referrer` will not be recognized and the field will silently remain empty.

**4. Step 4**

Deploy the configuration with a Cloud Manager **Config Pipeline** (Targeted Deployment).

> [!WARNING]
>
> `logForwarding.yaml` and CDN config files are deployed only by a Config Pipeline. Full Stack and Web Tier pipelines do not deploy these files.

1. In Cloud Manager, go to **Pipelines → Add Pipeline → Targeted Deployment → Config Pipeline**.

2. Point the pipeline at the branch holding `config/logForwarding.yaml` (and `config/cdn.yaml` if you configured one).

3. Target the environment(s) where log forwarding should be active.

4. Run the pipeline.

> [!NOTE]
>
> On first delivery, AEM's underlying Fastly delivery layer issues a one-time HTTP challenge to `/.well-known/fastly/logging/challenge` to validate the destination. Profound's endpoint handles this automatically: no action required on your side. If you inspect Cloud Manager logs and see a single challenge request, this is expected.

> [!TIP]
>
> That's it! AEM will now send CDN logs to Profound. Data should begin appearing in your dashboard within a few minutes.

## What gets captured

AEM CDN log entries are mapped to Profound's visit schema as follows:

* Time and IP

  * `timestamp` — Time the CDN received the request.

  * `cli_ip` — Client IP address.

* Request

  * `host` — Host header from the client request.

  * `url` — Path and query string requested.

  * `method` — HTTP method.

  * `req_ua` — User-Agent header reported by the client.

  * `referer` — Referer header (only populated when the optional CDN request transformation in Step 3 is deployed).

* Response

  * `status` — HTTP response status code.

  * `ttfb` — Time to first byte, in milliseconds.

  * Response body size — Adobe's CDN log schema does not emit a native response-size field, so this is reported as `0`.

## Troubleshooting

* If logs aren't appearing, confirm the pipeline that deployed was a **Config Pipeline**, not Full Stack or Web Tier

* If the pipeline succeeds but no requests reach Profound, the program may still be on Adobe's legacy support-ticket log forwarding. File a ticket with Adobe to migrate to self-serve Log Forwarding before retrying

* If volume looks roughly 2× expected, legacy and self-serve log forwarding may be running in parallel during a migration window (Adobe typically resolves this within a few hours)

* For `401` or `403` responses, verify `PROFOUND_LOG_INGESTION_TOKEN` is set with **Service Applied: All** and that the token is active in Profound

* Confirm the `url` begins with `https://` (AEM rejects the configuration at deploy time otherwise)

* Ensure `aem.enabled: false` is present so non-CDN log families are not forwarded

## Additional Resources

* [Adobe — Log Forwarding for AEM as a Cloud Service](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/developing/log-forwarding)

* [Adobe — CDN log entry format](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/developing/logging#cdn-log)

* [Adobe — Use Config Pipelines](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/operations/config-pipeline)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions

## Security Considerations

* Store the Log Ingestion Token as a Cloud Manager secret variable, never as a plaintext environment value

* Regularly rotate your Log Ingestion Token

* Monitor log forwarding activity for unusual patterns


# Akamai Datastream2 Integration
Source: https://docs.tryprofound.com/agent-analytics/akamai

This documentation explains how to set up Akamai Datastream2 to forward your CDN logs to the Profound Agent Analytics platform.

## Overview

The integration uses Akamai Datastream2 to forward your CDN logs to our Agent Analytics API. Akamai Datastream2 is a service that enables reliable delivery of streaming data to various destinations including HTTP endpoints. For more information about Akamai Datastream2, [visit the Akamai documentation](https://techdocs.akamai.com/datastream2/docs/welcome-datastream2).

## Prerequisites

* An Akamai account with Datastream2 add-on

* Access to your Akamai Datastream2 configuration

* A Profound Log Ingestion Token for Agent Analytics

## Configuration

![Akamai Detected](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/akamai_setup.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=d70903421c141d21d2f4beb6cb900afc)

**1. Step 1**

Sign in to Akamai Control Center and navigate to [DataStream](https://control.akamai.com/apps/data-stream-ui/#/streams/group/all) in the left sidebar. Click `Create a stream`

![DataStream create stream Navigation](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/create_a_stream.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=14c3af737e62235af30ed3f7db30cb07)

**2. Step 2**

Select your property and click `Next`

![DataStream property selection](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/datastream_start.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=2486ffc50acbe122039bc01bdb64076a)

**3. Step 3**

Select the following options in each section:

**Log Information**

![log information](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/log_data.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=4d71519f3395e9515b691d54839133cd)

**Message Exchange Data**

![message data](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/message_data.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=fbe0bc758691c5aaa77de1fc6404c5b8)

**Request Header Data**

![request header data](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/header_data.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=16bfab22e601431817b5d61acfb24f7d)

**Network Data**

![network data](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/network.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=28c02304416dfd93365626ce8cfa3f28)

**Log Format**

![log format](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/log_format.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=76d889d5b9ed3dc9c59b435fe7428019)

Read more about these fields in the [Akamai documentation](https://techdocs.akamai.com/datastream2/docs/data-set-parameters)

**4. Step 4**

Add authentication information:

* From the **Destination** dropdown, select **Custom HTTPS**.

* Select **BASIC** as the **Authentication** type.

* Enter `https://artemis.api.tryprofound.com/v1/logs/akamai` as the **URL**.

* Enter your organization name in the **Username** field.

* Enter your Log Ingestion Token in the **Password** field.

* Check **Send compressed data**.

  ![authentication](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/authentication.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=fd8855bdeebf8f995ba954a7ebe86a33)

**5. Step 5**

Select **JSON** as the **Log format**.

![output format](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/log_format.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=76d889d5b9ed3dc9c59b435fe7428019)

**6. Step 6**

Choose **30** seconds as the push frequency.

![push frequency](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/frequency.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=92de55c3ba5382bce59cac531094a217)

**7. Step 7**

Review the configuration. It should look like the following:

![Review configuration](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/review_configuration.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=620cb181cc50241f6db1a75fe1c60503)

When ready, click `Save stream`.

> [!WARNING]
>
> DataStream2 takes approximately **60 minutes** to start sending logs.
>
> Please wait for **60 minutes** before checking for logs.

**8. Step 8**

After creating the stream, you will now be instructed to activate the stream.

![activate stream](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/activate_stream.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=b3154f5ac1dbc08cec3ea597d1352300)

Click `Proceed to Property Manager` and select the property you created the stream for.

**9. Step 9**

Click **Edit a new version** to add the stream to the property.

Navigate to the **Property Configuration Settings**. Select **Standard property behavior** from the `+ Behavior` dropdown.

![add behavior](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/add_behavior.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=31859c483fef04f5143b991482e11c26)

Search for **DataStream** and click `Insert Behavior`

![datastream behavior](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/datastream_behavior.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=10ece92a22b6a37d5abf9a765a0e66ad)

> [!NOTE]
>
> The behavior will show **DataStream 1**. Ignore this as you will select version 2 later.

**10. Step 10**

Scroll down to the **DataStream** section and select:

* **DataStream2** from the **Stream Version** dropdown.

* Toggle **Enable** to **On**.

* Select the stream you created in the **Stream Name**.

* Select a sample rate. **100%** is recommended.

  ![datastream settings](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/add_stream_to_property.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=2c1d7e35b4578b1d9d5088abca0c9992)

**11. Step 11**

Select **Log Delivery** from the left sidebar.

Enable:

* **Log Host Header**

* **Log Referer Header**

* **Log User-Agent Header**

Leave the other settings as their default values.

![log delivery settings](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/akamai/log_delivery_behaviors.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=e5ab3c16f61a247431ddd4f38a9ad445)

**12. Step 12**

Save the configuration and activate the new version.

> [!TIP]
>
> That's it! Akamai will now send real-time logs to Profound. Data should begin appearing in your dashboard within **60 minutes** of activating the stream.

## Troubleshooting

* Ensure your stream is activated, connected to the property, and that the property has been updated

* Ensure your Log Ingestion Token is correct

## Additional Resources

* [Akamai Datastream2 Documentation](https://techdocs.akamai.com/datastream2/docs/welcome-datastream2)

* [Creating a Stream Documentation](https://techdocs.akamai.com/datastream2/docs/create-stream)

* [Data Set Parameters Documentation](https://techdocs.akamai.com/datastream2/docs/data-set-parameters)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions

## Security Considerations

* Regularly rotate your Log Ingestion Token

* Monitor your log drain usage for unusual patterns or unexpected traffic spikes


# Akamai
Source: https://docs.tryprofound.com/agent-analytics/akamai-s3

Deliver Akamai access logs to Profound through Amazon S3.

## Overview

Send Akamai access logs to your assigned Profound S3 destination to see agent traffic in Agent Analytics. This guide covers CDN request logs delivered through DataStream 2.

> [!NOTE]
>
> Contact [Profound support](mailto:support@tryprofound.com) to confirm availability and arrange your destination settings and credentials before configuring delivery.

For delivery over HTTP, use the [Akamai HTTPS integration](/agent-analytics/akamai).

## Prerequisites

* An Akamai account with DataStream 2 enabled.
* Permission to create and activate streams for your properties.
* Access to Property Manager to configure request header logging and any required DataStream behavior.

## Encryption

This guide uses **SSE-S3 (AES256)** to encrypt your logs at rest with keys managed by Amazon S3. Profound configures encryption on your destination.

## Configuration

**1. Arrange your destination**

Share the domains you want to monitor with Profound support. Support will confirm source readiness and coordinate delivery of your bucket name, folder path, region, and AWS credentials.

Keep the secret access key in your credential manager and enter it directly into Akamai. Avoid including it in support messages or screenshots.

**2. Create a stream**

In Akamai Control Center, open **Analytics & monitor > DataStream** and select **Create stream**. Choose CDN delivery logs, enter a descriptive name, and select the properties that serve your domains.

**3. Select log fields**

Select **JSON** as the log format. Profound processes one JSON object per line with the following fields:

| Field | Used for |
| - | - |
| `reqTimeSec` | Request time |
| `cliIP` | Client address |
| `reqHost` | Requested domain |
| `reqMethod`, `reqPath` | Requested resource |
| `statusCode` | Response status |
| `UA` | Agent identification |

Include `queryStr`, `referer`, and `totalBytes` for query strings, referral information, and response size.

In Property Manager, enable logging for the Host, User-Agent, and Referer headers so the selected fields contain request values. See Akamai's [data set parameters](https://techdocs.akamai.com/datastream2/docs/data-set-parameters) for the corresponding settings.

**4. Configure the S3 destination**

Select **S3** on the **Destination** tab and enter:

| Setting | Value |
| - | - |
| Name | A descriptive name |
| Bucket | The bucket supplied by Profound |
| Folder path | The folder path supplied by Profound |
| Region | `us-west-2` (Oregon) |
| Access key ID | The AWS access key ID arranged with Profound |
| Secret access key | The matching AWS secret access key |

Select **Validate & Save** to check access. Akamai writes a verification file to the destination during validation.

DataStream 2 delivers gzip-compressed files to S3. Keep the default log filenames; Profound expects compressed filenames to end in `.gz`.

See Akamai's [Amazon S3 destination guide](https://techdocs.akamai.com/datastream2/docs/stream-amazon-s3) for details.

**5. Activate delivery**

After Profound confirms readiness, review and activate your stream. Set sampling to **100%** to capture all matching requests.

For Property Manager-dependent streams, enable the **DataStream** behavior, select your DataStream 2 stream, and activate the property version. DataStream-managed streams use the stream configuration directly.

Follow Akamai's [activation steps](https://techdocs.akamai.com/datastream2/docs/quick-start) for your integration type. Activation can take up to an hour.

## Verify ingestion

Generate traffic to your configured domain. Once delivery and processing have completed, open Agent Analytics and select a date range covering those requests to check for expected agent traffic.

If data is missing, share your domain, stream ID, request time, and any delivery error with Profound support.

## Troubleshooting

| Problem | What to check |
| - | - |
| Destination validation fails | Compare the bucket, folder path, region, and credentials with your assigned settings. Ask support to check destination permissions. |
| No logs arrive | Confirm the stream and any required property version are active and receiving traffic. |
| Files arrive but data is missing | Check the JSON field selection, request header logging, and your selected date range. |
| Access fails after a credential change | Update the destination with the replacement credentials and validate again. |

## Additional Resources

* [Akamai: Stream logs to Amazon S3](https://techdocs.akamai.com/datastream2/docs/stream-amazon-s3)
* [Akamai: DataStream 2 quick start](https://techdocs.akamai.com/datastream2/docs/quick-start)
* [Akamai: Log format](https://techdocs.akamai.com/datastream2/docs/log-format)
* [Akamai HTTPS integration](/agent-analytics/akamai)
* [Contact Profound support](mailto:support@tryprofound.com)


# Amazon CloudFront Integration (Data Firehose)
Source: https://docs.tryprofound.com/agent-analytics/aws_firehose_cloudfront

This documentation explains how to set up Amazon CloudFront real-time logs delivery to Profound Agent Analytics platform using Amazon Data Firehose.

## Overview

The integration uses Amazon Data Firehose to forward CloudFront real-time logs to our Agent Analytics API. Amazon Data Firehose is an AWS service that enables reliable delivery of streaming data to various destinations including HTTP endpoints. For more information about CloudFront real-time logs, [visit the AWS documentation](https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/real-time-logs.html).

## Prerequisites

* An AWS Account with CloudFront and Amazon Data Firehose permissions

* Access to your CloudFront distribution configuration

* A Profound Log Ingestion Token for Agent Analytics

## Configuration

![Cloudfront Detected](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/cloudfront_setup.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=5356d4e17071b78c4a18d7916b638f61)

**1. Step 1**

Sign in to the AWS Console and navigate to the Amazon Data Firehose console

![Firehose Console Navigation](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/firehose_console_nav.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=31ed6890bce8b6d3c2ea4b5a097256da)

**2. Step 2**

Create a new delivery stream, select "Direct PUT" as source and "HTTP Endpoint" as destination

![Firehose HTTP Destination](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/firehose_http_destination.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=b854af041135c268be901b808c13fd31)

**3. Step 3**

1. Configure the HTTP endpoint with the following URL format:

```http theme={null}
https://artemis.api.tryprofound.com/v1/logs/aws_data_firehose_cloudfront
```

2. For authentication, provide your Profound Log Ingestion Token as the access key (we recommend using AWS Secrets Manager for secure token storage).

> [!TIP]
>
> If you are using the [AWS Secrets Manager](https://docs.aws.amazon.com/secretsmanager/latest/userguide/intro.html), please use the following [JSON format](https://docs.aws.amazon.com/firehose/latest/dev/secrets-manager-whats-secret.html) when creating the secret:

```json theme={null}
{
  "api_key": "bot_PROFOUND_LOG_INGESTION_TOKEN"
}
```

3. Enable GZIP content encoding.

   ![Firehose Destination Settings](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/firehose_http_destination_settings.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=f39d1fa3ee17e2fd13a63fafad1f60b8)

4. Create a new S3 bucket to store failed delivery logs (required by AWS)

   ![Firehose S3 Destination](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/s3_backup.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=d345ec2b32d1099f661055cd025b2bc0)

**4. Step 4**

Go to your CloudFront distribution and navigate to the "Logging" tab.

![CloudFront Logging Tab](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/cloudfront_logging.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=cb780dbc9c00f2efcc10fab968fbd21e)

Click the "Add" button and select "Kinesis Data Firehose" as the destination. (Kinesis Data Firehose is the legacy name for Amazon Data Firehose)

![CloudFront Logging Add](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/cloudfront_add_kinesis_data_firehose.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=7526bea9c504503d2019e7021784f368)

**5. Step 5**

Now you should be in the "Add standard logging destination" screen. Select the delivery stream you created in Step 3.

![CloudFront Logging Select](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/cloudfront_select_stream.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=30d591f5607b44e2e98bb3b827f45361)

Under "Additional settings - optional", select the following fields:

![CloudFront Logging Additional Settings](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/cloudfront_select.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=885ebac7639ad545f865d39ba7825f7c)

* Time and IP

  * `date` - Date when the request was completed

  * `time` - Time when the request was completed

  * `c-ip` - Client IP address

* Request Details

  * `cs-method` - HTTP request method

  * `x-host-header` - Host header value of the request

  * `cs-uri-stem` - Request URI path

  * `cs-uri-query` - Request query string

  * `cs(User-Agent)` - Client user agent

  * `cs(Referer)` - Request referrer

* Response Details

  * `sc-status` - HTTP response status

  * `sc-bytes` - Response size in bytes

  * `time-taken` - Request processing time

**6. Step 6**

Select `JSON` as the **Output format**.

![Output format](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudfront_firehose/log_format.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=9958635a099e3a3ab88597d0df3fcea8)

Click **Submit** to save the configuration.

> [!TIP]
>
> That's it! CloudFront will now send real-time logs to Data Firehose, which forwards them to Profound. Data should begin appearing in your dashboard within a few minutes.

## Troubleshooting

* If logs aren't appearing, verify your Firehose delivery stream status in CloudWatch

* Check Data Firehose monitoring for delivery errors

* Ensure your Log Ingestion Token is correct

* Verify CloudFront real-time logging is enabled for your distribution

* Check IAM roles have proper permissions for both CloudFront and Firehose

## Additional Resources

* [Amazon CloudFront Real-time Logs Documentation](https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/real-time-logs.html)

* [Amazon Kinesis Data Firehose Documentation](https://docs.aws.amazon.com/firehose/latest/dev/what-is-this-service.html)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions

## Security Considerations

* Store Log Ingestion Tokens in AWS Secrets Manager

* Regularly rotate your Log Ingestion Token

* Monitor CloudWatch logs for unusual patterns

* Enable CloudTrail for API activity monitoring

* Use IAM roles with least privilege access


# Cloudflare
Source: https://docs.tryprofound.com/agent-analytics/cloudflare-s3

Deliver Cloudflare HTTP request logs to Profound through Amazon S3.

## Overview

Send Cloudflare Logpush logs to your assigned Profound S3 destination to see agent traffic in Agent Analytics. This guide covers Cloudflare HTTP request logs and uses SSE-S3 encryption.

> [!NOTE]
>
> Contact [Profound support](mailto:support@tryprofound.com) to receive your S3 bucket name and destination prefix. The destination region is `us-west-2` (US West, Oregon). When Cloudflare asks you to verify ownership, contact support for your ownership token.

For delivery over HTTP, use the [Cloudflare Logpush streaming integration](/agent-analytics/cloudflare_logpush).

## Prerequisites

* Access to Cloudflare HTTP request Logpush.
* Permission to create and manage Logpush jobs for your domain.

## Encryption

Use **SSE-S3 (AES256)** for this Cloudflare integration. Amazon S3 encrypts your stored logs. You do not need to configure a KMS key.

## Configuration

**1. Get your destination settings**

Contact [Profound support](mailto:support@tryprofound.com) with your domain and Cloudflare zone details to receive your S3 bucket name and destination prefix. The destination region is `us-west-2` (US West, Oregon). Use these values exactly as supplied.

Profound configures the source, destination bucket, and write permissions for your organization.

| Setting | Value |
| - | - |
| Bucket | Your assigned Profound bucket |
| Path | Your complete assigned prefix |
| Region | `us-west-2` (US West, Oregon) |
| Encryption | SSE-S3 (AES256) |

Your S3 destination uses this format:

```text theme={null}
s3://<assigned-bucket>/<assigned-prefix>/
```

Use the prefix exactly as provided by Profound.

**2. Open Cloudflare Logpush**

Sign in to the Cloudflare dashboard, select your domain, and open **Investigate → Logpush**. Select **Create a Logpush job**.

![Cloudflare Investigate menu with Logpush selected and the Create a Logpush job button](https://mintcdn.com/profound-37face47/0hJzfTbdqU5C4M5V/images/agent-analytics/cloudflare_s3/logpush-navigation.png?fit=max\&auto=format\&n=0hJzfTbdqU5C4M5V\&q=85\&s=172f44804eab0eaab7d87cc225c8d2f6)

**3. Select Amazon S3**

Choose **Amazon S3** as the destination.

![Cloudflare destination selection with the Amazon S3 option](https://mintcdn.com/profound-37face47/0hJzfTbdqU5C4M5V/images/agent-analytics/cloudflare_s3/select-destination.png?fit=max\&auto=format\&n=0hJzfTbdqU5C4M5V\&q=85\&s=537076e0406c3ed0f6a74e220ea34f54)

**4. Enter your destination details**

1. Enter the bucket name and destination path supplied by Profound.
2. Set **Bucket region** to **US West (Oregon) — `us-west-2`**.
3. Select **My policy requires AWS SSE-S3 AES256 Server Side Encryption**.
4. Select **Continue** to start ownership verification.

![Cloudflare S3 destination form showing bucket, path, region, and the SSE-S3 encryption checkbox](https://mintcdn.com/profound-37face47/0hJzfTbdqU5C4M5V/images/agent-analytics/cloudflare_s3/destination-details.png?fit=max\&auto=format\&n=0hJzfTbdqU5C4M5V\&q=85\&s=8e3bea5c0f18e1dd096f679fcf1504ff)

The screenshot uses placeholder bucket and path values. Replace them with the values supplied by Profound.

Profound handles the bucket policy. If Cloudflare reports an access error, contact support with the source details.

**5. Complete ownership verification**

Cloudflare writes a challenge file to the destination. Ask Profound support for the token associated with your source, then enter it in Cloudflare to verify the destination.

![Cloudflare ownership verification with the token and challenge-file path blurred](https://mintcdn.com/profound-37face47/0hJzfTbdqU5C4M5V/images/agent-analytics/cloudflare_s3/prove-ownership.png?fit=max\&auto=format\&n=0hJzfTbdqU5C4M5V\&q=85\&s=b034854b36798a9ae46abd9e5310ff2e)

If verification fails, confirm that the token and destination belong to the same setup attempt. Ask support for help if you need to repeat verification.

**6. Configure HTTP request logs**

Select the **HTTP requests** dataset. Include the fields below, filter the hostname to your configured domain, and use RFC3339 timestamps. Keep the standard JSON output, with one record per line.

![Cloudflare dataset selection with the HTTP requests option](https://mintcdn.com/profound-37face47/0hJzfTbdqU5C4M5V/images/agent-analytics/cloudflare_s3/select-dataset.png?fit=max\&auto=format\&n=0hJzfTbdqU5C4M5V\&q=85\&s=4bbac60cd1b29d8c6c0c98aac5973120)

| Fields | Used for |
| - | - |
| `ClientIP` | Client address |
| `ClientRequestHost` | Requested domain |
| `ClientRequestMethod`, `ClientRequestURI` | Requested resource |
| `ClientRequestReferer`, `ClientRequestUserAgent` | Referral and agent analysis |
| `EdgeStartTimestamp`, `EdgeEndTimestamp` | Request timing |
| `EdgeResponseStatus`, `EdgeResponseBytes` | Response details |

These are the same record fields used by Profound's Cloudflare streaming integration. To capture all traffic, turn off sampling.

**7. Enable delivery**

Once ownership verification is complete and Profound confirms the source is ready, save and enable your job. Generate traffic to your configured domain and check Cloudflare for delivery errors.

## Verify ingestion

Open Agent Analytics for your domain and select a date range covering the delivered records. Confirm that expected agent traffic appears. Upload success alone does not confirm successful processing.

Logs appear after Cloudflare delivers a batch and Profound processes it. If data is missing, provide support with your domain, Logpush job ID, delivery time, and the S3 file path or error shown by Cloudflare. Do not include credentials.

## Troubleshooting

| Problem | What to check |
| - | - |
| Destination access denied | Confirm the supplied bucket, region, prefix, and SSE-S3 setting. Ask Profound to verify destination permissions. |
| Ownership challenge fails | Confirm the token matches the exact destination and current setup attempt. Request help retrieving or renewing it. |
| Delivery succeeds but data is missing | Check source readiness, `ClientRequestHost`, the field selection, and the selected date range. |
| Traffic counts are incomplete | Review hostname filters and sampling, and allow time for delivery and processing. |

## Additional Resources

* [Cloudflare: Configure Logpush to Amazon S3](https://developers.cloudflare.com/logs/logpush/logpush-job/enable-destinations/aws-s3/)
* [Cloudflare streaming integration](/agent-analytics/cloudflare_logpush)
* [Contact Profound support](mailto:support@tryprofound.com)


# Cloudflare Integration (Logpush)
Source: https://docs.tryprofound.com/agent-analytics/cloudflare_logpush

This documentation explains how to set up and deploy a Cloudflare Worker that captures HTTP request data and forwards it to Profound Agent Analytics platform using Cloudflare Logpush.

## Overview

The integration uses a Cloudflare Logpush that runs as middleware to collect request metadata and send it to our Agent Analytics API. Cloudflare Logpush is a Cloudflare Enterprise feature that allows you to forward Cloudflare logs to a non-Cloudflare destination like Profound. For more information related to Cloudflare Logpush, [visit the documentation site](https://developers.cloudflare.com/logs/about/).

## Prerequisites

* A Cloudflare account with Enterprise plan

* Access to your domain's Cloudflare configuration

* A Profound Log Ingestion Token for Agent Analytics

> [!TIP]
>
> Using Cloudflare Free/Pro plan? Check out our [integration guide using Cloudflare Workers](/agent-analytics/cloudflare_worker).

![Logpush Detected](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_logpush/logpush_setup.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=1f3d1936ae852ac98962c9167a90e126)

## Configuration

**1. Step 1**

Select your domain on Cloudflare Dashboard, go to **Analytics & Logs -> Logpush**

![Cloudflare Logpush Navigation](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_logpush/cloudflare_logpush_nav.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=625b1175af5e6297333789cb96f6da7b)

**2. Step 2**

Create a new Logpush job, select HTTP destination

![Cloudflare HTTP Destination](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_logpush/cloudflare_http_destination.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=ed84b38394cdeec008e1ce996f9115d6)

**3. Step 3**

Use the following API URL format to include your Profound Log Ingestion Token and continue

```http theme={null}
https://artemis.api.tryprofound.com/v1/logs/cloudflare_logpush?header_X-API-Key=bot_PROFOUND_LOG_INGESTION_TOKEN
```

**4. Step 4**

Select `Filtered logs` under **If logs match...**

And filter by the `ClientRequestHost` field to match your domain

> [!TIP]
>
> Ensure you use your domain in the filter. Do not use `www.tryprofound.com`, this is only an example.

![Cloudflare Filter Logs](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_logpush/cloudflare_filter_logs.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=39352a097e7fc3cfd9248d64b684c3ee)

**5. Step 5**

Select HTTP requests dataset, and choose the following fields to send:

* Request

  * `ClientIP` - IP Address of the client.

  * `ClientRequestHost` - Host requested by the client.

  * `ClientRequestMethod` - HTTP method of client request.

  * `ClientRequestReferer` - HTTP request referrer.

  * `ClientRequestURI` - URI requested by the client.

  * `ClientRequestUserAgent` - User agent reported by the client.

* Performance

  * `EdgeStartTimestamp` - Timestamp at which the edge received request from the client.

  * `EdgeEndTimestamp` - Timestamp at which the edge finished sending response to the client.

* Response

  * `EdgeResponseBytes` - Number of bytes returned by the edge to the client.

  * `EdgeResponseStatus` - HTTP status code returned by Cloudflare to the client.

That's it! You have now successfully configured Profound Agent Analytics for your website. Data should begin to populate on your dashboard within an hour.

## Troubleshooting

* If logs aren't appearing, verify your Log Ingestion Token and URL

* Check Workers > Analytics for any execution errors

* Ensure your route pattern matches your domain configuration

* Verify the Worker is receiving requests by checking the Cloudflare dashboard metrics

## Additional Resources

* [Cloudflare Logpush Documentation](https://developers.cloudflare.com/logs/about/)

* [Cloudflare Enable HTTP destination Documentation](https://developers.cloudflare.com/logs/get-started/enable-destinations/http/)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions

## Security Considerations

* Store Log Ingestion Tokens as secrets in production environments

* Regularly rotate Log Ingestion Tokens

* Monitor Worker usage and logs for unusual patterns


# Cloudflare Integration (Worker)
Source: https://docs.tryprofound.com/agent-analytics/cloudflare_worker

Set up and deploy a Cloudflare Worker that captures HTTP request data and forwards it to Profound Agent Analytics.

## Overview

The integration uses a Cloudflare Worker that runs as middleware to collect request metadata and send it to our Agent Analytics API. The Worker captures important request information like IP addresses, user agents, and referrers without affecting the actual request handling.

## Prerequisites

* A Cloudflare account with access to Workers

* Node.js installed on your development machine

* Access to your domain's Cloudflare configuration

* A Profound Log Ingestion Token for Agent Analytics

> [!TIP]
>
> Using Cloudflare Enterprise plan? Check out our [integration guide using Cloudflare Logpush](https://docs.tryprofound.com/agent-analytics/cloudflare_logpush). Logpush is always preferable to a Worker: it is out-of-band and cannot affect live traffic.

> [!WARNING]
>
> A Worker on a `*` route sits in the request path for **every** request to that hostname, so it can affect production traffic. Use the code below as-is, and in particular:
>
> * Never clone the response or read the response body inside the Worker. Buffering bodies can exceed the Worker's 128 MB isolate memory limit, which Cloudflare surfaces to the visitor (or crawler) as [Error 1102 `Worker exceeded resource limits`](https://developers.cloudflare.com/support/troubleshooting/http-status-codes/cloudflare-1xxx-errors/error-1102/).
>
> * Keep the logging call inside `ctx.waitUntil()` and swallow its errors so a slow or failing log request can never turn into a 5xx.
>
> * Keep sensitive paths out of the Worker entirely. The path list in the code below only skips logging; requests still pass through the Worker. To take checkout, cart, and admin paths out of the Worker's path completely, narrow the route pattern itself (or add a higher-priority route that is not bound to the Worker).
>
> Symptoms of a Worker that breaks these rules are intermittent 5xx responses on large or streamed responses, and downstream fallout such as Google Merchant Center / Google Ads disapprovals ("destination not working", "image not accessible") when `AdsBot-Google` hits one of those errors during a crawl.

> [!NOTE]
>
> Running a Shopify storefront? Do not put a Worker in front of it. Use one of the log drain options in our [Shopify guide](https://docs.tryprofound.com/agent-analytics/shopify) instead, which collects logs out-of-band.

![Logpush Detected](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/worker_setup.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=03401e0694f16e2b2076d70eb045e534)

## Implementation guide

**1. Set up your development environment**

Create a new Worker project and install dependencies:

### Create a new Worker project

First, use the Cloudflare Worker CLI to create a new Worker project:

In this example, we're using version `2.37.4` of the Cloudflare Worker CLI. You may use the latest version, but there might be some minor process differences.

```bash theme={null}
npm create cloudflare@2.37.4 -- log-collector
```

After npm launches, you'll be prompted to select a starting point. Select "Hello World" as your starting category.

![Cloudflare Worker Step 1 - Choose Starting Point](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/create-cloudflare-application-step1-1.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=49f53028fff87639f1bbe844d9591ade)

Select the "Hello World" worker template.

![Cloudflare Worker Step 1 - Select Worker Template](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/create-cloudflare-application-step1-2.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=f225e20e3063a3c6dfd647c8d8835e09)

Select the "TypeScript" language.

![Cloudflare Worker Step 1 - Select Language](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/create-cloudflare-application-step1-3.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=96537f92fa73752faa35b98a68351484)

Now the CLI will create a new project with the name `log-collector` and install the necessary dependencies.

Select Git for version control.

![Cloudflare Worker Step 2 - Select Version Control](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/create-cloudflare-application-step2-1.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=7d671bd7a44948a1f6ec8264a89c0a39)

Deploy the application now. The application will be deployed with a Cloudflare development domain and will not affect your production environment.

![Cloudflare Worker Step 3 - Deploy Application](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/create-cloudflare-application-step3-1.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=96bd057ccbecc1d746170780e362b8b7)

Cloudflare CLI will prompt you to login and select the account you want to deploy the application to. Please choose the account your domain is associated with.

Cloudflare should automatically open a browser window with "Hello World" displayed. You can navigate to the project directory once the application is deployed.

```bash theme={null}
cd log-collector
```

**2. Configure your Worker**

Edit your `wrangler.json` file to configure the `PROFOUND_API_URL` environment variable and the route binding:

> [!WARNING]
>
> Replace pattern `example.com/*` with your actual domain. Use your target site URL (usually marketing site). For example, if your marketing site is `https://www.example.com`, you should use `www.example.com/*` as the pattern.
> The `zone_name` should be your canonical domain without the [www](http://www).

> [!TIP]
>
> If you are not sure about the correct configuration, please contact [Profound support](mailto:support@tryprofound.com).

```json wrangler.json theme={null}
{
  "$schema": "node_modules/wrangler/config-schema.json",
  "name": "log-collector",
  "main": "src/index.ts",
  "compatibility_date": "2025-01-29",
  "observability": {
    "enabled": true
  },
  "route": {
    "pattern": "example.com/*",
    "zone_name": "example.com"
  },
  "vars": { "PROFOUND_API_URL": "https://artemis.api.tryprofound.com/v1/logs/cloudflare_worker" }
}
```

Then copy the TypeScript code into `src/index.ts`:

```typescript src/index.ts theme={null}
/**
 * Cloudflare Worker for Log Collection
 *
 * Forwards request metadata to Profound's log collection API. The origin
 * response body is never read or modified, and every error in the logging
 * path is swallowed, so logging cannot alter the response. Failures are
 * logged with enough detail to tell a rejected request apart from a
 * timeout or a connection error.
 */

export interface Env {
    PROFOUND_API_URL: string;
    PROFOUND_LOG_INGESTION_TOKEN: string;
}

// Paths that are not logged. Matched on full path segments, so '/cart' does
// not match '/cartography'.
const EXCLUDED_PATHS = ['/checkout', '/cart', '/admin', '/api'];

// Characters kept from a rejected response body when logging the failure.
const DETAIL_LIMIT = 512;

// Percent-decoded and lowercased so encoded variants such as '/%63heckout'
// and '/checkout%2Fpayment' match the same way the origin routes them.
function normalizePath(pathname: string): string {
    try {
        return decodeURIComponent(pathname).toLowerCase();
    } catch {
        return pathname.toLowerCase();
    }
}

function isExcluded(pathname: string): boolean {
    const path = normalizePath(pathname);
    return EXCLUDED_PATHS.some(
        (excluded) => path === excluded || path.startsWith(`${excluded}/`),
    );
}

export default {
    async fetch(request: Request, env: Env, ctx: ExecutionContext): Promise<Response> {
        const response = await fetch(request);

        const skip = response.status === 101 || isExcluded(new URL(request.url).pathname);

        if (!skip) {
            ctx.waitUntil(
                sendLog(request, response, env).catch((error: unknown) =>
                    console.error(
                        'Failed to send logs:',
                        error instanceof Error ? `${error.name}: ${error.message}` : error,
                    ),
                ),
            );
        }

        return response;
    },
} satisfies ExportedHandler<Env>;

async function sendLog(request: Request, response: Response, env: Env) {
    const requestUrl = new URL(request.url);

    // +2 for ': ' and +2 for '\r\n' per header
    const headerSize = Array.from(response.headers.entries()).reduce(
        (total, [key, value]) => total + key.length + value.length + 4,
        0,
    );

    // Derived from content-length so the body is never buffered. Streamed or
    // chunked responses have no content-length and are reported as headers only.
    // HEAD and bodyless statuses advertise a content-length that is never sent,
    // so their body size is counted as zero.
    const contentLength = Number(response.headers.get('content-length'));
    const transmitsBody =
        request.method !== 'HEAD' && response.status !== 204 && response.status !== 304;
    const bodySize = transmitsBody && Number.isFinite(contentLength) ? contentLength : 0;
    const bytes = headerSize + bodySize;

    const logData = {
        timestamp: Date.now(),
        host: requestUrl.hostname,
        method: request.method,
        pathname: requestUrl.pathname,
        query_params: Object.fromEntries(requestUrl.searchParams),
        ip: request.headers.get('cf-connecting-ip'),
        userAgent: request.headers.get('user-agent'),
        referer: request.headers.get('referer'),
        bytes,
        status: response.status,
    };

    const logResponse = await fetch(env.PROFOUND_API_URL, {
        method: 'POST',
        headers: {
            'Content-Type': 'application/json',
            'X-API-Key': env.PROFOUND_LOG_INGESTION_TOKEN,
        },
        body: JSON.stringify([logData]),
        signal: AbortSignal.timeout(5000),
    });

    // Only the first chunk of a rejected response is read, and only up to
    // DETAIL_LIMIT characters of it are kept, so a misconfigured URL that
    // returns a large body can never fill the isolate. The origin response
    // is still never read.
    if (!logResponse.ok) {
        const reader = logResponse.body?.getReader();
        const chunk = await reader?.read();
        await reader?.cancel();

        const detail = chunk?.value
            ? new TextDecoder().decode(chunk.value).slice(0, DETAIL_LIMIT)
            : '';

        console.error(`Log ingestion rejected the request: ${logResponse.status} ${detail}`);
        return;
    }

    await logResponse.body?.cancel();
}
```

> [!NOTE]
>
> The Worker logs two distinct failures. `Log ingestion rejected the request: <status> <body>` means Profound received the request and refused it. This is usually an authentication or payload problem, and the status and body identify which. `Failed to send logs: <name>: <message>` means the request never completed. `TimeoutError` is the five-second timeout. Other names point to a connection or configuration problem. Both messages go to the Worker's logs. You can read them under Workers & Pages > your Worker > Logs.

> [!NOTE]
>
> `bytes` is derived from the `content-length` response header instead of the response body. Reading the body (for example via `response.clone()` and `blob()`) forces Cloudflare to buffer the whole payload in the Worker isolate, which fails with a 5xx on large or streamed responses. This Worker never reads or modifies the response body. It streams the origin response through unchanged, but it is still in the request path, so origin fetch failures surface as they would without it. `bytes` therefore counts headers plus `content-length`, and only when the response actually transmits a body (not `HEAD`, `204`, or `304`).

**3. Deploy your Worker**

### Log in to Cloudflare

Use the Wrangler CLI to deploy the Worker:

```bash theme={null}
# Login to Cloudflare
npx wrangler login
```

### Configure the Profound Log Ingestion Token

Secrets is a feature of Cloudflare Workers that allows you to store sensitive information like Log Ingestion Tokens in a secure environment.

```bash theme={null}
# Configure the Log Ingestion Token secret
npx wrangler secret put PROFOUND_LOG_INGESTION_TOKEN
```

You will be prompted to enter the Log Ingestion Token. Copy and paste the token and press enter.

![Configure Log Ingestion Token](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/cloudflare_worker/configure-api-key.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=7ea6fb5162663140e753640037802093)

### Deploy the Worker

```bash theme={null}
# Deploy the Worker
npx wrangler deploy
```

**4. Test your implementation**

Verify your Worker is functioning correctly:

Navigate to **Agent Analytics** in Profound and check that logs appear in the **Logs** panel. The AI log filter is on by default. Disable it using the filter in the top of the **Logs** panel.

If logs appear, you are all set. You should see data populate in the Analytics dashboard.

## Troubleshooting

* If logs aren't appearing, verify your `PROFOUND_API_URL` environment variable and `PROFOUND_LOG_INGESTION_TOKEN` secret are configured correctly

* `Log ingestion rejected the request: 401` or `403` in the Worker logs: the `PROFOUND_LOG_INGESTION_TOKEN` secret is missing, misspelled, or not the token issued for this site. Set it again with `npx wrangler secret put PROFOUND_LOG_INGESTION_TOKEN`

* `Failed to send logs: TimeoutError`: the log request passed the five-second timeout. A small number of these is expected and costs you only the affected log lines, since the timeout keeps the logging call away from your visitors. A sustained rate is worth reporting to [Profound support](mailto:support@tryprofound.com) with a few `requestId` values from the Worker logs

* `Failed to send logs:` with any other error name: check `PROFOUND_API_URL` against the value in the configuration step above. A URL that resolves to a hostname on your own Cloudflare zone makes the Worker call itself, which fails on every request

* Check Cloudflare Workers > Analytics for any execution errors

* Ensure your route pattern matches your domain configuration

* Verify the Worker is receiving requests by checking the Cloudflare dashboard metrics

* Intermittent 5xx after deploying the Worker: check Metrics > Errors > Invocation Statuses for `Exceeded Memory` (Error 1102) or `Script threw exception` (Error 1101). Both mean the Worker itself is failing, not your origin. Confirm you are running the code above (no `clone()`, no body reads) and that the logging call is wrapped in `ctx.waitUntil()` with a `.catch()`

* Google Ads / Merchant Center disapprovals such as "destination not working" or "image not accessible" after deploying the Worker: these come from `AdsBot-Google` or the Merchant Center fetcher receiving one of the Worker errors above. Fix the Worker errors, then request a re-review

* To rule the Worker out entirely, remove the route binding (`npx wrangler triggers delete` or delete the route in the dashboard) and confirm the errors stop

## Additional resources

* [Cloudflare Workers Documentation](https://developers.cloudflare.com/workers/)

* [Wrangler CLI Documentation](https://developers.cloudflare.com/workers/wrangler/)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions

## Security considerations

* Store Log Ingestion Tokens as secrets in production environments

* Regularly rotate Log Ingestion Tokens

* Monitor Worker usage and logs for unusual patterns


# Amazon CloudFront
Source: https://docs.tryprofound.com/agent-analytics/cloudfront-s3

Deliver CloudFront access logs to Profound through Amazon S3.

## Overview

Send CloudFront access logs to your assigned Profound S3 destination to see agent traffic in Agent Analytics. This guide uses standard logging (v2), cross-account delivery, and SSE-S3 encryption.

> [!NOTE]
>
> Contact [Profound support](mailto:support@tryprofound.com) with your distribution ARN to receive your destination settings and delivery source name.

For the Firehose integration, see [CloudFront through Amazon Data Firehose](/agent-analytics/aws_firehose_cloudfront).

## Prerequisites

* Access to the AWS account that owns your CloudFront distribution.
* Permission to configure CloudFront standard logging and CloudWatch Logs delivery resources.
* An AWS CLI version that supports `aws logs put-delivery-source` and `aws logs create-delivery`, authenticated to that account.
* Your CloudFront distribution ARN.

## Encryption

This guide uses **SSE-S3 (AES256)** to encrypt your logs at rest with keys managed by Amazon S3. Profound configures encryption on your destination.

## Configuration

**1. Get your destination settings**

Open your distribution in the CloudFront console and copy its ARN. Send it to Profound support. It identifies both your distribution and its AWS account:

```text theme={null}
arn:aws:cloudfront::<your-aws-account-id>:distribution/<distribution-id>
```

Profound prepares the source record, S3 destination, and cross-account permissions. You will receive:

| Setting | Purpose |
| - | - |
| Bucket and prefix | Where Profound receives this source's logs |
| Bucket region | `us-west-2` (US West, Oregon) |
| Delivery source name | The exact name to use in your AWS account |
| Delivery destination ARN | The Profound destination to connect to |
| Output format | W3C, configured by Profound |

The delivery destination ARN is a CloudWatch Logs resource ARN, not the S3 bucket ARN. Use all assigned values exactly as supplied.

**2. Create the delivery source**

Run this command in the AWS account that owns your distribution. Replace the placeholders with your assigned source name and distribution ARN:

```bash theme={null}
aws logs put-delivery-source \
  --region us-east-1 \
  --name '<assigned-delivery-source-name>' \
  --resource-arn 'arn:aws:cloudfront::<your-aws-account-id>:distribution/<distribution-id>' \
  --log-type ACCESS_LOGS
```

> [!NOTE]
>
> Run CloudFront delivery commands in **`us-east-1`**, even though the destination bucket is in **`us-west-2`**.

Use the source name supplied by Profound so it matches the configured delivery permissions.

**3. Connect to the Profound destination**

After Profound confirms the destination is ready, create the delivery in your AWS account:

```bash theme={null}
aws logs create-delivery \
  --region us-east-1 \
  --delivery-source-name '<assigned-delivery-source-name>' \
  --delivery-destination-arn '<profound-delivery-destination-arn>'
```

Keep the returned delivery ID for troubleshooting. This creates a delivery; do not rerun it to check status.

Profound manages the receiving bucket policy and delivery destination policy. You do not need to create a destination bucket or share AWS access keys with Profound.

**4. Keep the expected log format**

Keep the default standard-log field selection. Profound's destination uses **W3C** output: tab-separated records with a `#Fields` header in each file.

The fields used for Agent Analytics include:

| Fields | Used for |
| - | - |
| `date`, `time` | Request time |
| `c-ip`, `x-forwarded-for` | Client address |
| `x-host-header` | Requested domain |
| `cs(Host)` | Distribution hostname fallback |
| `cs-method`, `cs-uri-stem`, `cs-uri-query` | Requested resource |
| `cs(User-Agent)`, `cs(Referer)` | Agent and referral analysis |
| `sc-status`, `sc-bytes` | Response details |

Preserve `x-host-header` to identify the domain requested by the visitor. Do not change this destination to JSON or Parquet, remove the header, or replace tabs with spaces.

Keep the assigned destination path. Coordinate any changes to fields or file paths with Profound before applying them.

**5. Confirm delivery**

Open your distribution's **Logging** tab and confirm the S3 cross-account delivery appears. Generate traffic and allow time for logs to arrive.

Ask Profound support to confirm receipt if you need to check the destination files. Access to your AWS account does not grant read access to Profound's bucket.

## Verify ingestion

Open Agent Analytics for your domain and select a date range covering your test requests. Confirm that expected agent traffic appears after delivery and processing.

If data is missing, provide support with the distribution ARN, delivery source name, delivery ID, request time, and any AWS error. Do not include credentials.

## Troubleshooting

| Problem | What to check |
| - | - |
| CreateDelivery reports AccessDenied | Confirm your AWS permissions and ask Profound to check the destination policy for your account. |
| AWS cannot find the source or destination | Use `us-east-1` and the exact supplied resource names and ARN. |
| Delivery exists but no data appears | Confirm source readiness, generate new requests, and check the Agent Analytics date range. |
| Records cannot be processed | Confirm W3C output, a `#Fields` header, tab-separated records, and the expected fields. |
| Traffic appears under the distribution hostname | Check that `x-host-header` is included and contains the requested domain. |

## Additional Resources

* [AWS: CloudFront standard logging (v2) and cross-account delivery](https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/standard-logging.html)
* [CloudFront through Amazon Data Firehose](/agent-analytics/aws_firehose_cloudfront)
* [Contact Profound support](mailto:support@tryprofound.com)


# Crawler Verification
Source: https://docs.tryprofound.com/agent-analytics/crawler-verification

How we verify AI and search engine crawlers

## Crawler Verification Process

Understanding who's accessing your website is crucial for security and analytics accuracy. Profound Agent Analytics platform employs robust verification methods to ensure that crawlers claiming to be from major AI and search platforms are genuine.

## Why Verification Matters

Accurate crawler identification is essential for:

* Protecting your website from malicious actors

* Ensuring data accuracy in your analytics

* Managing resource allocation effectively

* Maintaining security compliance

* Optimizing content delivery for legitimate AI platforms

## Verification Methods

### Primary Verification Techniques

Our platform employs multiple verification methods to ensure accuracy:

1. **Reverse DNS Lookup**

   * Verifies the crawler's hostname matches the claimed organization

   * Provides an additional layer of authenticity checking

   * Used by established platforms like Google and Apple

2. **IP Range Validation**

   * Confirms the crawler originates from the organization's known IP ranges

   * Particularly effective for platforms like OpenAI and You.com

   * Updated regularly to maintain accuracy

3. **Heuristic Detection**

   * Analyzes crawler behavior patterns

   * Identifies characteristic signatures

   * Helps verify crawlers without published verification methods

## Platform-Specific Verification

### Fully Verified Platforms

* [**Google**](https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers)

  * Google-CloudVertexBot: Reverse DNS verification

  * Googlebot: Reverse DNS verification + ASN verification

  * Google-Extended: Reverse DNS verification

  * Storebot-Google: Reverse DNS verification

  * Googlebot-News: Reverse DNS verification

  * Googlebot-Video: Reverse DNS verification

  * Googlebot-Image: Reverse DNS verification

  * Google-NotebookLM: ASN verification

  * Google-Agent: Reverse DNS verification + IP range verification + ASN verification

  * GoogleAgent-Mariner: IP range verification

  * Google-Other: Heuristic detection only

* [**Microsoft Bing**](https://www.bing.com/webmasters/help/which-crawlers-does-bing-use-8c184ec0)

  * BingBot: Reverse DNS verification

  * MicrosoftPreview: Reverse DNS verification

* [**Apple**](https://support.apple.com/en-us/119829)

  * Applebot: Reverse DNS verification + IP range verification

  * Applebot-Extended: Reverse DNS verification + IP range verification

* [**OpenAI**](https://platform.openai.com/docs/bots)

  * OAI-SearchBot: IP range verification + ASN verification

  * ChatGPT-User: IP range verification + ASN verification

  * GPTBot: IP range verification + ASN verification

  * OAI-Operator: ASN verification

* [**Duck Duck Go**](https://duckduckgo.com/duckduckgo-help-pages/results/duckduckbot)

  * DuckDuckBot: IP range verification

  * DuckAssistBot: ASN verification

* [**Perplexity**](https://docs.perplexity.ai/guides/bots)

  * PerplexityBot: IP range verification

  * Perplexity-User: IP range verification

* [**Meta**](https://developers.facebook.com/docs/sharing/webmasters/web-crawlers/#identify-2)

  * Meta-ExternalAgent: ASN verification

  * Meta-ExternalFetcher: ASN verification

  * Meta-WebIndexer: ASN verification

  * FacebookBot: Heuristic detection only

* [**Amazon**](https://developer.amazon.com/amazonbot)

  * AmazonBot: Reverse DNS verification + IP range verification

  * Amazon-SearchBot: IP range verification

  * Amazon-User: IP range verification

  * AmazonBuyForMe: ASN verification

### Partially Verified Platforms

Some platforms use common cloud provider IPs or don't publish verification methods, making complete verification challenging:

* [**Anthropic**](https://privacy.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler)

  * ClaudeBot: ASN + Heuristic detection

  * Claude-User: ASN + Heuristic detection

  * Claude-SearchBot: ASN + Heuristic detection

* **You.com**

  * YouBot: Reverse DNS verification + IP range verification

* **Bytedance**

  * Bytespider: Reverse DNS verification

  * TikTokSpider: Heuristic detection only

* [**Common Crawl**](https://commoncrawl.org/ccbot)

  * CCBot: IP range verification

* **Yahoo**

  * Slurp: Reverse DNS verification

* **OpenClaw**

  * OpenClaw: Heuristic detection only

* **DeepSeek**

  * DeepSeek: Reverse DNS verification

* **Baidu**

  * BaiduSpider: Reverse DNS verification

* **Huawei**

  * PetalBot: ASN verification

* [**Mistral**](https://docs.mistral.ai/robots)

  * MistralAI-User: IP range verification

* **Yandex**

  * YandexBot: IP range verification

* **Gemini**

  * Google: Heuristic detection

  * Gemini-Deep-Research: Reverse DNS verification + ASN verification

## Stay Updated

Our platform continuously updates verification methods as:

* New AI platforms emerge

* Existing platforms modify their crawler infrastructure

* Additional verification methods become available

* Security requirements evolve

## Important Note About Data Updates

We continuously monitor and improve our verification processes as the AI crawler landscape evolves. As we enhance our detection methods and crawler identification techniques, you may notice changes in your historical and current analytics data. These updates reflect our commitment to providing the most accurate and reliable crawler identification possible.

If you observe any significant changes in your data, it's likely due to improvements in our verification system. We recommend regularly reviewing your analytics dashboard to stay informed about the latest insights into AI crawler behavior on your site.


# Custom Log Integration
Source: https://docs.tryprofound.com/agent-analytics/custom

This documentation explains how to send your application logs directly to Profound's analytics platform using our Custom Log Integration API.

## Overview

The Custom Log Integration allows you to send your application logs directly to Profound's analytics platform using a standardized JSON format. This integration supports batch processing of logs and includes robust validation to ensure data quality while providing detailed feedback on any validation issues.

## Prerequisites

* Your Profound Log Ingestion Token

* The ability to send HTTP POST requests from your application/infrastructure

## API Specification

### Endpoint

```http theme={null}
POST https://artemis.api.tryprofound.com/v1/logs/custom
```

### Authentication

Include your Profound Log Ingestion Token in the request header:

```http theme={null}
'x-api-key': 'bot_PROFOUND_LOG_INGESTION_TOKEN'
```

### Request Format

Logs must be sent as an array of JSON objects. Each log entry requires specific fields:

#### Required Fields

* `timestamp` - Event timestamp (Unix timestamp or ISO 8601 string)

* `method` - HTTP method (max 10 characters)

* `host` - Request hostname (max 255 characters)

* `path` - Request path (max 2048 characters)

* `status_code` - HTTP status code (range: 100-599)

* `ip` - Client IP address (max 45 characters)

* `user_agent` - User agent string (max 1024 characters)

#### Optional Fields

> [!WARNING]
>
> Both `query_params` and `referer` are optional. However, if you do not provide either, human referral data will **not be available** in your dashboard.

* `query_params` - Query parameters (key max: 100 chars, value max: 1000 chars)

* `referer` - Request referer (max 2048 characters)

* `bytes_sent` - Response size in bytes (must be >= 0)

* `duration_ms` - Request duration in milliseconds (must be >= 0)

> [!TIP]
>
> Each request can contain up to 1,000 log entries. For larger datasets, split your logs into multiple requests.

## Implementation Guide

![Custom Detected](https://mintcdn.com/profound-37face47/cWnNS0qcU3zRxQWi/images/custom_setup.png?fit=max\&auto=format\&n=cWnNS0qcU3zRxQWi\&q=85\&s=5dfad431ec225ef0cd700f3ffc201494)

**1. Format Your Logs**

Prepare your log data in the required JSON format:

```json theme={null}
[
  {
    "timestamp": "2024-01-11T12:00:00Z",
    "method": "GET",
    "host": "example.com",
    "path": "/products",
    "status_code": 200,
    "ip": "192.168.1.1",
    "user_agent": "Mozilla/5.0...",
    "query_params": {
      "category": "electronics",
      "page": "1"
    },
    "referer": "https://example.com",
    "bytes_sent": 1024,
    "duration_ms": 150
  }
]
```

**2. Send the Request**

Here's how to send logs using popular programming languages:

Python:

```python theme={null}
import requests
import json

logs = [{
    "timestamp": "2024-01-11T12:00:00Z",
    "method": "GET",
    "host": "example.com",
    "path": "/products",
    "status_code": 200,
    "ip": "192.168.1.1",
    "user_agent": "Mozilla/5.0..."
}]

response = requests.post(
    "https://artemis.api.tryprofound.com/v1/logs/custom",
    headers={
        "x-api-key": "bot_PROFOUND_LOG_INGESTION_TOKEN",
        "Content-Type": "application/json"
    },
    json=logs
)

print(response.json())
```

Node.js:

```javascript theme={null}
const axios = require('axios');

const logs = [{
    timestamp: "2024-01-11T12:00:00Z",
    method: "GET",
    host: "example.com",
    path: "/products",
    status_code: 200,
    ip: "192.168.1.1",
    user_agent: "Mozilla/5.0..."
}];

axios.post('https://artemis.api.tryprofound.com/v1/logs/custom', logs, {
    headers: {
        'x-api-key': 'bot_PROFOUND_LOG_INGESTION_TOKEN',
        'Content-Type': 'application/json'
    }
})
.then(response => console.log(response.data))
.catch(error => console.error(error));
```

**3. Handle the Response**

Process the API response to confirm successful ingestion or handle any errors:

Successful Response:

```json theme={null}
{
  "status": "accepted",
  "message": "Processing X log entries",
  "first_visit_id": "uuid-string",
  "errors": []
}
```

Response with Validation Errors:

```json theme={null}
{
  "status": "accepted",
  "message": "Processing 98 log entries",
  "first_visit_id": "uuid-string",
  "errors": [
    "Invalid log format: timestamp out of reasonable range",
    "Invalid log format: status_code must be between 100 and 599"
  ]
}
```

## Error Handling

The API handles errors in two ways:

### Validation Errors

For expected validation issues (invalid timestamps, malformed data, etc.), the API will:

* Continue processing valid entries in the batch

* Add validation errors to the `errors` array in the response

* Return a 200 status code with processed entries and errors

### Unexpected Errors

For unexpected errors (server issues, database problems, etc.), the API will:

* Return a 500 status code

* Return an error message in the response detail field

* Not process any entries in the batch

## Best Practices

* Group logs into batches of up to 1,000 entries

* Implement retry logic for failed requests

* Use background processing for log submission

* Send logs asynchronously to avoid impact on application performance

* Consider implementing local buffering

* Use compression for large payloads

## Support

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for assistance

## Security Considerations

* Store Log Ingestion Tokens securely

* Regularly rotate Log Ingestion Tokens

* Monitor request logs for unusual patterns

* Use HTTPS for all API requests

***


# Adobe Analytics Extension
Source: https://docs.tryprofound.com/agent-analytics/extensions/adobe-analytics

Connect Adobe Analytics with Profound Agent Analytics to correlate AI-referred traffic with human engagement, orders, and revenue

## Overview

Profound Agent Analytics now offers seamless integration with Adobe Analytics, enhancing your ability to correlate AI interactions with human engagement and business outcomes. This integration bridges the gap between AI visibility and traditional web analytics, providing a comprehensive view of your digital performance ecosystem.

By connecting your Adobe Analytics report suite, Agent Analytics can import critical engagement and revenue metrics alongside our specialized AI tracking capabilities. This integration works both as a complementary extension to our core platform or as a standalone solution, giving you flexibility in how you monitor and optimize your digital presence for both human and AI visitors.

The Adobe Analytics integration automatically synchronizes your analytics data in real-time when accessing the Adobe Analytics dashboard, allowing you to make data-driven decisions with the most current information available while maintaining strict privacy controls.

### Adobe Analytics Metrics

The following metrics are retrieved from the [Adobe Analytics Reporting API](https://developer.adobe.com/analytics-apis/docs/2.0/guides/endpoints/reports/):

* **Referrer** (`referrer`): The referring source of the visit, used to identify traffic originating from AI systems
* **Visits** (`visits`): The number of human sessions referred by AI platforms
* **Orders** (`orders`): Completed purchase events from AI-referred traffic
* **Revenue** (`revenue`): Total monetary value generated from AI-referred visits

### Dashboard

Once connected, the Adobe Analytics tab in your Profound dashboard provides a unified view of how AI-referred traffic converts on your website.

#### Summary Metrics

At the top of the dashboard, you'll see high-level metrics for your selected date range:

* **Humans visited from AI systems**: The total number of human visitors who arrived at your site after being referred by AI assistants
* **Revenue from AI systems**: The total revenue attributed to AI-referred traffic

Both metrics include period-over-period comparison to help you track trends.

#### Platforms

The Platforms table breaks down AI-referred traffic by the AI system that sent the visit. Each platform row includes:

* **Visits**: Number of human visits referred by the platform
* **Orders**: Completed purchase events from those visits
* **Revenue**: Total revenue generated from AI-referred visitors
* **Conversion Rate**: Percentage of visits that resulted in an order

#### Human Visits by Page

The page-level breakdown shows which pages humans visit most after being referred by AI assistants. Each row includes:

* **Visits**: Number of AI-referred visits to the page
* **Orders**: Purchase events that occurred on or through the page
* **Revenue**: Revenue attributed to the page

## Prerequisites

* Adobe Analytics account

* Adobe Analytics configured for your website

* Access to a target Report Suite via your Adobe account

## Configuration within Adobe Admin Console

**1. Step 1**

Go to the [Adobe Admin Console](https://adminconsole.adobe.com/) and select your Adobe Analytics product.

![Adobe Admin Console](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/within_adobe_admin_console.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=8f7dfcf5b3e2451d788753958ed4a816)

**2. Step 2**

In the **Product Profiles** section, create a new profile for Profound's Agent Analytics Integration using the **New Profile** button. You'll be prompted to give it a name and optionally a description.

![New Profile](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/within_adobe_new_profile.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=e4ba6f14aae851f53de34cee44407b6e)

**3. Step 3**

Click into the new product profile and navigate to the **Permissions** tab.

![Product Profile Permissions](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/within_adobe_product_profile.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=5608216ab8251f514552740b92b12f14)

**4. Step 4**

Click on the edit icon within the **Report Suites** row.

![Edit Icon](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/within_adobe_edit_icon.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=dccbfaf4e5e3402df714c9c02a9c082f)

**5. Step 5**

For **Report Suites**, **Metrics**, and **Dimensions**, toggle on **Auto-include** for each one.

![Auto-include](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/within_adobe_auto_include_on.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=b73db93438130776b2ee37fe7005e1cf)

**6. Step 6**

Within the **Report Suite Tools** tab, add the **Report Suite Mgmt** permission.

![Report Suite Tools](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/within_adobe_report_suites_tools_addon.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=c960519ed5319a35873798a3dc963672)

**7. Step 7**

Within the **Analytics Tools** tab, add the **Web Services** and **Web Service Access** permissions.

![Analytics Tools Permissions](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/wihtin_adobe_analytics_tools_perms.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=3fef9b3b8f6c11a52ef3c616b5160ca1)

**8. Step 8**

Click **Save**.

## Configuration within Adobe Developer Console

**1. Step 1**

Navigate to the [Adobe Developer Console](https://developer.adobe.com/) and click **Create new project**. Make sure you are creating the project under the same organization that the product profile was created under in the Admin Console.

![Create New Project](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/adobe_dev_create_new_project.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=cd911c9b5b99de8179e79640de389476)

**2. Step 2**

Select **Add API**.

![Add API](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/adobe_dev_add_api.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=ba6e14417b0c88071f33127ca30a9c91)

**3. Step 3**

Select **Adobe Analytics**.

![Select Adobe Analytics](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/adobe_dev_add_adove_analytics.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=0ffaeee7bb6070479b70bec43d4e970e)

**4. Step 4**

Select **Server-to-Server Authentication** and click **Next**. After clicking next, input what you'd like to name the credential (for example, Profound Agent Analytics) and then click **Next** again.

![Server-to-Server Authentication](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/adobe_dev_s2s.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=9cbea7e4ca1a57e01e758556982939b6)

**5. Step 5**

From the **Product Profile** list, select the product profile that was created in the Admin Console in the previous steps. Then click **Save configured API** on the bottom right.

![Select Product Profile](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/adobe_dev_product_profile_select.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=d679ec8b6a7e19354304a3f3a3b3ab16)

**6. Step 6**

In the newly created project, navigate to the **OAuth Server-to-Server** tab and retrieve the **Client ID**, **Client Secret**, and **Organization ID**. These values will be necessary when connecting the project within Profound.

![OAuth Credentials](https://mintcdn.com/profound-37face47/pZin5bi5tFO0ur3R/images/agent-analytics/extensions/adobe_dev_creds.png?fit=max\&auto=format\&n=pZin5bi5tFO0ur3R\&q=85\&s=36c3180711b99618ead0b5b8d7d09794)

## Configuration within Profound

**Agent Analytics + Adobe Analytics**

Learn how to integrate Adobe Analytics with your existing Agent Analytics configuration.

**1. Step 1**

Navigate to the **Adobe Analytics** tab. Here you will find a button to **Connect account**.

![Adobe Analytics Tab](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_tab.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=af26d46a12795bbe38b69364ac4cd472)

**2. Step 2**

Click **Connect account**.

![Set up Adobe Analytics](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_setup.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=4efec5292fb490eda425b8b236588f63)

**3. Step 3**

Input the **Client Secret**, **Client ID**, and **Organization ID** retrieved from the Adobe Developer Console from the previous steps.

![Credentials Input](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/profound_credentials_inputs.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=c84f29c22c7f832e08524d0b9d0458b9)

**4. Step 4**

Once your Adobe account is connected, choose your Report Suite. Select a report suite to connect to this domain from the dropdown.

![Choose Report Suite](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_report_suite.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=e0246611f73ff39c11d3238b9af0351f)

**5. Step 5**

Click **Complete setup** to finalize the connection.

![Complete Setup](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_complete_setup.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=4a47a5c46d77cbc1c96cd3fea91f5007)

> [!TIP]
>
> That's it! Adobe Analytics is now connected. Reload your Adobe Analytics tab to see data from your selected report suite.

**Adobe Analytics Alone**

Learn how to integrate Adobe Analytics without Agent Analytics

**1. Step 1**

Navigate to the **Adobe Analytics Setup** tab. Here you will find a button to **Connect account**.

![Adobe Analytics Tab](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_setup_alone.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=a7e87461009e5ae25ab7f9792704e79a)

**2. Step 2**

Click **Connect account**.

![Set up Adobe Analytics](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_setup_alone.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=a7e87461009e5ae25ab7f9792704e79a)

**3. Step 3**

Input the **Client Secret**, **Client ID**, and **Organization ID** retrieved from the Adobe Developer Console from the previous steps.

![Credentials Input](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/profound_credentials_inputs.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=c84f29c22c7f832e08524d0b9d0458b9)

**4. Step 4**

Once your Adobe account is connected, choose your Report Suite. Select a report suite to connect to this domain from the dropdown.

![Choose Report Suite](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_report_suite.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=e0246611f73ff39c11d3238b9af0351f)

**5. Step 5**

Click **Complete setup** to finalize the connection.

![Complete Setup](https://mintcdn.com/profound-37face47/XgLNXwd99BklcUAl/images/agent-analytics/extensions/adobe_analytics_complete_setup.png?fit=max\&auto=format\&n=XgLNXwd99BklcUAl\&q=85\&s=4a47a5c46d77cbc1c96cd3fea91f5007)

> [!TIP]
>
> That's it! Adobe Analytics is now connected. Reload your Adobe Analytics tab to see revenue data from your selected report suite.

## Disconnecting

You can disconnect the Adobe Analytics integration at any time in the Settings tab. Your integration will be disabled immediately.

## Security and Privacy

> [!NOTE]
>
> Profound requests READ ONLY access to your Adobe Analytics data and does not download, save, or otherwise retain your Adobe Analytics data.

Your data is fetched on demand for processing and display purposes only. We persist minimal report suite information (ID, name, and association with your selected website) to maintain the connection until you choose to disconnect the Adobe Analytics integration. You can opt out at any time through the Settings tab, giving you complete control over your analytics data.

## Considerations

* Only one report suite can be connected per website at this time

* Only one member of your organization must complete the Adobe authentication process. They must have access to the target Adobe Analytics account and report suites.

* Adobe Analytics can be blocked by various browsers or browser extensions. For best results, we recommend using Adobe Analytics in conjunction with the Agent Analytics platform

## Additional Resources

* [Adobe Analytics API Documentation](https://developer.adobe.com/analytics-apis/docs/2.0/)

* [Adobe Exchange Listing](https://exchange.adobe.com/apps/ec/600242/adobe-analytics-profound-agent-analytics-integration)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions


# Google Analytics (GA4) Extension
Source: https://docs.tryprofound.com/agent-analytics/extensions/google-analytics

Connect AI visibility with human engagement metrics to create a complete picture of your digital performance and revenue impact

## Overview

Profound Agent Analytics now offers seamless integration with Google Analytics 4 (GA4), enhancing your ability to correlate AI interactions with human engagement and business outcomes. This integration bridges the gap between AI visibility and traditional web analytics, providing a comprehensive view of your digital performance ecosystem.

By connecting your GA4 property, Agent Analytics can import critical conversion and revenue metrics alongside our specialized AI tracking capabilities. This integration works both as a complementary extension to our core platform or as a standalone solution, giving you flexibility in how you monitor and optimize your digital presence for both human and AI visitors.

The GA4 integration automatically synchronizes your analytics data in real-time when accessing the Google Analytics dashboard, allowing you to make data-driven decisions with the most current information available while maintaining strict privacy controls.

### GA4 Metrics

* Sessions (`sessions`): Track human visitor engagement based on GA4 session data
* Transactions (`transactions`): Monitor purchase-related events including completed sales and refunds
* Purchase Revenue (`purchaseRevenue`): Quantify the financial impact of your digital presence
* Session Key Event Rate (`sessionKeyEventRate`): Evaluate conversion efficiency through the percentage of sessions resulting in transactions
* Page Path Analysis (`pagePath`): Identify your most valuable content for both humans and AI systems

## Prerequisites

* Google Analytics account

* Google Analytics configured for your website

* Access to a target Property via your Google account

## Configuration

**Agent Analytics + GA4**

Learn how to integrate GA4 with your existing Agent Analytics Configuration.

**1. Step 1**

Navigate to the Google Analytics tab. Here you will find a button to Connect Google Analytics.

![Google Analytics Tab](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/extensions/google_analytics_tab.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=c7064407885861b97f929e4c0e294aac)

**2. Step 2**

Click **Configure Google Analytics**. This will direct you to connect your Google account to Profound. When prompted, *you should select the Google account which has access to the GA4 Project you want to integrate.*

![Connect Google Account](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/connect_google.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=ec7225779e5cf28187e5703f6f732308)

![Google OAuth](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/extensions/google_oauth.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9d98364e79234d16cc4443b0617c192b)

**3. Step 3**

Select the website you want to connect. This will determine which domain displays GA4 data. Then, select the GA4 property to retrieve data from.

![Select Website and Property](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/extensions/property_website_selection.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=a0765bd872511f27a2b6c3aa71f4c187)

**Step 4**

Profound will now verify the connection.

![Configured](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/configured.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=2cd3fda73d5efd49ae2c9c9c1c86c536)

> [!TIP]
>
> That's it! Google Analytics is now connected. Reload your Google Analytics tab to see revenue data from your selected property.

**GA4 Alone**

Learn how to integrate GA4 without Agent Analytics

**1. Step 1**

Navigate to Website tab in the sidebar. If you have not configured Agent Analytics, you will see the following window.

![Referrals Tab](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/ga4_only.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=24b6bcebd7ef0638f110fc5322a723dd)

**2. Step 2**

Click **Google Analytics** near the bottom. This will direct you to connect your Google account to Profound. When prompted, *you should select the Google account which has access to the GA4 Project you want to integrate.*

![Connect Google Account](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/connect_google.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=ec7225779e5cf28187e5703f6f732308)

![Google OAuth](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/extensions/google_oauth.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9d98364e79234d16cc4443b0617c192b)

**3. Step 3**

Select **+ Add a new website** and enter your domain name. Use the domain for which your target GA4 property is configured. Then, select the GA4 property to retrieve data from.

![Add New Website](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/add_new_domain.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=57feefd90475e5d74ec6d97543255053)

**Step 4**

Profound will now verify the connection.

![Configured](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/configured.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=2cd3fda73d5efd49ae2c9c9c1c86c536)

## Disconnecting

You can disconnect GA4 integration at any time in the Settings tab. Your integration will be disabled immediately.

![Disconnecting](https://mintcdn.com/profound-37face47/dU1U8PbY91PjiXyk/images/agent-analytics/extensions/disconnect_ga4.png?fit=max\&auto=format\&n=dU1U8PbY91PjiXyk\&q=85\&s=198c6f5053a1f10720024ad8aa6f106e)

## Security and Privacy

> [!NOTE]
>
> Profound requests READ ONLY access to your Google Analytics data and does not download, save, or otherwise retain your Google Analytics data.

Your data is fetched on demand for processing and display purposes only. We persist minimal property information (ID, name, and association with your selected website) to maintain the connection until you choose to disconnect the GA4 integration. You can opt out at any time through the Settings tab, giving you complete control over your analytics data.

## Considerations

* Only one property can be connected per website at this time

* Every team member must complete the Google authentication process. They must have access to the target GA4 project and property

* Google Analytics is blocked by various browsers or browser extensions. For best results, we recommend using GA4 in conjunction with the Agent Analytics platform

## Additional Resources

* [Google Analytics Data API](https://developers.google.com/analytics/devguides/reporting/data/v1/rest)

* [Google Analytics Metric Information](https://developers.google.com/analytics/devguides/reporting/data/v1/api-schema)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions


# Frequently Asked Questions
Source: https://docs.tryprofound.com/agent-analytics/faq



## Agent Analytics

> **My site is hosted on Shopify / Squarespace / WP / Kinsta / Liquid / etc. How can I connect Agent Analytics?**
>
> Profound supports integrations with many providers. However, managed providers cannot be integrated at this time.
> Managed providers like those above offer a streamlined experience that abstracts the technical requirements of setting up and hosting a website.
> This abstraction makes setting up a website much easier, but consequently obfuscates the niche features necessary to integrate Agent Analytics.
>
> Generally, Profound requires a means to access your site's server logs to analyze traffic and surface insights about bot activity. In most cases,
> managed providers simply do not offer such a mechanism. Without the ability to access your site's server logs, we cannot integrate with your site.
>
> If you are using a managed provider, please contact your provider to inquire about server logs. If you have some means of accessing your site's server logs,
> we offer a custom endpoint where these can be sent. Please contact support to learn more.

> **Do I need technical expertise and access to my site's code / CDN to configure Agent Analytics?**
>
> Yes, you will need to have access to your site's hosting platform and permissions to configure the CDN.
> Profound provides comprehensive documentation for each supported platform, including step-by-step instructions and configuration examples.
> However, we cannot edit your site's code or CDN configuration for you. Thus, it is necessary to have technical expertise and access to your site's code / CDN
> to complete the integration.

> **How do I know if Agent Analytics is working?**
>
> You can verify that the log collection is working by checking the **Logs** tab in **My Website**. After completing the onboarding process,
> please allow a few minutes for data to begin populating. If there is no data after 24 hours, please contact support.

> **I used the Cloudfront integration via Firehose. Why am I not seeing any data?**
>
> Cloudfront sometimes sends logs from a unique domain that differs from your site's domain. This is normal and expected, but means the
> key tied to your real site's domain does not match the domain name coming from Cloudfront. Please contact support to alter the domain.
> Data will begin appearing shortly after the domain is updated.

> **I used the Cloudflare integration via Logpush. Why is there no HTTP destination option at Step 2 / no ClientRequestHost filter at step 4?**
>
> Cloudflare integrations need to be configured at the domain level. This can be accessed by signing into Cloudflare > select an account >
> **Account Home** (on the left sidebar) > select your domain > complete the steps in the documentation.

> **I registered two (or more) domains with Agent Analytics. I see one domain by default. How do I see the other domains?**
>
> The domain name at the top of the page is a dropdown menu. Click once to see a list of registered domains, then select the domain you want to view.

## Google Analytics

> **What domain should I use when registering my site with Google Analytics?**
>
> The domain name to enter when registering via Google Analytics can be any label of your choosing.
> If you are configuring only GA4, use whichever label makes sense to you.
> If you are configuring both GA4 and Agent Analytics for the same site, use the site name you registered with Agent Analytics.
> This will combine the data to see it all in one place.

> **Why can't I click on the Overview / Bots / Pages / Logs tabs?**
>
> Google Analytics exposes fundamentally different data than the server logs coming from a CDN integration.
> Using a CDN integration allows Profound to track traffic on your site at the most granular level.
> GA4 does not have the same level of granularity in its data. We are able to track human sessions only, including identifying
> a referral from AI platforms.
>
> Additionally, GA4 provides data that is not exposed in the server logs. Namely, we can see transaction and revenue data that is associated
> with AI referred traffic.

> **My team registered GA4 already. Why can't I see any data?**
>
> If someone on your team has registered a Google Analytics property linked to your Profound account, and you are not able to see any data,
> please ensure you have registered your Google account as well. For security purposes, we do not share the access tokens from one account to another.
> If your teammates are able to see data, they have valid Google tokens associated with their account. To see the data from your account, please ensure
> you have authorized Profound to access your Google account. You can do this in **My Website** → **Settings** → **Connect Google Analytics** at the bottom of the page.

> **I have configured both GA4 and Agent Analytics. Why am I seeing different data in the Referrals tab?**
>
> If you have just registered Agent Analytics via a CDN integration within the last week, Profound has not yet collected enough data to accurately
> represent your site's traffic. Google Analytics will have historical data available immediately, but Agent Analytics will take some time to catch up.
>
> Additionally, regardless of sign up date, GA4 can be blocked by certain browsers or extensions. The Agent Analytics data cannot be blocked. Thus, your GA4
> data may be slightly different than the Agent Analytics data.


# Fastly HTTPS Endpoint
Source: https://docs.tryprofound.com/agent-analytics/fastly

This documentation explains how to set up Fastly real-time logs delivery to Profound Agent Analytics platform using Fastly custom HTTPS Endpoint

## Overview

The integration uses Fastly's custom HTTPS endpoint to forward Fastly real-time logs to our Agent Analytics API. Fastly's real-time log streaming offers instant visibility into traffic, performance, threats, and troubleshooting, and streams logs to various destinations including HTTP endpoints. For more information about Fastly real-time logs, [visit the Fastly documentation](https://docs.fastly.com/en/guides/streaming-logs).

## Prerequisites

* A Fastly account

* Access/permission to edit Logging Endpoints

* A Profound Log Ingestion Token for Agent Analytics

## Configuration

![Fastly Detected](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/fastly_setup.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=376e9c92e922aed26a434d2028bc0af0)

**1. Step 1**

Sign into the Fastly console, select your service, and navigate to the Service configuration tab.

![Service Configuration Tab](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/service_configuration.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9f5c9b84ec0be5034c9e3d29a998bc68)

**2. Step 2**

Select "Edit configuration" and navigate to the Logging section.

Find "HTTPS" and click `+ Create endpoint`

![Logging Endpoint](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/logging_endpoint.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=1fdf94eb3baeea16c21a0b2d651602f7)

**3. Step 3**

Configure your HTTPS endpoint with the following settings:

![Configuration](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/create_https_endpoint.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=f229d92714f0e359e77fe14b9e625acf)

* **Name:** Add an informative name

* **Placement:** Use the default

* **URL:** Enter `https://artemis.api.tryprofound.com/v1/logs/fastly`

* **Maximum logs & Maximum bytes:** Optional, configure as needed.

* **Log format**: Paste the following log format. This governs what information will be sent to the Agent Analytics API. If you use a different format, your logs may be rejected by the API.

  ```
    {
        "timestamp": "%{strftime(\{"%Y-%m-%dT%H:%M:%S%z"\}, time.start)}V",
        "client_ip": "%{req.http.Fastly-Client-IP}V",
        "start_time": "%{begin:%Y-%m-%dT%H:%M:%S%z}t",
        "end_time": "%{end:%Y-%m-%dT%H:%M:%S%z}t",
        "host": "%{req.http.Host}V",
        "url": "%{json.escape(req.url)}V",
        "request_method": "%{json.escape(req.method)}V",
        "request_referer": "%{json.escape(req.http.referer)}V",
        "request_user_agent": "%{json.escape(req.http.User-Agent)}V",
        "response_status": %{resp.status}V,
        "response_body_size": %{resp.body_bytes_written}V
    }
  ```

  * Time and IP

    * `timestamp` - Time when the request was completed

    * `client_ip` - Client IP address

  * Request Details

    * `request_method` - HTTP request method

    * `host` - Requested host header

    * `url` - Request url path

    * `request_referer` - Request referrer

    * `request_user_agent` - Client user agent

  * Response Details

    * `start_time` - The time at which the request began

    * `end_time` - The time at which the request finished

    * `response_status` - HTTP response status

    * `response_body_size` - Response size in bytes

  *For more information see the [documentation](https://docs.fastly.com/en/guides/useful-variables-to-log)*

  > [!NOTE]
  >
  > You may see a warning like:
  >
  > ![Warning Message](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/warning_message.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=05aab745a4e1946bb2ae3cec2df9527c)
  >
  > You can safely ignore this message. The validation has been configured for you.

**4. Step 4**

Next, click "Attach a condition"

![Attach a condition 1](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/attach_a_condition1.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=70409e193f42a771baa45bb24c161099)

![Attach a condition 2](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/attach_a_condition2.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=a9c05f001e10b3e81e3b9d7e63bd3817)

Enter the following using your domain name:

`req.http.host == "my-website.com"`

![Filter to host](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/filter_to_host.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=70fb508bcb2ebfc02de72b1603b7ee9b)

This filter will ensure that only logs for your selected domain are sent to Profound. Select `Save and apply`.

**5. Step 5**

Scroll down to "Advanced options"

![Advanced Options](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/fastly/advanced_options.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=f604cc2170c571c08d0f610874d8c6c2)

Configure:

* **Content type** - `application/json`

* **Custom header name** - `X-API-Key`

* **Custom header value** - Your provisioned Log Ingestion Token for Profound's Agent Analytics Platform (`bot_PROFOUND_LOG_INGESTION_TOKEN`)

* **Method** - `POST`

* **JSON log entry format** - `Array of JSON`

* **Select a log line format** - `Blank`

**6. Finished**

Save your configuration and activate the new service version.

> [!TIP]
>
> That's it! Fastly will now send real-time logs to Profound. Data should begin appearing in your dashboard within a few minutes.

## Troubleshooting

* If logs aren't appearing, verify your HTTPS endpoint configuration

* Check Fastly for delivery errors

* Verify your Log Ingestion Token is correct

* Verify your API endpoint is correct

## Additional Resources

* [Fastly's real-time log streaming](https://docs.fastly.com/en/guides/about-fastlys-realtime-log-streaming-features)

* [Log streaming: HTTPS](https://docs.fastly.com/en/guides/log-streaming-https)

* [Log Formats](https://docs.fastly.com/en/guides/useful-log-formats)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions


# Fastly
Source: https://docs.tryprofound.com/agent-analytics/fastly-s3

Deliver Fastly access logs to Profound through Amazon S3.

## Overview

Send Fastly access logs to your assigned Profound S3 destination to see agent traffic in Agent Analytics. This guide covers Fastly CDN services using VCL and SSE-S3 encryption.

> [!NOTE]
>
> Contact [Profound support](mailto:support@tryprofound.com) to receive your bucket name, destination prefix, region, and IAM role ARN before enabling delivery.

For delivery over HTTP, use the [Fastly HTTPS integration](/agent-analytics/fastly). For a Fastly Compute service, contact support to confirm the logging configuration.

## Prerequisites

* Access to the Fastly service that handles your website's traffic.
* Permission to edit logging endpoints and activate a service version.
* Your Fastly account ID, which Profound needs to configure delivery permissions.

## Encryption

This guide uses **SSE-S3 (AES256)** to encrypt your logs at rest with keys managed by Amazon S3.

## Configuration

**1. Get your destination settings**

Contact Profound support with your domain and Fastly account ID. Profound configures the source and a role that allows Fastly to write to your assigned destination.

You will receive a bucket name, destination prefix, region, and role ARN. Use the prefix exactly as provided, including the trailing slash. The prefix identifies your source.

```text theme={null}
s3://<assigned-bucket>/<assigned-prefix>/
```

The role ARN is the authorization value for this setup. You do not need AWS access keys or a Profound HTTP ingestion token.

**2. Create an S3 logging endpoint**

Open your Fastly service and edit a service version. In its logging settings, create an **Amazon S3** endpoint.

| Setting | Value |
| - | - |
| Name | A descriptive name |
| Bucket name | The bucket supplied by Profound |
| Path | Your complete assigned prefix, including the trailing slash |
| Domain | The S3 endpoint for the supplied region; for `us-west-2`, use `s3.us-west-2.amazonaws.com` |
| Access method | **IAM Role** |
| IAM role ARN | The role supplied by Profound |
| Server side encryption | **Amazon S3-managed keys (SSE-S3 / AES256)** |

Keep the default logging placement. Follow Fastly's [S3 endpoint guide](https://www.fastly.com/documentation/guides/integrations/logging-endpoints/object-and-cloud-storage/log-streaming-amazon-s3/) for the current control-panel layout.

**3. Set the log format**

Paste this template into **Log format** as a single line. It emits the fields expected by Profound's Fastly processor.

```text theme={null}
{"timestamp":"%{begin:%Y-%m-%dT%H:%M:%S%z}t","client_ip":"%{req.http.Fastly-Client-IP}V","host":"%{json.escape(req.http.Host)}V","url":"%{json.escape(req.url)}V","request_method":"%{json.escape(req.method)}V","request_referer":"%{json.escape(req.http.Referer)}V","request_user_agent":"%{json.escape(req.http.User-Agent)}V","response_status":%{resp.status}V,"response_body_size":%{resp.body_bytes_written}V}
```

| Fields | Used for |
| - | - |
| `timestamp` | Request time |
| `client_ip` | Client address |
| `host` | Requested domain |
| `url`, `request_method` | Requested resource |
| `request_referer`, `request_user_agent` | Referral and agent analysis |
| `response_status`, `response_body_size` | Response details |

Use **Blank** for the log line format (`message_type=blank`) so no syslog prefix is added. Each line must contain one JSON object. Profound does not accept the default space-separated log format or a JSON array for this S3 integration.

Use uncompressed output or gzip. Gzip filenames must end in `.gz` so Profound can decompress them. Do not enable PGP encryption, Snappy, or Zstandard for this setup.

**4. Filter to your domain**

Attach a logging condition that matches the hostname you want to monitor. For example:

```text theme={null}
req.http.host == "example.com"
```

Replace `example.com` with your hostname. Include any additional hostnames you intend to monitor in your condition.

**5. Enable delivery**

Once Profound confirms your source is ready, save the endpoint and activate the service version. Generate traffic to the configured domain and allow time for Fastly to deliver a log file.

The endpoint's **Period** setting controls how frequently files are finalized. Check that setting when estimating when your test traffic will arrive.

## Verify ingestion

Open Agent Analytics for your domain and select a date range covering your test requests. Confirm that expected agent traffic appears after delivery and processing.

If data is missing, contact support with your domain, Fastly service ID, endpoint name, request time, and any delivery error or S3 file path. Do not include credentials.

## Troubleshooting

| Problem | What to check |
| - | - |
| Fastly cannot assume the role | Verify the role ARN and the Fastly account ID supplied during setup. Ask Profound to check the role's trust configuration. |
| S3 reports AccessDenied | Check the bucket, prefix, regional endpoint, and encryption setting against your assigned values. |
| Delivery succeeds but data is missing | Confirm source readiness, the `host` field, and your selected date range. |
| Records cannot be processed | Use the single-line JSON template and Blank log line format. Check compression settings. |
| No files arrive | Confirm that the service version is active, traffic matches the condition, and the file delivery period has elapsed. |

## Additional Resources

* [Fastly: Log streaming to Amazon S3](https://www.fastly.com/documentation/guides/integrations/logging-endpoints/object-and-cloud-storage/log-streaming-amazon-s3/)
* [Fastly S3 endpoint settings](https://www.fastly.com/documentation/reference/api/logging/s3/)
* [Fastly HTTPS integration](/agent-analytics/fastly)
* [Contact Profound support](mailto:support@tryprofound.com)


# Google Cloud CDN
Source: https://docs.tryprofound.com/agent-analytics/gcp_log_sink

This documentation explains how to set up Google Cloud CDN log delivery to Profound Agent Analytics platform using a log sink.

## Overview

This integration uses a topic, subscription, and log sink to capture Google Cloud CDN logs and forward them to Profound Agent Analytics platform.

## Prerequisites

* A GCP account with Google Cloud CDN enabled

* Access and edit permissions to the GCP project

* API endpoint details for the Profound Agent Analytics platform

## Configuration

![GCP setup](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/gcp_setup.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=afcaea65e6de8c8cf41bf68e5da11fe4)

**1. Step 1**

Sign into the GCP Console and navigate to `Topics`, you can search for 'topics' in the search bar.

**2. Step 2**

Create a new topic by clicking the `Create Topic` button near the top.

![Create a topic](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/create_topic.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=b562468b14ddb5f2958332a7cf80ded3)

Give your topic a name and click `Create`

**3. Step 3**

Navigate to the `Subscriptions` tab on the left sidebar. Click the `Create Subscription` button near the top.

![Create subscription](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/create_a_subscription.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9dd8d288cad258acaed3fba9b4fc23f2)

1. Give your subscription a name.
2. Select the topic you created in step 2.
3. Use delivery type `Push`
   * Enter `https://artemis.api.tryprofound.com/v1/logs/gcp_log_sink` as the endpoint
   * Enable authentication
   * Paste the service account this subscription will use into the `Service account` field
   * Leave `Audience` blank
   * Enable payload unwrapping

**4. Step 4**

Add retention policy configurations at your discretion.

![Retention policy](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/subscription_retention_policy.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=99f1706b09abb9289f362878cd3c0239)

**5. Step 5**

Click `Add a Transform` button. Enter `profound_agent_analytics_reduce` as the function name.

Paste the following code into the function body:

```javascript theme={null}
/* This UDF is used to filter the unused fields before sending to Profound.
 * 
 * @param {Object} message - The message to reduce.
 * @param {Object} metadata - The metadata of the message.
 * @returns {Object} The reduced message.
*/

function profound_agent_analytics_reduce(message, metadata) {
    // parse the message
    const data = JSON.parse(message.data);

    // remove unused fields
    delete data['insertId'];
    delete data['jsonPayload'];
    delete data['logName'];
    delete data['receiveTimestamp'];
    delete data['resource'];
    delete data['severity'];
    delete data['spanId'];
    delete data['trace'];

    message.data = JSON.stringify(data);

    return message;

}
```

> [!NOTE]
>
> This function is used to drop extraneous fields before sending to Profound. This reduces log export volume, improves performance, and enhances data security.

![Transform function](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/subscription_transform.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=30c80cbdc54364c310f5b1aaeea9477d)

Click `Validate` to ensure the function is valid.

**6. Step 6**

Add additional configurations at your discretion.

![Additional configurations](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/subscription_optional_config.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=c863a14ff66be1824ce1878df86bee24)

Click `Create` to create the subscription.

**7. Step 7**

Navigate to the `Log Router` tab by searching 'log router' in the search bar. Click the `Create sink` button near the top.

Give your sink a name and optional description.

**8. Step 8**

In the `Sink destination` section, select `Cloud Pub/Sub topic` as the `sink service`. In the `sink destination` field, select the topic you created in step 2.

![Sink service](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/sink_service.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=fc04900889359ece1c61f29f3c0f6645)

**9. Step 9**

Add the following filters to your sink in the `Choose logs to include in sink` section.

> [!WARNING]
>
> Make sure to replace the placeholders with your real load balancer and url map names. **This is only an example.**

```toml theme={null}
resource.type="http_load_balancer"
resource.labels.forwarding_rule_name="MY-FORWARDING-RULE-NAME"
resource.labels.url_map_name="MY-URL-MAP-NAME"
```

![Sink filters](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/sink_filters.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=5b2447b42a4b3bb85515a7fff4f9c275)

Finally, click `Create sink` at the bottom.

> [!NOTE]
>
> If you have multiple load balancers and url maps, you can modify the filters to include multiple resources like:
>
> ```toml theme={null}
> resource.type="http_load_balancer"
> AND (
> (resource.labels.forwarding_rule_name="MY-FORWARDING-RULE-NAME"
> AND resource.labels.url_map_name="MY-URL-MAP-NAME")
> OR
> (resource.labels.forwarding_rule_name="MY-FORWARDING-RULE-NAME-2"
> AND resource.labels.url_map_name="MY-URL-MAP-NAME-2")
> )
> ```

**10. Step 10**

> [!WARNING]
>
> If you are registering multiple GCP domains with Profound, you will need to create a separate service account for each domain. The account is used to decode the JWT token and confirm the source domain.

Navigate back to Profound. On the Agent Analytics onboarding page, paste the same service account email you used in step 3 into the `GCP Service Account` field. Ensure you click `Save` after pasting the email.

![Service account](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/gcp/gcp_setup.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=afcaea65e6de8c8cf41bf68e5da11fe4)

Click `Continue` to complete the onboarding process.

> [!TIP]
>
> That's it! Google Cloud CDN will now send real-time logs to Profound. Data should begin appearing in your dashboard within a few minutes.

## Troubleshooting

* Ensure you have shared the same service account email with Profound that you used to create the subscription.

* Check the `Logs` tab in Profound to verify the logs are being received.

* Verify the service account has the necessary permissions to access the topic as pub/sub.

* Verify the topic is created and active.

* Verify the subscription is created and active.

## Additional Resources

* [Third Party Log Routing Documentation](https://docs.cloud.google.com/logging/docs/export/pubsub#integrate-thru-pubsub)

* [Service Account Management Documentation](https://docs.cloud.google.com/iam/docs/create-service-agents#grant-role)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for API-related questions


# Netlify Integration
Source: https://docs.tryprofound.com/agent-analytics/netlify

This guide explains how to configure Netlify Log Drains to forward your application logs to Profound Agent Analytics platform for comprehensive monitoring and analysis.

> [!NOTE]
>
> Netlify Log Drains are only available on Enterprise plans.

## Overview

Netlify General HTTP Endpoint Log Drains enable you to stream your application logs to external services for enhanced monitoring and analytics. By integrating with Profound's Agent Analytics platform, you can gain deeper insights into your site's performance, track visitor behavior, and monitor AI agent interactions in real-time.

## Prerequisites

* A Netlify account with Enterprise plan

* Administrative access to your Netlify team settings

* Your Profound Log Ingestion Token (provisioned during onboarding)

* A deployed Netlify site with production traffic

## Configuration

![Netlify Detected](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/netlify/netlify_setup.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=e9bd35718c1d642d9c8499aae3bb1f6b)

**1. Step 1**

Navigate to your project in the Netlify dashboard. On the left, click **Logs > Log Drains**

![Log Drains](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/netlify/log_drains.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=403dc144e903f43f448cbcdd2a779ecd)

**2. Step 2**

Click **Enable Log Drain** and select `General HTTP Endpoint`

![Http Endpoint](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/netlify/http_endpoint.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=f4db6bad9062aa6106f8003f57be678b)

**3. Step 3**

Prepare your log drain. Uncheck:

* **Function logs**
* **Edge function logs**
* **Deploy logs**
* **WAF logs**
* **Exclude personally identifiable information (PII)**

You should only use **Traffic Logs** for the log drain.

Prepare your endpoint and authorization:

* URL: `https://artemis.api.tryprofound.com/v1/logs/netlify`
* Authorization Header: `Bearer bot_xxx`

Leave **JSON** selected in the **Log Drain Format**.

![Configure Http Endpoint](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/netlify/configure_endpoint.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=a0ba4afdaec87c3c26193e447b594039)

> [!NOTE]
>
> Your configuration should match the screenshot above.

**4. Step 4**

Click **Connect** to connect the log drain.

> [!WARNING]
>
> If you see the following error:
>
> `incorrectly configured GetURL, validation request returns status code 400`
>
> double check you have prefixed your token with `Bearer `

![Configuration Complete](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/netlify/configuration_complete.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9aa26eb73af4ec97e9c0144de06d4b49)

> [!TIP]
>
> That's it! Netlify will now send real-time logs to Profound. Data should begin appearing in your dashboard within a few minutes.

## Troubleshooting

* If logs aren't appearing in your Profound dashboard, verify your endpoint targets `/v1/logs/netlify`

* For authentication or Log Ingestion Token issues ensure you have prefixed the token with `Bearer `

## Additional Resources

* [Netlify Log Drains Documentation](https://docs.netlify.com/monitor-sites/log-drains/?monitoring-providers=general)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for integration assistance

## Security Considerations

* Store Log Ingestion Tokens securely in environment variables

* Regularly rotate your Log Ingestion Token

* Monitor log drain activity for unusual patterns

* Use HTTPS endpoints for all log forwarding


# About Agent Analytics
Source: https://docs.tryprofound.com/agent-analytics/overview

Gain unprecedented visibility into how AI systems interact with your website

## Overview

Profound Agent Analytics provides enterprise-grade visibility into how AI systems interact with your digital presence. As AI search engines and assistants become primary discovery channels, understanding and optimizing for AI visibility is crucial for modern digital strategy.

## The AI Analytics Challenge

### The Rise of AI Web Interactions

AI systems are revolutionizing how information is discovered and consumed online. Search engines like ChatGPT, Claude, and Google Gemini don't just index your content: they interpret it, synthesize it, and present it to users in entirely new ways.

### Why Traditional Analytics Fall Short

Traditional web analytics tools were designed for human visitors who:

* Execute JavaScript

* Maintain cookies and sessions

* Follow predictable browsing patterns

AI systems operate differently:

* They bypass JavaScript execution

* Don't maintain session state

* Access content through specialized crawlers

* Exhibit varying behavioral patterns based on their purpose (indexing, training, or real-time assistance)

This fundamental mismatch makes traditional tracking methods ineffective for monitoring AI interactions.

## The Profound Solution

### Architecture Built for AI

Profound's analytics platform is engineered specifically for tracking AI systems:

* **Server-Side First**: Implementation at the infrastructure level ensures complete visibility of AI interactions

* **Edge Integration**: Direct CDN integration captures requests before caching, providing the most accurate data

* **Advanced Bot Detection**: Sophisticated algorithms differentiate between various types of AI crawlers

* **High-Performance Backend**: Columnar OLAP databases enable real-time analysis of billions of requests

### Comprehensive AI Visibility

Track interactions from major AI platforms:

* OpenAI (ChatGPT)

* Anthropic (Claude)

* Google (Gemini/VertexAI)

* Perplexity

* Microsoft (Bing)

* Apple

* Meta

* DeepSeek

* Emerging AI systems

### Real-Time Intelligence

Monitor AI interactions as they happen:

* Track which content AI systems access

* Understand how your content appears in AI responses

* Identify content gaps and optimization opportunities

* Analyze AI crawling patterns and frequencies

## Implementation Approaches

### CDN Integration

For websites using Content Delivery Networks, we recommend integrating at the CDN level:

* **Maximum Accuracy**: Capture requests before caching occurs

* **Complete Coverage**: Track all AI interactions, including cached content

* **Simplified Setup**: Native integration with major providers:

### Roadmap

> [!NOTE]
>
> Roadmap timelines are estimates and subject to change as Profound continues to expand its integration coverage.

#### 2026

* General AWS S3 Bucket Integration (Q3 2026)

* Microsoft Azure CDN (Q4 2026)

### Direct Server Integration

For applications without CDN infrastructure:

* **API Integration**: Direct integration with your application servers

* **Log Ingestion**: Process server logs for AI interaction data

* **Custom Solutions**: Tailored implementations for unique infrastructures

## Key Features

### AI Crawler Analytics

* Detailed crawler identification and classification

* Real-time tracking of crawl events

* Pattern analysis and trend monitoring

* Coverage analysis and gap detection

### Performance Metrics

* Response time analysis

* Content accessibility metrics

* Cache effectiveness monitoring

* Geographic distribution of AI access

### Content Visibility

* Page-level AI interaction tracking

* Optimization recommendations

* Content effectiveness scoring

### Advanced Reporting

* Trend analysis

* Export capabilities

## Technology Stack

Profound is built on enterprise-grade technology:

* **Detection Engine**: Advanced machine learning models for accurate bot classification

* **Data Processing**: High-performance columnar databases for real-time analytics

* **API Layer**: RESTful APIs for seamless integration

* **Security**: Enterprise-grade encryption and access controls

## Getting Started

Ready to gain visibility into your AI presence?

> **[Integration Guides](/agent-analytics/self-serve-onboarding)**
>
> Step-by-step guides for implementing Profound Analytics

> **[Support](mailto:support@tryprofound.com)**
>
> Connect with our integration specialists


# Page type methodology
Source: https://docs.tryprofound.com/agent-analytics/page-type-methodology

How Profound classifies website pages for Agent Analytics and benchmarking.

Profound classifies website URLs into page types so Agent Analytics can show how AI traffic and benchmarking performance differ across kinds of content.

Page types are used in page-level reporting and benchmarking views. They help compare similar pages against similar pages instead of mixing fundamentally different URL types, such as blog posts, product pages, documentation, pricing pages, and technical infrastructure.

## How page types are assigned

Profound classifies URL patterns, not every individual URL independently. For each website, Profound studies the host's URL structure, groups similar paths like `/blog/`, `/docs/`, `/products/*`, or `/pricing`, and assigns durable page types to those reusable patterns.

Once patterns are saved, future URLs are tagged deterministically by matching their host and path against the active mappings.

When a path is ambiguous or risky, Profound may use page evidence such as the title, meta description, headings, and visible page text to make a second-pass classification decision. This helps avoid relying on URL shape alone.

## Page type definitions

| Page type | Definition |
| - | - |
| `blog_editorial` | Informational or editorial content such as articles, guides, thought leadership, news, press releases, case studies, glossary pages, research pages, media posts, and SEO content. |
| `product_page` | Pages where a user evaluates, discovers, compares, or selects an offering. This includes product, feature, solution, ecommerce product detail, category, marketplace, catalog, listing, inventory, integration, and template pages. |
| `docs_support` | Help, support, and documentation pages. This includes technical docs, help center articles, API references, manuals, troubleshooting pages, warranty pages, and support content. |
| `landing_page` | Campaign-specific or conversion-focused pages, such as promotional pages, seasonal pages, partner campaigns, paid acquisition pages, product launch pages, event or webinar registration pages, trial pages, sign up pages, and lead-capture pages. |
| `brand_about` | Pages about the company, brand, organization, or physical presence. This includes about pages, company pages, mission, team, leadership, careers, contact, locations, store locators, and investor pages. Home, security, and trust pages may classify here when supported by host-specific mappings or page evidence. |
| `pricing` | Pricing, plans, subscription tiers, packages, and plan comparison pages. |
| `ugc_page` | Public pages created or owned by users, customers, members, creators, vendors, or community participants rather than the brand itself. Examples include public profiles, creator pages, storefronts, vendor profiles, forum threads, community questions, reviews, and donation or tip pages. |
| `technical_infra` | Machine-readable or infrastructure URLs used for crawling, indexing, AEO/SEO, APIs, tracking, or site operation. Examples include `robots.txt`, `llms.txt`, sitemaps, manifests, service workers, XML/JSON feeds, API endpoints, tracking endpoints, metrics endpoints, CDN/proxy URLs, and similar infrastructure files. |
| `file_asset` | Direct downloadable or static file URLs, such as PDFs, images, videos, ZIP files, spreadsheets, presentations, documents, calendar files, catalogs, reports, and other file-extension-based assets. |
| `other` | Utility, legal, transactional, or unmatched pages that should not be forced into another page type. Includes privacy, terms, login, sign up, account, cart, checkout, and internal search pages. Also used as the fallback when no mapping confidently classifies the page. |

## Fallback behavior

Profound uses a closed taxonomy: every page used in Agent Analytics benchmarking resolves to a page type.

If no active host-specific mapping matches a URL, Profound applies generic fallback rules for common patterns such as `/blog/`, `/docs/`, `/pricing`, file extensions, infrastructure files, login pages, and legal pages.

If neither host-specific mappings nor generic fallback rules confidently classify the page, the page resolves to `other`. This prevents ambiguous URLs from being over-classified while preserving them for reporting.

## How page types are used in benchmarking

Benchmarking uses page types to compare pages against the appropriate peer distribution. For example, a product page should be compared against product pages, and a docs page should be compared against docs/support pages.

Some benchmarking views show all page types together, while page-specific views may filter to a selected page type. The `other` page type remains available in metrics, but it should be interpreted as a utility or fallback bucket rather than a strategic content category.

## Limitations

Page type classification is based on URL patterns and available page evidence. Some sites use ambiguous URL structures, reuse paths for multiple content types, or change templates over time. In those cases, classification may require review and mapping updates.


# Getting Started
Source: https://docs.tryprofound.com/agent-analytics/self-serve-onboarding

Set up Agent Analytics in just a few clicks

## Overview

Agent Analytics is Profound's powerful tracking solution that provides comprehensive insights into user interactions on your website. In this guide, you'll learn how to quickly add and configure a domain for tracking, allowing you to monitor traffic patterns, user behavior, and performance metrics.

Setting up Agent Analytics takes just a few minutes and requires minimal technical expertise. Once configured, you'll gain access to an intuitive dashboard that displays real-time data, helping you make informed decisions to optimize your website's performance and enhance the user experience.

## Prerequisites

* A Profound account

* Access to your domain's hosting provider (for example, Cloudflare, Vercel, etc.)

## Onboarding

**1. Step 1**

Navigate to the Agent Analytics tab on the left sidebar.

![My Website Tab](https://mintcdn.com/profound-37face47/Mo0VGkZHXuvRdZKy/images/self-serve/sidebar.png?fit=max\&auto=format\&n=Mo0VGkZHXuvRdZKy\&q=85\&s=aabcb6884453ae153d343776141c2b21)

**2. Step 2**

> [!NOTE]
>
> If you are configuring your first domain, you can skip this step.

Click on your existing domain near the top and select `+ Add a new website` from the dropdown menu or use the button in the top right.

![Add a New Website](https://mintcdn.com/profound-37face47/Mo0VGkZHXuvRdZKy/images/self-serve/add_new_website.png?fit=max\&auto=format\&n=Mo0VGkZHXuvRdZKy\&q=85\&s=c507aa995d22ec7d84c2a4281af48810)

**3. Step 3**

You will then see the following window. Use the input field to enter the apex domain you want to track.

![Enter Domain](https://mintcdn.com/profound-37face47/Mo0VGkZHXuvRdZKy/images/self-serve/registration_modal.png?fit=max\&auto=format\&n=Mo0VGkZHXuvRdZKy\&q=85\&s=db31e197e97c5bb725f3e7b0d152d428)

**4. Step 4**

Profound will attempt to automatically detect your domain's hosting provider. Ensure the correct provider is selected before continuing.

**5. Step 5**

To continue, select the appropriate provider and follow the setup guide.

> **[Cloudflare (Worker)](/agent-analytics/cloudflare_worker)**
>
> Connect your Profound account to your Cloudflare account with a lightweight worker

> **[Cloudflare (Logpush)](/agent-analytics/cloudflare_logpush)**
>
> Connect your Profound account to your Cloudflare account with Logpush

> **[Vercel (Connectable Account)](/agent-analytics/vercel_native)**
>
> Integrate your Vercel account with Agent Analytics in just a few clicks

> **[Amazon CloudFront](/agent-analytics/aws_firehose_cloudfront)**
>
> Use Amazon Data Firehose to deliver real-time logs to Profound

> **[Fastly](/agent-analytics/fastly)**
>
> Connect your Fastly traffic to Profound with a Custom HTTPS Endpoint

> **[Netlify](/agent-analytics/netlify)**
>
> Connect your Netlify traffic to Profound with Log Drains

> **[Akamai](/agent-analytics/akamai)**
>
> Connect your Akamai traffic to Profound with DataStream2

> **[Google Cloud CDN](/agent-analytics/gcp_log_sink)**
>
> Connect your Google Cloud CDN traffic to Profound with a log sink

> **[WordPress Plugin](/agent-analytics/wordpress/introduction)**
>
> Use the WordPress plugin to send request logs to Profound

> **[Custom Integration](/agent-analytics/custom)**
>
> Build your own custom integration with standardized log drain formats

### Need Help?

Our team is here to support you every step of the way. For personalized assistance with your integration, reach out through your dedicated Slack channel.

> **[FAQ](/agent-analytics/faq)**
>
> Frequently Asked Questions about Agent Analytics and Google Analytics

> **[Profound Status](https://status.tryprofound.com/en)**
>
> Monitor our service health and stay updated on system performance in real-time

> **[Contact](mailto:support@tryprofound.com)**
>
> Work with our team to design and build integrations tailored to your unique infrastructure needs

## Troubleshooting

* Ensure you have configured the proper domain you want to track

* Ensure you selected your hosting provider

* Verify you have completed all the steps in the relevant documentation

* Be aware it may take several minutes to see data

* Reach out to [support](mailto:support@tryprofound.com)


# Shopify
Source: https://docs.tryprofound.com/agent-analytics/shopify

This documentation details 3 potential Shopify log drain integrations for Profound Agent Analytics platform.

## Overview

> [!NOTE]
>
> Please carefully read the documentation to determine which integration is best for your Shopify site. Not sure which to use? Please contact [support](mailto:support@tryprofound.com) for assistance.

This integration is designed to capture logs from a Shopify site. Shopify is unlike other integrations in that there is no out-of-the-box solution for log exports. We detail three potential integrations for Shopify:

1. Shopify via Nostra (Recommended)
2. Shopify Log Drain via Hydrogen (Requires Hydrogen)
3. Shopify Trace Exporter (Requires Hydrogen)

## Prerequisites

* A Shopify site

* Access and management permissions to the Shopify site

* If connecting via **Nostra**, you must be a customer of both Profound and Nostra.

* If connecting via **Hydrogen**, you must have a Shopify Plus account.

## Configuration

![Shopify setup](https://mintcdn.com/profound-37face47/jkiOf0G56QSlBLCH/images/agent-analytics/shopify/shopify_setup.png?fit=max\&auto=format\&n=jkiOf0G56QSlBLCH\&q=85\&s=3ac484e400fcb734ec29cdea1c7c4608)

**Shopify via Nostra (Recommended)**

Learn how to integrate your Shopify site with Profound Partner Nostra

> [!NOTE]
>
> Integrating your Shopify site via Nostra is the recommended method with optimal analytics data. This method requires you to use Nostra's platform to track your Shopify site.

**1. Step 1**

Ensure you are a Nostra customer hosting your Shopify site on their platform.

**2. Step 2**

Contact your Nostra account representative to request a log drain integration.

**3. Step 3**

Once the integration is complete, you will be able to see your Shopify site in the Profound dashboard.

> [!TIP]
>
> That's it! Your Shopify site is now integrated with Profound via Nostra.

**Shopify Log Drain via Hydrogen (Requires Hydrogen)**

Learn how to integrate your Shopify site with Profound via Hydrogen.

> [!WARNING]
>
> This method requires your site uses Hydrogen and you have a Shopify Plus account.

**1. Step 1**

Ensure your Shopify site uses Hydrogen and you have a Shopify Plus account.

**2. Step 2**

Navigate to the Hydrogen Log Drain setup [documentation](https://shopify.dev/docs/storefronts/headless/hydrogen/logging#connecting-a-log-drain) and follow the instructions to configure the log drain.

**3. Step 3**

Select an `HTTP` log drain for generic integrations. Enter the following URL:

`https://artemis.api.tryprofound.com/v1/logs/shopify_hydrogen`

**4. Step 4**

Provide a header value `Bearer` with your Profound Log Ingestion Token. Your token should start with `bot_PROFOUND_LOG_INGESTION_TOKEN`.

**5. Step 5**

Click **Connect** to connect the log drain.

![Hydrogen Log Drain Configuration](https://mintcdn.com/profound-37face47/zh0y2yz9HzqrrnAo/images/agent-analytics/shopify/shopify_hydrogen.png?fit=max\&auto=format\&n=zh0y2yz9HzqrrnAo\&q=85\&s=89e5772d381aee0d39da5c638669a872)

> [!TIP]
>
> That's it! Your Shopify site is now integrated with Profound via Hydrogen.

**Shopify Trace Exporter (Requires Hydrogen)**

Learn how to integrate your Shopify site with Profound via the Trace Exporter.

> [!NOTE]
>
> This method provides sampled tracing data, not a full log export.

> [!WARNING]
>
> This method requires your site uses Hydrogen and you have a Shopify Plus account.

**1. Step 1**

Ensure your Shopify site uses Hydrogen and you have a Shopify Plus account.

**2. Step 2**

Navigate to the Hydrogen Trace Export setup [documentation](https://shopify.dev/docs/storefronts/headless/hydrogen/trace-exports#configure-trace-exports) and follow the instructions to configure trace exports.

**3. Step 3**

Select an `HTTP` trace export for generic integrations. Enter the following URL:

`https://artemis.api.tryprofound.com/v1/logs/shopify_hydrogen`

**4. Step 4**

Provide a header value `x-api-key` with your Profound Log Ingestion Token. Your token should start with `bot_PROFOUND_LOG_INGESTION_TOKEN`.

**5. Step 5**

Click **Save** to save the trace export configuration.

> [!TIP]
>
> That's it! Your Shopify site is now integrated with Profound via the Trace Exporter.

## Limitations of Hydrogen-Based Integrations

> [!WARNING]
>
> **Important**: Shopify's Hydrogen platform does not expose client IP addresses in log drains or trace exports. This is a platform limitation, not a Profound limitation.

### The Issue

Profound's Agent Analytics uses a multi-step verification process to accurately classify traffic, distinguishing real users from bots, crawlers, and automated agents. A key input to this process is the client IP address.

Shopify's Hydrogen runtime, hosted on Oxygen, strips client IP information from logs and traces before they reach external consumers. Because Profound never receives IP data through these integrations, the multi-step verification process cannot be performed.

This is an unavoidable platform constraint imposed by Shopify. There is no configuration, workaround, or API option to include IP addresses in Hydrogen log drains or trace exports.

### Impact on Analytics

Without client IP addresses, Profound cannot:

* Perform IP-based reputation checks against known bot networks
* Correlate requests across sessions for behavioral analysis
* Apply geographic and network-level classification signals
* Execute the full multi-step verification pipeline

Analytics data from Hydrogen-based integrations will still be collected and displayed, but classification accuracy is reduced compared to integrations that provide complete request metadata.

### Recommendation

We recommend integrating your Shopify site via **Nostra** (option 1 above). Nostra's platform uses a Cloudflare integration under the hood, which provides full request metadata, including client IP addresses. This means the Nostra integration does not suffer from any of the limitations described above, and Profound can run the complete multi-step verification process.

Using the Nostra integration requires you to be a customer of both **Profound** and **Nostra**.

> [!NOTE]
>
> This limitation is not unique to Agent Analytics. Any third-party service consuming Hydrogen log drains or trace exports faces the same constraint: Shopify simply does not include client IP data in these exports. The recommended path for complete traffic visibility on Shopify is through a CDN-native integration like Nostra.


# Vercel Integration (Connectable Account)
Source: https://docs.tryprofound.com/agent-analytics/vercel_native

This guide explains how to set up a Vercel Log Drain via the Profound Agent Analytics Integration

## Overview

Vercel Log Drains provide a way to forward your Vercel application logs to [Profound Agent Analytics](https://vercel.com/cooper-square-technologies/~/integrations/profound) platform. By integrating with Profound's Agent Analytics platform, you gain access to comprehensive monitoring features including automated agent detection and real-time performance analysis.

## Prerequisites

Before beginning the integration process, ensure you have:

* A Vercel account with administrative access

* A Profound account with an active subscription

* Administrative access to your Vercel project settings

## Configuration

![Vercel Detected](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/vercel_setup.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9cad98d8b0967c2d6d55f8ffb7810633)

**1. Access Project Settings**

Log into your Vercel Dashboard and navigate to your project's settings at **Settings** > **Log Drains**.

![Settings](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/settings_log_drains.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=50597b2fcc00c1607de43afb9da5b943)

**2. Add Log Drain**

In the top right, press the **Add Log Drain** button. This will open a list of integrations. Search for "Profound Agent Analytics" and select it. Press **Continue**.

![Profound AA Integration](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/profound_aa_integration.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=465f19a2ab530c9948b074d975712abb)

**3. Select Project Access**

You will now be prompted to configure the log drain. Select **Specific Projects** and add the projects you would like to track via Profound's Agent Analytics service.
Press **Connect Account** when ready.

> [!NOTE]
>
> All projects you give access to will be configured for Profound's Agent Analytics.
> It is critical you select only the projects you wish to track in Profound.
>
> **Do not share access to projects you do not intend to track via Profound.**

![Select Projects](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/select_specific_projects.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=37f68d4f1c8a494f141c40add6ad3f06)

**4. Validate Projects and Domains**

You will be redirected to Profound to finish the configuration.

> **Logged out?**
>
> If you are logged out, you will see a login prompt.
>
> ![Logged Out](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/logged_out.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=b1e298f8f29bf67402ffde47a03b2f7f)
>
> Simply enter your email, complete any OAuth or SSO steps, and you will be returned to the page below.

> **No Profound account?**
>
> Agent Analytics requires access to the Profound platform and thus a Profound account. If you do not have an account, press **I'm new to Profound**.
>
> ![Logged Out](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/logged_out.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=b1e298f8f29bf67402ffde47a03b2f7f)
>
> This will open the pricing page in a new tab. Press **Start** to proceed.
>
> ![Pricing](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/pricing.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=66ecfd365afad858ac1dfd1ee31db5cc)
>
> Because a Profound account is required, you must complete this process **before** completing the integration.
>
> ![Onboard](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/create_account.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=8a8e9e7bce7b42b487e57091454b4015)

Double check the projects and domains in the list. These are the domains that will be tracked via Agent Analytics.
Likewise, ensure your Profound Organization is selected.
If you are a member of many organizations, double check the proper one is selected.

Press **Continue** when ready.

![Verify Projects and Domains](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/verify_projects_and_domains.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=0ddfa6ff5c47d4bab470ad46b57484d8)

**5. Finish Up**

After pressing **Continue**, Vercel will verify the integration.

![Vercel Verify](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/completing_install.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=edbd937994c19c127019b1f7dfd8d7a7)

You will be returned to the integration homepage.

![Integration Page](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/integration_page.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=0244a67a55a472247a157fba152b5283)

> [!TIP]
>
> That's it! Vercel will now send real-time logs to Profound. Data should begin appearing in your dashboard within a few minutes.

## Remove Integration

**1. Access the Integration Settings**

There are a few ways to disable the integration.

1. Option 1

   Select the project you configured. Navigate to **Settings** > **Log Drains**.

   ![Log Drains](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/disable_log_drain.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=9b553f4cd4d9795f6c5bf39af9e861be)

   Press **Manage**.

2. Option 2

   Navigate to **Project** > **Integrations** in the top toolbar.

   ![Manage Integrations](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/manage_integration.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=4c0cdf1cbdd4a09f35fc69986f6ee224)

   Press **Manage**.

**2. Integration Settings**

Either option will redirect you to the integration page.

![Integration Page](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/integration_page.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=0244a67a55a472247a157fba152b5283)

Press **Settings** in the navigation tabs at the top.

**3. Uninstall Integration**

To remove the integration, scroll to the bottom of the **Settings** page and press **Uninstall Integration** and complete the steps required by Vercel.

![Uninstall](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/vercel_integration/uninstall_integration.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=98802c92eabb07c3cd261796f6635df0)

> [!TIP]
>
> You have now removed the integration. Logs will no longer be sent to Profound Agent Analytics.

## Additional Resources

* [Integration Documentation](https://vercel.com/cooper-square-technologies/~/integrations/profound-test)

* [Vercel Log Drain Documentation](https://vercel.com/docs/observability/log-drains)

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for integration assistance


# WordPress Plugin Installation
Source: https://docs.tryprofound.com/agent-analytics/wordpress/installation

This guide explains how to install and configure the Agent Analytics plugin for WordPress to send HTTP request logs to the Profound platform

> [!WARNING]
>
> The WordPress plugin can have limited functionality by nature of the WordPress platform. Please read the [limitations](/agent-analytics/wordpress/limitations) page to understand more.

## Overview

The Agent Analytics [WordPress plugin](https://wordpress.org/plugins/profound-agent-analytics/) enables automatic log collection from your WordPress site, forwarding visitor traffic data to the Profound Agent Analytics platform. The plugin operates transparently in the background, capturing only public traffic while excluding all administrative and system activity.

## Prerequisites

Before beginning the installation, ensure you have:

* A WordPress site running version 6.0 or higher

* Administrative access to your WordPress dashboard

* A Profound account with a Log Ingestion Token

* PHP 7.4 or higher on your hosting environment

## Installation

![Wordpress Detected](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/wordpress/wordpress_setup.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=f42646e73b1c7a0b855229e799d0d5a4)

**1. Step 1**

Log into your WordPress admin dashboard and navigate to **Plugins** in the left sidebar.
Click **Add New** at the top of the Plugins page. Search for "Profound Agent Analytics" or visit the [WordPress plugin page](https://wordpress.org/plugins/profound-agent-analytics/) to download the plugin.

![WordPress Plugins Navigation](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/wordpress/install.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=d5b2c8c215318441535107aea7c5c8b9)

Click **Install Now** to install the plugin.

**2. Step 3**

After installation completes, click **Activate** to activate the plugin.

Alternatively, navigate to **Plugins** → **Installed Plugins**, find "Agent Analytics" in the list, and click **Activate**.

![Activate Plugin](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/wordpress/activate.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=295589131a4bab05245026c263d1ca7a)

**3. Step 4**

Once activated, navigate to **Settings** → **Agent Analytics** in your WordPress admin menu.

**4. Step 5**

In the General settings tab:

1. Check the **Enable Logging** checkbox to activate log collection

2. Enter your Profound Log Ingestion Token (starts with `bot_`) in the **Log Ingestion Token** field

3. Leave the **API Endpoint** at its default value unless instructed otherwise

4. Optionally adjust the **Batch Size** and **Send Interval** settings.

   ![API Configuration](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/wordpress/settings.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=bba8eb1b9fb0ec5f5b70f27300ea1b16)

> [!WARNING]
>
> Your Log Ingestion Token is sensitive information. For production sites, consider using environment variables instead of database storage. See the [security documentation](/agent-analytics/wordpress/security) for details.

**5. Step 6 (Optional)**

Navigate to the **Privacy** tab to adjust filtering options:

* **Query Parameter Redaction**: Add any additional sensitive parameters to redact (default includes passwords, tokens, cards)

* **Path Filtering**: Add custom paths to exclude from logging

* **Skip Options**: Toggle exclusion of admin, cron, AJAX, or REST API requests

  ![Privacy Settings](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/wordpress/privacy.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=166226249d5a8483233df5348a382640)

> [!NOTE]
>
> The plugin automatically excludes all WordPress admin activity and system paths. These settings allow additional customization beyond the defaults.

**6. Step 7**

Click **Save Changes** at the bottom of the settings page.

Navigate to the **Status** tab to verify your configuration:

1. Check that the **Queue Status** shows as "Active"

2. Click **Send Test Event** to verify connectivity

3. Confirm you see "Test event sent successfully"

   ![Status Verification](https://mintcdn.com/profound-37face47/tg0qbLRw_vFMInTl/images/agent-analytics/wordpress/status.png?fit=max\&auto=format\&n=tg0qbLRw_vFMInTl\&q=85\&s=25dfddfd4b78cdb856107eb25eba4762)

> [!TIP]
>
> If the test fails, double-check your Log Ingestion Token and ensure your server can make outbound HTTPS requests to `artemis.api.tryprofound.com`.

**7. Step 8**

After configuration:

1. Visit a few pages on your public website (not logged into WordPress admin)

2. Return to **Settings** → **Agent Analytics** → **Status**

3. Check that the **Next automatic send** shows a time within the next minute

4. Verify the **Queue size** increases as you browse, then decreases after automatic sending

That's it! You have successfully configured Agent Analytics for your WordPress site. Data should begin appearing in your Profound dashboard within minutes.

## Post-Installation Verification

To confirm everything is working correctly:

1. **Check the Status Tab**: The queue should periodically drain (every 60 seconds by default)

2. **Monitor Your Dashboard**: Log into your Profound account and verify data is appearing

3. **Review Processing Schedule**: Ensure "Next automatic send" updates regularly

## Troubleshooting

### Logs Not Sending

* Verify your Log Ingestion Token is correct and starts with `bot_`

* Check that **Enable Logging** is checked in General settings

* Ensure WordPress Cron is running (visit your site frontend to trigger it)

* Look for any error messages in the Status tab

### Test Event Fails

* Confirm your server allows outbound HTTPS connections

* Check if a firewall is blocking connections to `artemis.api.tryprofound.com`

* Verify your Log Ingestion Token has no extra spaces or characters

### Queue Growing But Not Sending

* The automatic sending requires site visits to trigger WordPress Cron

* Consider setting up a real cron job for consistent processing

* Click **Re-schedule Automatic Sending** if the schedule appears missing

### Wrong IP Addresses

* The plugin automatically detects proxy headers from CDNs and managed hosts

* If IPs appear incorrect, your host may use non-standard headers

* Contact support for assistance with custom IP detection

## Additional Resources

* [Security & Privacy Configuration](/agent-analytics/wordpress/security) - Detailed privacy settings and data handling

* [Plugin Overview](/agent-analytics/wordpress/introduction) - Complete feature documentation

* Contact [support@tryprofound.com](mailto:support@tryprofound.com) for integration assistance


# Introduction
Source: https://docs.tryprofound.com/agent-analytics/wordpress/introduction

Send request logs from your WordPress site to Agent Analytics for advanced traffic analysis

> [!WARNING]
>
> The WordPress plugin can have limited functionality by nature of the WordPress platform. Please read the [limitations](/agent-analytics/wordpress/limitations) page to understand more.

## Overview

The Agent Analytics WordPress plugin provides a seamless integration for WordPress sites to send HTTP request logs to the Profound Analytics platform. The plugin is designed with performance and privacy in mind, capturing only public visitor traffic while excluding all administrative and system activity.

> **Zero Performance Impact**
>
> Asynchronous queue system ensures page delivery is never blocked. Adds less than 1ms to request processing.

> **Public Traffic Only**
>
> Automatically excludes WordPress admin, REST API, and system paths. Only tracks real visitor traffic.

> **Privacy Controls**
>
> Redacts sensitive query parameters and excludes login pages. No cookies or client-side tracking.

> **Enterprise Ready**
>
> Supports environment variables for Log Ingestion Tokens and includes circuit breaker protection for reliability.

## How It Works

The plugin operates using a lightweight, non-blocking architecture:

1. **Capture** - Essential HTTP metadata is captured during request processing
2. **Queue** - Data is immediately stored in a local database queue with minimal overhead
3. **Batch** - Background process sends logs in efficient batches (up to 1,000 per request)
4. **Retry** - Failed sends are automatically retried with exponential backoff
5. **Protect** - Circuit breaker prevents cascade failures if the API is unavailable

## System Requirements

* WordPress 6.0 or higher
* PHP 7.4 or higher
* MySQL 5.7+ or MariaDB 10.2+
* Outbound HTTPS connectivity
* WordPress Cron enabled (or system cron configured)
* Sodium PHP extension (recommended for Log Ingestion Token encryption)

## Hosting Compatibility

The plugin is tested and optimized for:

* **Managed WordPress Hosts**: WP Engine, Kinsta, Flywheel
* **Cloud Platforms**: AWS, Google Cloud, Azure
* **Traditional Hosts**: SiteGround, Bluehost, GoDaddy
* **Local Development**: LocalWP, MAMP, XAMPP

> [!TIP]
>
> The plugin automatically detects and handles reverse proxy headers used by managed WordPress hosts and CDNs to ensure accurate IP detection.

## Next Steps

> **[Security & Privacy](/agent-analytics/wordpress/security)**
>
> Configure Log Ingestion Tokens, privacy settings, and filtering options

> **[Plugin Limitations](/agent-analytics/wordpress/limitations)**
>
> Understand the limitations of the WordPress plugin when used with a CDN for caching

> **[Installation Guide](/agent-analytics/wordpress/installation)**
>
> Step-by-step instructions for installing and activating the plugin

***


# CDN Caching Limitations
Source: https://docs.tryprofound.com/agent-analytics/wordpress/limitations

Understand the limitations of the WordPress plugin when used with a CDN for caching

> [!WARNING]
>
> **Important**: If your WordPress host uses a CDN for caching (Cloudflare, Fastly, etc.), the plugin will **only capture cache misses**, not cache hits. This is a platform limitation, not a plugin limitation.

### The Issue

Many managed WordPress hosts (WP Engine, Kinsta, Flywheel, etc.) place a CDN in front of your WordPress site for performance. When a visitor requests a page:

1. **Cache Miss** → Request passes through to WordPress → Plugin captures the request → Logged successfully
2. **Cache Hit** → CDN serves cached content directly → **Request never reaches WordPress** → Not logged by plugin

This means you'll only see a portion of your actual traffic: specifically, the traffic that wasn't served from cache.

#### Example

In this example, the site is hosted on WordPress Engine with WPE application cache layer enabled.

In this example, we are fetching the page for the first time, so it is a cache miss.

```shell theme={null}
curl -I "https://my-wp-site.com/some-page"


HTTP/2 200 OK
...
x-cache: MISS
# Cache missed, request reaches WordPress and is logged by the plugin
...
```

On the second request, the page is live in the cache, so it is served from the cache layer and never reaches WordPress.

```shell theme={null}
curl -I "https://my-wp-site.com/some-page"


HTTP/2 200 OK
...
x-cache: HIT: 1
# Cache hit, request never reaches WordPress and is not logged by the plugin
...
```

### Why This Happens

The plugin operates at the WordPress application layer. It can only log requests that actually reach your WordPress server. When a CDN serves cached content, the request terminates at the CDN's edge servers and never makes it to your WordPress installation.

**The complete solution** is to use CDN-native log drains, which Agent Analytics already fully supports. CDN logs capture *everything*: cache hits, cache misses, bot traffic, DDoS attempts, the complete picture.

### The Platform Problem

Unfortunately, most managed WordPress hosts like WP Engine and Kinsta do not provide customers access to configure their CDN log drains.
Profound has contacted these providers to request access for customers, they declined at this time. We are pursuing partnership opportunities with these providers to enable this functionality.

### Impact on Analytics

> **What Gets Captured**
>
> * Initial page visits (cache miss)
> * Dynamic/personalized pages that bypass cache
> * POST requests and form submissions
> * Admin-generated traffic to public pages
> * Pages with query parameters (often uncached)
> * Requests during cache purge/rebuild periods

> **What Gets Missed**
>
> * Subsequent visits to the same page (cache hits)
> * Popular pages with high cache hit rates (90%+ typical)
> * Static assets served from CDN (CSS, JS, images)
> * Bot traffic filtered or served at CDN edge
> * Attacks/DDoS blocked at CDN layer

> **Typical Cache Hit Rates**
>
> * **Well-optimized sites**: 85-95% cache hit rate
> * **Average sites**: 60-80% cache hit rate
> * **Dynamic/personalized sites**: 20-40% cache hit rate

### Workarounds

If you need complete traffic visibility, consider:

1. **Switch to a host that provides CDN control** - Many hosts (AWS, Google Cloud, etc.) allow you to configure your own CDN
2. **Request enterprise CDN access** - Some managed hosts provide log access to enterprise customers
3. **Accept partial data** - For bot detection and abuse monitoring, cache misses are often sufficient since bots typically don't benefit from cache
4. **Use Agent Analytics directly with a supported CDN** - If you have the ability to configure your own Cloudflare, Fastly, or other supported CDN, use our native integrations instead

> [!NOTE]
>
> This limitation is not unique to Agent Analytics. Any WordPress plugin attempting to log HTTP requests faces the same constraint: they can only see what reaches WordPress. The industry standard solution is CDN-native log collection, which we fully support for customers with CDN access.


# Security & Privacy Configuration
Source: https://docs.tryprofound.com/agent-analytics/wordpress/security

Configure data collection, filtering, and privacy settings for the WordPress plugin

## Privacy & Compliance

The plugin is designed to work with analytics while respecting user privacy:

* **No cookies**: Pure server-side implementation
* **No JavaScript**: No client-side tracking code
* **Data minimization**: Only HTTP metadata, no content
* **Automatic redaction**: Sensitive parameters removed
* **Public only**: Admin and user areas excluded

> [!WARNING]
>
> Full IP addresses are collected as they are essential for accurate bot detection and traffic analysis. If your jurisdiction requires IP anonymization, additional configuration may be needed.

## Data Collection

The plugin collects standard HTTP request metadata from public pages:

### Required Fields

The following fields are **required** for every log entry sent to Agent Analytics:

| Field | Format | Max Length | Description |
| - | - | - | - |
| `timestamp` | ISO 8601 | - | UTC timestamp of the request |
| `method` | String | 10 chars | HTTP method (GET, POST, etc.) |
| `host` | String | 255 chars | Request hostname |
| `path` | String | 2048 chars | URL path (without query string) |
| `status_code` | Integer | 100-599 | HTTP response status code |
| `ip` | String | 45 chars | Client IP address (supports IPv4/IPv6) |
| `user_agent` | String | 1024 chars | User agent string |

> [!NOTE]
>
> All required fields must be present for successful log ingestion. Missing fields will result in validation errors on the API side.

### Optional Fields

These fields are included when available but not required:

| Field | Format | Max Length | Description |
| - | - | - | - |
| `query_params` | Object | 100 keys | Query string parameters (after redaction) |
| `referer` | String | 2048 chars | HTTP referer header |
| `bytes_sent` | Integer | - | Response size in bytes |
| `duration_ms` | Integer | - | Request processing time |

> [!NOTE]
>
> The plugin **does not** collect: page content, form data, WordPress user information, admin activity, database queries, or any server-side application data.

## Automatic Path Filtering

### Always Excluded Paths

The following paths are **automatically excluded** from logging and cannot be overridden:

```
wp-admin         # All admin pages and subdirectories
wp-login.php     # Login page
wp-cron.php      # Cron jobs
admin-ajax.php   # Admin AJAX calls
xmlrpc.php       # XML-RPC interface
wp-json/wp/*     # WordPress core REST API
```

> [!WARNING]
>
> These exclusions use substring matching, so they work regardless of WordPress installation directory (e.g., `/blog/wp-admin/` is also excluded).

### Request Type Exclusions

The plugin automatically skips:

* **Admin-referred requests**: Any request with a referer containing `wp-admin`
* **REST API calls**: Requests with `rest_route=/wp/*` query parameter
* **CLI operations**: WP-CLI commands
* **Edit context**: REST requests with `context=edit` parameter

### Path Filtering Priority

Paths are evaluated in this order:

1. **Built-in exclusions** (wp-admin, wp-login.php, etc.)
2. **Custom deny paths** (user-configured patterns)
3. **Referer check** (requests from wp-admin)
4. **REST API check** (WordPress core API)

First match wins - if any check matches, the request is excluded.

## Query Parameter Redaction

### Default Redacted Parameters

The following query parameter keys are automatically redacted (replaced with `[REDACTED]`):

```
password    # Password fields
pass        # Alternative password fields
token       # Authentication tokens
secret      # Secret keys
card        # Credit card fields
cc          # Credit card abbreviation
ssn         # Social security numbers
```

> [!NOTE]
>
> Redaction uses substring matching. Any parameter containing these strings will be redacted (e.g., `user_password`, `api_token`, `card_number`).

### Redaction Example

Original query string:

```
?username=john&password=secret123&product=widget&api_token=abc123
```

After redaction:

```json theme={null}
{
  "username": "john",
  "password": "[REDACTED]",
  "product": "widget",
  "api_token": "[REDACTED]"
}
```

## Log Ingestion Token Security

### Storage Methods

The plugin supports three methods for Log Ingestion Token configuration, in order of security preference:

**Environment Variable (Most Secure)**

Add to your server environment:

```bash theme={null}
WP_ENV_AGENT_ANALYTICS_API_KEY=bot_PROFOUND_LOG_INGESTION_TOKEN
```

Or in `wp-config.php`:

```php theme={null}
define('WP_ENV_AGENT_ANALYTICS_API_KEY', 'bot_PROFOUND_LOG_INGESTION_TOKEN');
```

**WordPress Constant**

Add to `wp-config.php`:

```php theme={null}
define('AGENT_ANALYTICS_API_KEY', 'bot_PROFOUND_LOG_INGESTION_TOKEN');
```

**Database (Encrypted)**

Enter via WordPress admin. The key is:

* Encrypted using Sodium library (if available)
* Falls back to WordPress salts for encryption
* Stored in `wp_options` table
* Masked in UI (only last 4 characters shown)

> [!WARNING]
>
> For production environments, use environment variables or constants. Database storage should only be used for testing or when file access is restricted.

## IP Address Handling

### Detection Order

The plugin attempts to detect the real client IP in this order:

1. `HTTP_X_FORWARDED_FOR` - First IP in the list (for proxies/load balancers)
2. `HTTP_CF_CONNECTING_IP` - Cloudflare's real IP header
3. `HTTP_X_REAL_IP` - Alternative proxy header
4. `HTTP_TRUE_CLIENT_IP` - Cloudflare Enterprise header
5. `REMOTE_ADDR` - Direct connection (fallback)

> [!NOTE]
>
> The plugin automatically handles reverse proxies used by managed WordPress hosts (WP Engine, Kinsta, etc.) and CDNs (Cloudflare, Fastly).

### IP Privacy Considerations

* **No anonymization**: Full IP addresses are captured for accurate bot detection
* **Required for service**: IP data is essential for geographic analysis and reverse DNS lookups
* **GDPR compliance**: Consider your legal requirements for IP data collection

## Configuration Options

### General Settings

| Setting | Default | Description |
| - | - | - |
| **Enable Logging** | Off | Master switch for log collection |
| **API Endpoint** | `https://artemis.api.tryprofound.com/v1/logs/wordpress` | Destination for logs |
| **Batch Size** | 500 | Number of logs per send (50-1000) |
| **Send Interval** | 60 seconds | How often to process queue (30-600 seconds) |

### Privacy Settings

| Setting | Default | Options | Description |
| - | - | - | - |
| **Skip Admin** | On | On/Off | Exclude `/wp-admin/` requests |
| **Skip Cron** | On | On/Off | Exclude cron job requests |
| **Skip AJAX** | On | On/Off | Exclude admin AJAX calls |
| **Skip REST** | Off | On/Off | Exclude ALL REST API requests (not just core) |
| **Redact Query Keys** | `password,pass,token,secret,card,cc,ssn` | Comma-separated list | Parameters to redact |

### Advanced Settings

| Setting | Default | Description |
| - | - | - |
| **Custom Deny Paths** | Empty | Newline-separated path patterns (supports wildcards) |
| **Max Queue Size** | 100,000 | Maximum logs to store before dropping old entries |
| **Retention Days** | 7 | Days to keep unsent logs |
| **Circuit Breaker Threshold** | 3 | Consecutive failures before pausing |
| **Circuit Breaker Cooldown** | 30 minutes | Pause duration after threshold reached |

## Custom Path Filtering

### Wildcard Pattern Examples

Add custom path exclusions using these patterns:

```
/private/*           # Exclude all paths starting with /private/
*/temp/*            # Exclude paths containing /temp/
*.pdf               # Exclude all PDF files
/api/v1/internal/*  # Exclude internal API endpoints
staging.*           # Exclude staging subdomains
```

## Data Retention

### Plugin-Side Retention

* **Queue entries**: Deleted after successful send
* **Failed entries**: Retried up to 6 times with exponential backoff
* **Old entries**: Auto-cleaned after retention period (default 7 days)
* **Maximum queue**: Oldest entries dropped when limit reached (100,000)

### API-Side Processing

* **Successful logs**: Processed immediately by Agent Analytics
* **Validation errors**: Logged but entry still deleted from queue
* **Authentication failures**: Queue paused, circuit breaker activated

## Performance Characteristics

> **Request Impact**
>
> * **Processing overhead**: \< 1ms per request
> * **Memory usage**: Minimal (single array per request)
> * **Database operations**: One INSERT per request
> * **No output buffering**: Does not delay response delivery

> **Background Processing**
>
> * **Send interval**: Every 60 seconds (configurable)
> * **Batch size**: Up to 500 logs per send (configurable)
> * **Timeout**: 5 second maximum per batch
> * **Queue limit**: 100,000 entries maximum

> **Reliability Features**
>
> * **Automatic retry**: Exponential backoff on failures
> * **Circuit breaker**: Pauses after 3 consecutive failures
> * **Queue persistence**: Logs preserved during outages
> * **Lock mechanism**: Prevents concurrent processing

## Security Best Practices

* Store Log Ingestion Tokens in environment variables rather than the database for production sites.

* Ensure all sensitive areas of your site are excluded from logging.

* Check the Status tab regularly to ensure logs are sending properly.

* Use the "Send Test Event" feature to verify connectivity without affecting real data.

* Ensure your WordPress site uses HTTPS to protect data in transit to your server.

> [!NOTE]
>
> If you discover any security issues or need assistance with configuration, please contact [support@tryprofound.com](mailto:support@tryprofound.com) immediately.


# About Agents
Source: https://docs.tryprofound.com/agents/about

What Profound Agents are, the problem they solve, and their key features

> [!NOTE]
>
> **Available on**: all plans
>
> **Find it in**: Workspace > Agents

Agents let you hand recurring marketing work to Profound. You describe the job in plain English, give the Agent the tools it needs, and decide when it runs: on demand, on a schedule, or when something happens in your data, on the web, or in your apps.

AI Marketer (Aim) builds Agents for you from a description. When you need full manual control over every step, use Workflows, the node-based automations you build in the Workflow Builder. You can also combine the two: an Agent can run a Workflow for the part of a task that must run the same way every time.

> [!NOTE]
>
> If you're looking for information on the legacy node-based Workflow Agents, visit [About Workflows](/agents/workflows/about-workflows).

## The problem Agents solve

Profound shows you where your brand is visible in AI search, which sources answer engines cite, how they talk about you, and where they get the facts wrong. Before Agents, acting on those insights meant manual work: pulling numbers into spreadsheets, briefing a writer, drafting outreach, and checking back next week to see if anything changed.

Agents take that work on. An Agent can draft and optimize content for the topics where you're losing ground, write outreach to a source that repeats an inaccurate claim or drags your sentiment down, and send a report every Monday without anyone opening a dashboard.

Agents work with the tools your team already uses. Combine Profound data with insights from services like G2 or Google Analytics, save the results as Documents and Sheets in Profound, or publish them directly to your content management system (CMS). Learn more in [Tools in Agents](/agents/tools-in-agents).

## Key features

### One page to create, edit, and run

Each Agent has a single page. On the left is a chat with Aim, where you run the Agent and ask for changes. On the right is the Agent's settings panel. Select **Run history** or **Artifacts** at the top of the page to see past runs and the files the Agent produced. The changes you make apply to the Agent right away, and every run uses the current version.

### Agent configuration

The settings panel holds everything the Agent needs to run: triggers, inputs, instructions, outputs, tools, context, and advanced settings. Instructions are the body of the Agent: what to do, in what order, and what to avoid. You write them in a full text editor that supports Markdown, so you can use headings, lists, and code blocks. Type **/** to insert tools, inputs, outputs, documents, Skills, Workflows, and other items, or **@** to mention items the Agent already has. An Agent can also run a Workflow you've built as one of its steps, for parts of the task that must give the same result every time. Switch the **Instructions** card to **Execution steps** to see the list of steps Aim works out from your instructions. Learn more in [Building blocks of an Agent](/agents/building-blocks-of-agent).

### Run and refine in chat

Select **New run** to run the Agent. The run happens in the Agent's own chat, so you can watch the reasoning, the tool calls, and the results as they appear, and then ask follow-up questions about them. To change how the Agent works, describe the change in the same chat and Aim updates the Agent. Every run is saved under **Run history**.

### Automatic runs on triggers

Triggers start a run without anyone selecting **New run**. The available kinds are Schedule (a recurring time), a connected app (an event in that app, such as a new message in a Slack channel), Monitor (a change on the web, such as a competitor publishing a page), and Profound data (a metric crossing a threshold, such as a 10% drop in Visibility Score or a rise in inaccurate FactCheck claims). Learn more in [Triggers in Agents](/agents/triggers-in-agents).

### Sharing and version history

Agents are private by default. Select **Share** to invite teammates with **Can view** or **Can edit** access, or set **General access** to everyone in your organization. Profound saves versions of the Agent as it changes. Open **Version history** from the **...** menu to review earlier versions and restore one.

### Templates

Templates are Agents that Profound built for common jobs, such as weekly reports, content briefs, FAQ optimization, and brand mention digests. Select **Templates** on the Agents page to browse them. To share your own Agent as a template, submit it for your organization or for all Profound customers. Profound reviews every submission before it appears in the template gallery.

## Next steps

* Follow [Get started with Agents](/agents/getting-started) to build your first Agent.
* Learn about the [building blocks of an Agent](/agents/building-blocks-of-agent).
* Read the [Agents FAQ](/agents/faq).


# Agents and Workflows: when to use which
Source: https://docs.tryprofound.com/agents/agents-vs-workflows

How Agents and Workflows differ, when to use each, and how to run a Workflow inside an Agent

Automate a task in Profound with an [Agent](/agents/about), a [Workflow](/agents/workflows/about-workflows), or both. Both live on the Agents page in Profound, both run on a schedule or on demand, and both produce Documents, Sheets, and other outputs. To see only one kind on the Agents page, use the **Type** filter.

## What's the difference?

Agents take written instructions, connected tools, and triggers, and AI Marketer (Aim) works out the steps on each run.

Workflows are automations you build yourself in the Workflow Builder as a chain of nodes that runs the same way every time.

The main difference is control. With Agents, you describe what you want to achieve and leave the implementation details to Aim. With Workflows, you build every step yourself and control the order of steps, the tools, and the inputs and outputs.

## When to use which?

Use an Agent for most tasks. If you're new to Agents, it helps you skip the implementation details and focus on the task you want your Agent to accomplish.

Use a Workflow when you need the same output every run, when a step must run code or call an API in a precise way, or when you need granular control over which steps run and in what order.

Workflows also tend to cost fewer credits for high-volume tasks. An Agent reasons through each step, and that reasoning uses credits.

If you need both, combine them: build an Agent and have it run a Workflow for the part that must stay fixed.

## Use a Workflow inside an Agent

An Agent can hand part of its run to a Workflow. The Agent handles the parts that need judgment, such as deciding what to look at, interpreting results, and writing, and the Workflow handles the parts that must run the same way every time.

### When to combine them

* **A step needs fixed logic.** For example, a calculation, a scoring model, code, or an API call that must run the same way on every run.
* **You already have a Workflow that works.** Reuse it as a step instead of rebuilding that part in the Agent's instructions.
* **You need an integration that Agents can't use as a tool yet.** Build a Workflow with that integration and have the Agent run it.
* **You want an Agent's triggers for an existing Workflow.** Start the Workflow from a monitor, an app event, or a Profound data change by running it from an Agent.

### Add a Workflow to an Agent

1. Make sure the Workflow is published. Only published Workflows can be added to an Agent.
2. Open the Agent and go to **Instructions** in the settings panel.
3. Place the cursor in the step where the Agent should run the Workflow, type **/**, and open **Agents**. Pick the Workflow from the list, or type its name to search.
4. The Workflow appears in the instructions as a pill with its name. Write around it what the Agent should give the Workflow and what to do with its results, for example, "For each topic from the input, run \[Workflow]. Then compare the results and write a summary."

Or ask Aim in the Agent's chat to add it, for example, "Use my Competitor Snapshot Workflow to pull the numbers."

To remove a Workflow from the Agent, hover over its pill in the instructions and select the **×**.

### What happens during a run

* When the Agent reaches the step, it starts a run of the Workflow and fills in the Workflow's inputs from the task.
* The Workflow's run shows in the Agent's chat with its status. Select **Open** to see the run in the Workflow, including what each node returned. The run also appears in the Workflow's own run history.
* If the Workflow includes a human review step, the run pauses until someone reviews the content.
* When the Workflow finishes, the Agent uses its results in the next steps.

### Requirements and limits

* **Access:** anyone who runs the Agent, including the person who set a trigger, needs access to the Workflow. When you share the Agent, share the Workflows it uses too.
* **Credits:** the Workflow's run uses credits the same way it does when you run it on its own.
* **Agents can't run other Agents.** Only Workflows can be added as steps.

## Side-by-side comparison

| | Agent | Workflow |
| - | - | - |
| **How you build it** | Describe the goal and the tools in plain English. Aim drafts the Agent, and you refine it in chat or edit the instructions. | Lay out each step as a node on a canvas, then configure and connect the nodes. |
| **How it runs** | Aim reasons through the steps on each run. | Follows the same fixed path every time. |
| **When the input doesn't match the expectation** | Works around it during the run and tells you what it did. | Stops or returns an error at the node that failed. |
| **Output** | Can differ between runs, within the guardrails you set. Wording and steps can differ between runs. | Identical structure every run. |
| **Seeing the work** | Every run is a chat: you see why each step happened, and the run is saved. | You see what each node returned. |
| **Tools and integrations** | Tools: connections to accounts in services like G2, Notion, Linear, or WordPress. Profound data is built in. | Integrations: nodes for each service, Profound data, code, and API calls. |
| **How it starts** | On demand, or on a Schedule, Monitor, Profound data, or app trigger. | On demand, on a schedule, or as a step in an Agent run. |
| **Using the other** | Can run published Workflows as steps. | Can't run Agents. |
| **API and Model Context Protocol (MCP)** | Not yet supported. Agents run in the Profound platform. | Supported through the Profound API and MCP. |
| **Credits per run** | Higher: reasoning on every step uses credits. | Lower for the same task. |
| **Best for** | Recurring tasks: monitoring, reporting with commentary, research, and drafting. | High-volume tasks that must be identical, code and API steps, and deterministic data operations. |


# Building blocks of an Agent
Source: https://docs.tryprofound.com/agents/building-blocks-of-agent

The triggers, inputs, instructions, outputs, tools, context, and settings that make up an Agent

An Agent is a single document made of a body of instructions and the properties it needs to run: triggers, inputs, outputs, tools, context, and settings. An Agent can also run Workflows you've built as steps.

You find them on the Agent's settings panel, in this order: **Run triggers**, **Inputs**, **Instructions**, **Outputs**, **Tools**, **Context**, and a collapsed **Advanced settings** section. Edit them directly, or describe the change to Aim in the Agent's chat. Changes apply to the Agent right away.

## Inputs

Information a person provides each time they run the Agent, such as a URL, a topic, or a client name. Most Agents take an input, but they don't have to: an Agent that always does the same job, like getting data from a connected tool, needs none.

Each input has a type, a label, and an optional description, and you can mark it as required (inputs are optional by default). When someone runs the Agent, it asks for the inputs before it starts.

Input types include text, numbers, URLs, dates and date ranges, lists, dropdowns, true/false values, Documents, Sheets, file uploads, and Profound data: assets, categories, topics, prompts, regions, and tags.

## Instructions

The body of the Agent: what it should do, in what order, which tools and Workflows it should use, and any other guidance you have for the Agent.

Aim drafts the instructions when you describe the Agent, and updates them when you ask for changes. To ask Aim about part of the instructions, select the text and choose **Ask in chat**.

Switch the **Instructions** card to **Execution steps** to see the flow Aim works out from your instructions as a list of steps. To change the steps, ask Aim in the chat, or select **Regenerate steps** after you edit the instructions.

### Add items with / and @

Type **/** anywhere in the instructions to open the insert menu. Pick a category, or keep typing to search. The item appears in the instructions as a pill, so the Agent knows exactly what to use at that step. The categories are:

* **Inputs**, **Outputs**, and **Tools**: refer to the Agent's inputs, outputs, and tools in the step that uses them, for example, "Save the summary as \[Output]."
* **Agents**: run a published Workflow as a step. Learn more in [Run a Workflow as a step](#run-a-workflow-as-a-step) below.
* **Skills**: point the Agent to a Skill at the step where it should apply it.
* **Documents**, **Sheets**, and **Knowledge Bases**: point the Agent to reference material in Profound.
* **Basic blocks**: formatting for the instructions, such as headings and lists.

Type **@** to mention an item the Agent already has, such as a tool or an input.

### Run a Workflow as a step

Have the Agent run a Workflow you've built in the Workflow Builder as one of its steps. Use it for the parts of a task that must run the same way every time, such as fixed calculations, code, API calls, or integrations that Agents can't use as tools yet.

* **Add a Workflow:** in the instructions, type **/**, open **Agents**, and pick the Workflow, or type its name to search. It appears as a pill with the Workflow's name. Place it in the step where the Agent should run it, for example, "Run \[Workflow] for each topic, then summarize the results."
* **What happens during a run:** the Agent starts a run of the Workflow, fills in its inputs, waits for it to finish, and uses its results in the next steps. The Workflow's run shows in the Agent's chat, and you can open it to see what each node returned.
* **Remove a Workflow:** hover over its pill in the instructions and select the **×**.

Only published Workflows that you have access to are listed. Anyone who runs the Agent needs access to the Workflows it uses, so share them along with the Agent. An Agent can run Workflows, but not other Agents.

## Outputs

The artifacts the Agent produces: a Document, a Sheet, an HTML file, a CSV, or a PDF. Declare an output when you want a file at the end of every run. Each output can be visible to **Everyone** or **Private**. Outputs are saved under **Artifacts** at the top of the Agent's page. Some Agents don't produce any artifacts (for example, an Agent that creates a Linear issue whenever your Visibility Score drops).

## Tools

The apps the Agent can use while it works, such as G2, Notion, or Linear. Each tool is a connection to an account in that service, so the Agent acts with that account's permissions and asks before it changes anything in the connected service. Your Profound data is available to every Agent without setup. Learn more in [Tools in Agents](/agents/tools-in-agents).

## Triggers

Conditions that start an Agent run on their own:

* A **Schedule** trigger runs an Agent on a recurring schedule.
* An app trigger, listed under the app's name (for example, **Slack**), runs an Agent when an event happens in that app, such as a new channel message.
* A **Monitor** trigger watches the web for a change or an event (for example, a competitor publishes a new blog post) and runs an Agent when it happens. Monitors check the web as often as every hour.
* A **Profound data** trigger watches your Profound data and runs an Agent when a metric crosses a threshold, including FactCheck metrics such as the inaccurate claim count.

The switch next to **Run triggers** turns all of the Agent's triggers on or off at once. Without a trigger, the Agent runs only when someone selects **New run**. Learn more in [Triggers in Agents](/agents/triggers-in-agents).

## Context

Material the Agent draws on during every run: Knowledge Bases, Skills, Profound Documents, and Sheets. Select **Add** next to **Context** to attach them, then mention them in the instructions so the Agent knows when to use them.

## Advanced settings

Expand **Advanced settings** to set:

* The **Default model** and **Default effort**.
* **Capabilities**: turn **Browser**, **Web search**, and **Image generation** on or off. Turn off what the Agent doesn't need.
* **Behavior**: **Allow self-updates** lets the Agent modify its own instructions; **Allow questions** lets it pause to ask the person running it for input; **When a tool needs approval** sets whether it asks (**Ask**), approves automatically (**Approve**), or refuses (**Deny**).

## Next steps

* Learn how to create an Agent in [Get started with Agents](/agents/getting-started).
* Learn when to use a Workflow inside an Agent in [Agents and Workflows: when to use which](/agents/agents-vs-workflows).
* Check out the [Agents FAQ](/agents/faq).


# Agents FAQ
Source: https://docs.tryprofound.com/agents/faq

Frequently asked questions about Profound Agents and Workflows

## General

> **Where are the Agents I built with the Workflow Builder?**
>
> Everything you built in the Workflow Builder is now called a Workflow. Your existing Workflows are on the Agents page, in the same list as all other Agents. To see only Workflows, filter the list by **Type** > **Workflow**. Nothing changes in the way they run unless you [migrate them to Agents](/agents/how-tos/migrate-workflow-to-agent).

> **Do Agents use credits?**
>
> While Agents are in public beta, running them doesn't use credits. Credit billing for Agents starts after the beta.
>
> Workflows still use credits as before, including Workflows that an Agent runs as a step.
>
> Once billing starts, an Agent run uses more credits than the same task run as a Workflow, because the Agent reasons through each step and that reasoning uses credits. To keep your credit use down:
>
> * **Keep high-volume, fixed tasks as Workflows.** Run them on their own, or as a step inside an Agent so the Agent only reasons about the parts that need judgment.
> * **Check how often triggers fire.** Every triggered run is a full run. An hourly monitor, for example, can start up to 24 runs a day.
> * **Match the effort to the task.** For simple, repetitive tasks, set a lower **Default effort** under **Advanced settings**.
>
> Learn more in [Profound usage pricing: Credits](/getting-started/credits).

> **Can I run an Agent across many inputs at once instead of one at a time?**
>
> No. Bulk runs with Sheets are currently supported only for Workflows. The Workflows you built in the Workflow Builder can still run in Sheets as normal.

> **Is my Agent data used to train models?**
>
> No. All data in Agents is contractually prohibited from being used by third-party AI providers (such as OpenAI, Anthropic, or Perplexity) to train or improve their underlying models.

> **Will I get the same output every run?**
>
> No. An Agent reasons through the task each time, so wording and steps can vary between runs. If you need certain things to stay the same (format, data sources, tone of voice), tell Aim in the Agent's chat. If you need identical output and the same sequence of steps on every run, use a Workflow, either on its own or as a step inside an Agent.

> **Do I need to publish or test an Agent before it runs?**
>
> No. Changes you make to an Agent apply right away, and every run uses the current version. There are no separate test runs: run the Agent with **New run**, check the result in its chat, and ask Aim for changes if you need them.

> **Why isn't my trigger firing?**
>
> Check the following in the Agent's settings panel:
>
> * **The switch next to Run triggers is on.** When it's off, all of the Agent's triggers are paused. Only the person who created a trigger can switch it.
> * **The trigger is listed under Run triggers.** A trigger described only in the instructions, such as "every Monday", doesn't fire.
> * **For Slack triggers, the Profound Slack app is in the channel.** The channel list only shows channels the app has joined.
> * **The Agent has been created.** An Agent that's missing a name or instructions shows as a draft, and its triggers stay paused until it's complete. Hover over a paused trigger to see why.

> **Why does the Agent ask permission for a tool again on a triggered run?**
>
> Each run is its own chat, and you grant tool permissions per chat, so **Allow once** or **Allow for this task** doesn't carry over. To stop the prompts for a tool action, select **Always allow** on the approval card, or set the action to **Allow** in the tool's **Tools & permissions** settings. To change this for the whole Agent, set **When a tool needs approval** under **Advanced settings** to **Approve**.

> **Can I create or run Agents through the API or MCP?**
>
> Not yet. Workflows work with both the Profound API and MCP. Agents currently work only in the Profound platform.

> **Who can see my Agent?**
>
> Only you, by default. Select **Share** to invite teammates with **Can view** or **Can edit** access, or set **General access** to everyone in your organization. Shared Agents appear under **Shared with me** for the people you share them with. Runs started by a trigger are visible to the person who set the trigger.

> **A teammate changed my Agent. Can I undo it?**
>
> Yes. Open **Version history** from the **...** menu on the Agent's page, review earlier versions, and select **Restore** to bring one back.

> **How do I stop an Agent from running on its own?**
>
> Turn off the switch next to **Run triggers** in the Agent's settings panel to pause all of its triggers, and turn it back on to resume. To stop a single trigger, remove it from **Run triggers**.

> **Can I share an Agent as a template?**
>
> Yes. In the **...** menu, select **Submit as template**, pick a category, and choose who can use it: your organization or all Profound customers. Profound reviews every submission before it appears. Track your submissions under **My template submissions**.

## Agents vs Workflows

> **Which should I use, an Agent or a Workflow?**
>
> Agents built with Aim work for most tasks, so we recommend using them. When you need the same steps on every run, or full control over the steps and their order, use a Workflow. You can also combine them: an Agent can run a Workflow as one of its steps. Learn more in [Agents and Workflows: when to use which](/agents/agents-vs-workflows).

> **Can an Agent run a Workflow?**
>
> Yes. In the Agent's instructions, type **/**, open **Agents**, and pick a published Workflow. During a run, the Agent starts the Workflow, fills in its inputs, and uses its results in the next steps. You can follow the Workflow's run in the Agent's chat and open it to see what each node returned. Learn more in [Agents and Workflows: when to use which](/agents/agents-vs-workflows).

> **Why doesn't my Workflow appear in the list?**
>
> Only published Workflows that you have access to are listed. Publish the Workflow in the Workflow Builder, or ask its owner to share it with you.

> **My teammate can't run an Agent that uses a Workflow. Why?**
>
> Anyone who runs the Agent needs access to every Workflow it uses. Share the Workflow with them, too.

> **Can an Agent run another Agent?**
>
> No. An Agent can run Workflows as steps, but not other Agents.

> **Do I have to migrate my existing Workflows to Agents?**
>
> No. Your Workflows keep running as they are, so you don't have to migrate any of them. To use a Workflow with an Agent's triggers or reasoning, add it to an Agent's instructions and the Agent runs it as a step.

> **My migrated Agent behaves differently from the Workflow. Is that expected?**
>
> Yes. When you migrate a Workflow, Aim writes instructions based on it and then reasons through them instead of following the nodes, so the results are close but not identical. Run the migrated Agent and adjust the instructions before you rely on it. Your original Workflow stays unchanged. See [Migrate a Workflow to an Agent](/agents/how-tos/migrate-workflow-to-agent).

> **Why would I migrate a Workflow to an Agent?**
>
> Migrate a Workflow if you need:
>
> * Easier updates: describe the change in the Agent's chat, and Aim makes it.
> * Flexibility with unexpected inputs, or a fix for a Workflow that keeps breaking: the Agent reasons through the run and works around common issues.
> * Triggers beyond a schedule. Learn more in [Triggers in Agents](/agents/triggers-in-agents).

> **Will the Workflow templates be available as Agent templates?**
>
> Profound is rebuilding the Workflow template library as Agent templates. To browse templates that aren't rebuilt yet, open the Templates page and switch the type filter from **Agents** to **Workflows**.

## Content creation

> **Do Agents implement content updates automatically, or do they only provide recommendations?**
>
> Both are possible, and you control which approach your Agents take. Agents can generate recommendations and surface them for human review before any content is published, or publish directly to your content management system (CMS) without a manual approval step. Most teams start with a review step and automate publishing once they're confident in the output quality.

> **Can Agents generate content briefs and articles based on identified gaps?**
>
> Yes. A common use case for Agents is to pull Profound visibility data or run a Google Search query, identify gaps or underperforming topics, and then generate content briefs or full articles. The Agent can then pause for a human review or publish the content directly to your CMS.

> **Do the content generation tools use public chat apps or connections?**
>
> No. All content generation in Agents goes through secure, enterprise-tier API connections rather than public-facing consumer apps, such as the public ChatGPT or Gemini apps.

## Tools and integrations

> **Does Profound integrate with my CMS so I can publish content automatically?**
>
> Yes. Agents can publish to WordPress, Webflow, Contentful, and Sanity through tools, and Workflows support more CMS platforms through integrations. To publish to a CMS that's only available to Workflows, have your Agent run a Workflow that publishes there. Learn more in [Integrations vs connectors: what's the difference?](/integrations/integrations-vs-connectors).

> **Does Profound integrate with Shopify?**
>
> Agents don't support Shopify as a tool yet. In Workflows, you can push data to and pull data from Shopify with an API call, which brings product content, catalog data, or e-commerce signals into your answer engine optimization (AEO) work. To use that data in an Agent, add the Workflow to the Agent's instructions. Native Shopify support is planned for a future release.

> **Does the platform integrate with Google Analytics or Looker Studio?**
>
> Agents support Google Analytics as a tool: add it to an Agent to pull Google Analytics 4 (GA4) data into a run. Looker Studio isn't available as a tool. As a workaround, connect your Agent's outputs to your Looker Studio reporting pipelines.

> **Does the platform integrate with Reddit?**
>
> There's no direct Reddit integration, but several Workflow templates cover Reddit analysis and engagement, such as the Reddit Insights Generator and Reddit Comment Generator. They identify top-cited Reddit threads, surface key discussion themes, and generate on-brand comments.

## Workflows

> **What models power the Prompt LLM node? Can I bring my own API keys?**
>
> The Prompt LLM node supports a variety of models, including the latest from Anthropic, OpenAI, and AWS Bedrock. Select any available model. Bringing your own API keys isn't supported.


# Get started with Agents
Source: https://docs.tryprofound.com/agents/getting-started

Build your first Agent with AI Marketer (Aim), run it, and refine it in chat

This guide takes you from an empty Agents page to an Agent that runs on its own. You describe the task, let AI Marketer (Aim) build the Agent, run it, and refine it in the same chat.

## Before you start

* **Pick a task with a clear outcome.** A weekly summary of a Visibility Score change, a content brief from a keyword, and a digest of new brand mentions are all good use cases for Agents.
* **Decide how the Agent should run:** on demand, on a schedule, or when something happens on the web or with your Profound data. Change this at any time.
* **Check for Workflows you can reuse.** If a Workflow already handles part of the task, the Agent can run it as a step instead of repeating that work.

## Build an Agent

**1. Start a new Agent**

In the Profound platform, go to **Workspace > Agents** in the left sidebar and select **Create**. Or select **+** next to **Agents** in the sidebar.

The **Create an agent** page opens with an empty Aim chat. Templates are listed below it under **Use a template**.

To build a Workflow instead, select the arrow next to **Create**, then **Workflows** > **New Workflow**.

**2. Describe the Agent**

Write what the Agent should do, when, and where the result goes. One or two sentences are enough. For example:

```txt wrap theme={null}
Every Monday, check whether our Visibility Score dropped week over week for the AI Visibility Products category and save a short summary as a Document.
```

If a Workflow should handle part of the job, name it in your description, for example, "Use the Competitor Snapshot Workflow to pull the numbers, then write the summary."

Send the message, and Aim starts working.

**3. Answer the questions**

Aim asks for anything it can't infer from your description. In the example above, it may ask what time on Monday to run and how detailed the summary should be.

The questions appear in the message box, one at a time. Select the option Aim suggests, select **Skip** to let Aim choose for you, or type your own answer in the **Other...** field. Select **Next** to move on and **Submit** after the last question.

When Aim has a name and instructions for the Agent, it creates the Agent and shows an **Agent created** message. The Agent's page opens with the chat on the left and the Agent's settings on the right.

**4. Review the Agent**

The settings panel lists the Agent's building blocks: **Run triggers**, **Inputs**, **Instructions**, **Outputs**, **Tools**, **Context**, and **Advanced settings**. Learn more about them in [Building blocks of an Agent](/agents/building-blocks-of-agent).

Review each one. To change something, edit it directly or describe the change in the chat. Changes apply to the Agent right away: there's no separate publish step.

To have the Agent run one of your Workflows as a step, type **/** in the instructions, open **Agents**, and pick the Workflow. Learn more in [Agents and Workflows: when to use which](/agents/agents-vs-workflows).

**5. Run the Agent**

Select **New run** at the top of the page, or above the message box. If the Agent has inputs, a short form opens in the message box: fill in each input, then select **Start run**.

The run happens in the Agent's chat. You see each step, each tool call, and the result as they happen. When the run finishes, ask follow-up questions about the results, or describe what to change in the Agent and Aim updates it. Run the Agent again after each change to check the output.

If the run button is unavailable because some of the tools the Agent needs aren't connected, connect the missing tools in the **Tools** section first.

**6. Review runs and artifacts**

The Agent's page keeps a record of its work. Select either option at the top of the page:

* **Run history**: every run, manual or triggered. Select a run to reopen it in the chat.
* **Artifacts**: the Documents, Sheets, and files the Agent produced.

**7. Refine the Agent**

Use what you learned from the first runs to improve the Agent. There are two ways to do it:

* **Edit the configuration yourself.** Change the instructions, inputs, outputs, tools, or context directly in the settings panel.
* **Give feedback in the chat.** Describe what you'd like to be different, and Aim updates the Agent. For example, "Pull in my brand kit to better align with my brand's tone of voice."

Changes apply right away. Select **New run** to check the result, and repeat until the output is what you need.

**8. Add triggers and share**

To have the Agent run on its own, add a trigger under **Run triggers** in the settings panel and make sure the switch next to it is on. Learn more in [Triggers in Agents](/agents/triggers-in-agents).

To give teammates access, select **Share** at the top of the page.

## Starting from a template

Templates are a quick way to get started with Agents. Select **Templates** at the top of the Agents page to open the **Templates** page. Filter templates by category or search for one. Templates also appear under **Use a template** on the **Create an agent** page.

Select a template to see its description, **Steps**, and **Instructions**. Select **Use template**, and Aim builds your own private copy of the template in a new chat and helps you set it up. Connect any tools it needs, and adjust it however works best for you.

## Finding your Agents

The Agents page lists your Agents. Use the **Recents**, **My Agents**, and **Shared with me** tabs to switch between Agents you used recently, Agents you created, and Agents shared with you. Filter by **Created by**, **Status** (Published or Draft), or **Type** to show only Agents or only Workflows, change the **Sort** order, or select **Search** to find one by name.

Each row shows the Agent's name, owner, and an icon that shows whether access is private or shared. Incomplete Agents have a **Draft** label. Select the star on a row to add an Agent to your favorites; favorites appear at the top of the list.


# Tools in Agents
Source: https://docs.tryprofound.com/agents/tools-in-agents

The apps an Agent can use, how to add them, and how tools differ from integrations and connectors

Tools are the apps an Agent can use while it works: read a Notion page, create a Linear issue, or publish a post to WordPress. Each tool is a connection to an account in that service, so the Agent does only what that account can do. Your Profound data is available to every Agent without any setup.

## Available tools

The external tools currently available are:

| Category | Tools |
| - | - |
| Project management and issue tracking | Asana, Atlassian, Linear |
| Content management (CMS) | Contentful, Sanity, Webflow, WordPress |
| Marketing automation and customer messaging | Customer.io, Customer.io (EU), HubSpot, Klaviyo, Knock |
| Web search, crawling, and SEO data | DataForSEO, Exa, Firecrawl, Google Search Console, Parallel |
| Customer support | Pylon |
| Reviews and market intelligence | G2 |
| Analytics | Google Analytics, Hex |
| Developer tools and source control | GitHub |
| Email and collaboration | Gmail, Google Drive, Microsoft SharePoint, Notion, Slack |
| Meetings and workplace knowledge | Gong, Granola, Microsoft Work IQ |
| Social media | Ordinal |
| Stock media | Getty Images |

Profound data (Visibility, Citations, Sentiment, FactCheck, prompts, and topics) is built in and isn't listed among an Agent's tools. It appears in Agent instructions as a Profound pill.

This list grows over time, and the tools available to your organization may differ. Open the **Add tool** dialog on an Agent to see what's available to you. To use a service that isn't listed, add it as a custom connector.

## Add or remove a tool

To add a tool to your Agent:

1. In the Agent's settings panel, select **Add** next to **Tools**.
2. Pick a tool from the list, or search for it. Tools the Agent already uses are listed under **Connected**, and common ones under **Most used**.
3. Read what the tool lets the Agent do, select **Connect account** and sign in to the service, and then select **Add tool**. If more than one account is connected, choose one under **Which account should the agent use?** A lock icon marks tools that only an admin can connect.

Or ask Aim in the Agent's chat to add a tool. For example, "Use Notion to read the brief" adds the Notion tool if it isn't there yet and prompts you to connect it.

To remove a tool, ask Aim to remove it, select the trash icon next to it in **Tools**, or select **Remove tool** in the **Add tool** dialog. Then update the instructions so they no longer refer to the tool. The change applies to the Agent right away.

## Tools, integrations, and connectors

Profound connects to outside services through tools, integrations, and connectors.

**Tools** (also called **connectors**) are what Agents and Aim use. A tool is a connection to a service through its own API or Model Context Protocol (MCP) server, made under your personal account or an organization account. You add tools from an Agent or from Integrations in the platform settings.

**Integrations** are what Workflows use. An integration adds Profound-built nodes to the Workflow Builder, one node per action. You set up integrations in Integrations in the platform settings.

Some services are available both ways, for example Slack, Contentful, G2, WordPress, or Sanity. Others exist only as Workflow integrations today, such as Gamma. Learn more in [Integrations vs connectors: what's the difference?](/integrations/integrations-vs-connectors).

To use a service that's only available as a Workflow integration, build a Workflow that uses it, then add that Workflow to your Agent's instructions. The Agent runs the Workflow as a step. Learn more in [Agents and Workflows: when to use which](/agents/agents-vs-workflows).

### Add a custom connector

If a service has a public MCP server but isn't in the list of tools, add it as a custom connector. Your Agents can then use it like any other tool.

1. Go to **Settings** in the organization name menu at the top left of the screen > **Integrations**.
2. Scroll to the bottom of the list and select **Add custom connector**.
3. Enter a **Name** for the connector, for example "Exa Search".
4. In **URL**, paste the public HTTPS URL of the MCP server, for example `https://mcp.example.com/mcp`.
5. Choose the **Connection type** the server uses: **HTTP** or **SSE**.
6. Choose how the connector signs in to the server under **Authentication** (see the options below).
7. Under **Access**, choose who can use the connector: **Only me**, or your organization's name to share it with everyone in your organization.
8. Optionally, open **Appearance settings** to add a **Logo URL** and a **Description**.
9. Select **Add connector**. Profound connects to the server and discovers its tools.

Authentication options:

* **None**: the server doesn't need credentials.
* **Secret**: Profound sends a header, such as an API key, with each request to the server. Enter the header name (for example, `x-api-key`) and the **Secret value**. Select **Add secret** to add more headers.
* **OAuth**: you sign in to the service. Enter a **Client ID** and **Client secret** only if the service requires them.

Then add the connector to an Agent from the **Add tool** dialog, the same way as other tools.

## FAQ

> **Does connecting a tool give Profound access to everything in my account?**
>
> The Agent can do only what the connected service allows for the connected account, and only what the instructions ask it to do. Actions that change something in the service need your permission during the run, unless you've set them to **Allow**.
>
> To see your permissions for each tool:
>
> 1. Go to **Settings** in the organization name menu at the top left of the screen > **Integrations**.
> 2. Select a service (for example, G2) to open its page.
> 3. Switch to the **Connector** tab. View and edit permissions in the **Tools & permissions** section. Tools are grouped into reading, editing, and destructive tools. Set each one to **Allow**, **Ask**, or **Block**.
>
> To open these settings from the **Add tool** dialog, select **Manage** next to **Tools & permissions**. Permissions apply to all your Agents.

> **Can my teammates use a tool I connected?**
>
> It depends on the account you used to connect the tool. If it's your personal account, the tool is yours: when you share an Agent, each person who runs it connects the tools they need under their own account. Anyone in your organization can use a tool connected to an organization account. When a tool isn't connected, the Agent can't run until someone connects it in the **Tools** section.


# Triggers in Agents
Source: https://docs.tryprofound.com/agents/triggers-in-agents

Start Agent runs automatically on a schedule, an app event, a change on the web, or a change in your Profound data

A run trigger starts an Agent run automatically when its condition is met. The condition can be a schedule, an event in a connected app or on the web, or a change in your Profound data.

Add triggers in the Agent's settings panel, or ask Aim in the Agent's chat to add them.

## Trigger types

| Trigger | Runs when | Example |
| - | - | - |
| **Schedule** | A scheduled time arrives | Every Monday at 9:00 AM |
| **App** (listed by app name, for example **Slack**) | An event happens in a connected app | A new message is posted in a Slack channel |
| **Monitor** | Something changes on the web | A competitor publishes a new blog post |
| **Profound data** | A Profound metric meets a condition | Visibility Score drops 10% compared with the last 7 days |

Add more than one trigger to an Agent, and it runs when any trigger's condition is met. By default, an Agent has no triggers and runs only when you select **New run**.

## How triggers behave

* **One switch turns triggers on and off.** The switch next to **Run triggers** in the settings panel pauses or resumes all of the Agent's triggers at once. It appears once the Agent has at least one trigger. Only the person who created a trigger can switch it.
* **Triggered runs use the current version of the Agent.** Changes you make apply to the next triggered run.
* **Triggers belong to the person who set them.** Only that person sees the triggered run and its outputs. People the Agent is shared with can run it themselves and see their own runs. Triggers aren't saved in version history.
* **Every triggered run is a chat.** Find it under **Run history** at the top of the Agent's page, with any outputs under **Artifacts**.
* **Triggered runs can run Workflows, too.** If the Agent's instructions include a Workflow, a triggered run runs it and uses its results, the same as a manual run. The person who set the trigger needs access to the Workflow.

## Add a schedule

In the settings panel, select **Add** next to **Run triggers** and point to **Schedule**. Select **Hourly**, **Daily**, or **Weekly** to add a schedule that starts now, or **Custom** to set it up yourself:

1. Set **Starts on** to the first date and time the Agent should run.
2. Choose a **Frequency**: **Doesn't repeat**, **Daily**, **Weekdays**, **Weekend**, **Weekly**, **Monthly**, **Annually**, or **Custom**. For a custom cadence, select **Custom**, set **Repeat every** (for example, 1 week or 4 hours), and select the days under **Repeat on**.
3. Select **Add trigger** to save.

The trigger appears under **Run triggers**, named after how often it runs, for example, "Every Monday at 9:00 AM". Select it to edit it, or select **Remove** in the dialog or the trash icon next to it to delete it.

To add a schedule in the chat, ask Aim, for example, "Run this Agent every Monday at 9 AM."

## Add an app trigger

1. Select **Add** next to **Run triggers** and point to the app, for example **Slack**. Only connected apps that support triggers appear in the menu.
2. Choose the event, for example **New channel message** or **Reaction added to a message**.
3. Complete the trigger sentence: for Slack, choose which messages or reactions count and select the channel. The trigger saves as soon as the sentence is complete.

For the Slack app trigger, add the Profound Slack app to the channel you want to use with an Agent. The channel list only shows channels the app has already joined. Learn more in [Connect Slack to Profound](/integrations/slack/connect-slack-to-profound).

## Add a monitor

A monitor watches the web for updates that match a query you write. Profound checks for new matches on the schedule you choose, ignores anything it has already reported, and starts a run only when something new appears.

1. Select **Add** next to **Run triggers** and choose **Monitor**.
2. Optionally, enter a **Name**, for example "Competitor blog monitor". If you leave it empty, the monitor is named "Monitor 1", "Monitor 2", and so on.
3. In the **Query** field, describe what to watch (up to 2,000 characters), for example, "a new post on competitor.com/blog" or "changes to the pricing page at competitor.com/pricing".
4. Choose a **Frequency**: **Every hour** (the default), **Every 2 hours**, **Every 4 hours**, **Every 6 hours**, **Every 12 hours**, or **Every day**.
5. Select **Add trigger** to save.

After you save a monitor, you can rename it, but you can't change the query or frequency. To change them, remove the monitor and add a new one.

### Writing a good monitor query

* Describe what you're looking for in plain language, the way you'd brief a colleague.
* Keep your query clear and specific: include what to watch for and where to watch for it.
* Leave out date ranges. The monitor already surfaces only what's new.

Prefer the queries in the first column and avoid the ones in the second.

| Prefer ✅ | Avoid ❌ |
| - | - |
| AI startup funding announcements | Find all AI funding news from the last 2 years |
| Tesla news and announcements | Tesla news after December 12, 2025 |
| New blog posts published on competitor.com | competitor.com blog |

## Add a Profound data trigger

1. Select **Add** next to **Run triggers** and choose **Profound data**.
2. Choose the **Category** and the **Asset** to watch.
3. Build the **Trigger** condition: set **When** a metric **drops by**, **rises by**, **drops below**, or **rises above** a value (in % or a number), **Against** a comparison window (**Last day**, **Last 7 days**, **Last 14 days**, **Last month**, or **Last 3 months**).
4. Narrow the scope with **Filters**: Topics, Platforms, Regions, Tags, or Personas.
5. Set the **Check frequency** (**Daily** by default, or **Weekly**, **Bi-weekly**, or **Monthly**) and select **Add trigger**.

Available metrics:

* **Visibility**: Visibility Score, Share of Voice, Average Position, and their ranks.
* **Sentiment**: positive percentage, negative percentage, and assessment count.
* **Citations**: Citation Share, citation count, Citation Rank, and Citation Coverage.
* **FactCheck**: Accuracy score, and inaccurate, accurate, and verified claim counts. These appear only when FactCheck is set up for the category.

## Edit, pause, or remove a trigger

In the Agent's settings panel, select the trigger under **Run triggers**. Change the settings and select **Save**, or select **Remove** to delete the trigger.

To pause all of an Agent's triggers without deleting them, turn off the switch next to **Run triggers**, and turn it back on to resume. To stop an Agent from running on its own permanently, remove its triggers.

Trigger changes apply right away.


# About Workflows
Source: https://docs.tryprofound.com/agents/workflows/about-workflows

What Profound Workflows are and how the node-based workflow builder automates the AEO cycle

![Frame 2147259079 (4).png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/about-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=678a0812b405ad356725299ca1a190a4)

## Overview

Profound Workflows is a modular, node-based Workflow Builder that lets your team automate the full AEO cycle, from pulling live visibility data all the way to publishing optimized content into your CMS. It brings together Profound's AEO data, leading AI models, and the tools your team already uses into a single, connected pipeline.

Workflows use cases include:

* Running Answer Engine Insights reports and automatically sending them to your team via Slack.
* Identifying content gaps and automatically creating optimized content to fill those gaps.
* Detecting underperforming content and automatically updating it to perform better in AI search.

## How Workflows work

A Workflow is built on a visual canvas made up of nodes. Each node performs a specific task: querying Profound data, prompting an AI model, fetching content from the web, or publishing to a connected platform. Nodes are connected in sequence, and data passes between them as variables.

This structure means you can build workflows of any complexity: a simple two-step Workflow that drafts a content brief and sends it to Slack, or a multi-step pipeline that pulls visibility data, identifies content gaps, generates optimized articles, and publishes them directly to your CMS.

## Workflow components

Every Workflow is made up of the same building blocks:

* **Inputs** — Information you provide each time you start a Workflow. Inputs can include text, Profound data, files, or dropdown selections.
* **Nodes** — The steps of your Workflow. Each node performs a specific task, like querying data, generating content, sending a message, or publishing to a CMS.
* **Variables** — Every piece of data that flows in and out of a node. Variables can be passed between steps and reused at any point in the workflow.
* **Workflow output** — The final result produced after all steps run. This can be generated content, a published article, a Slack notification, or a confirmation that the Workflow ran successfully.

### Conditionals and iterations

Workflows support branching logic and loops, so your workflows can handle real-world complexity.

**[Conditionals](/agents/how-tos/add-conditionals-to-workflows)** let your Workflow take different paths based on the value of a variable. For example, routing high-visibility content to one workflow and low-visibility content to another, or only publishing content if a quality check passes.

**[Iterations](/agents/how-tos/add-iterations-to-workflows)** let your Workflow apply the same sequence of steps to every item in a list. For example, iterate over a list of competitor websites to extract their pricing, then pass that combined output to an LLM node to generate a new pricing page for your site.

### Sheets

[Sheets](/sheets/getting-started) lets you run Workflows at scale by supplying a spreadsheet of inputs instead of running a Workflow one item at a time. Upload a CSV or connect a Google Sheet, and Profound processes each row as a separate Workflow run in parallel.

This is particularly useful for content operations teams managing large keyword lists, site audits, or bulk content generation projects.


# Create a Workflow
Source: https://docs.tryprofound.com/agents/workflows/create-workflow

Build and run your first Workflow

## Step 1: Open the Workflow Builder

1. Go to **Agents** in the left navigation sidebar.
2. Select **+ Create** on the top right of the screen.
3. In the dropdown menu, select **Workflows > + New Workflow**.

The Workflow Builder opens to a blank canvas.

## Step 2: Configure your inputs

Every Workflow begins with a **Start** block where you configure Workflow inputs. Inputs are the values you supply each time the Workflow runs: a keyword, a topic, a URL, and so on.

Select the **Start** block to open its settings. Add each input you need, give it a clear label (for example, `target_keyword` or `article_url`), choose a type (text, number, list, date range, and so on), and indicate whether it's required. These labels become the variable names you'll reference in later steps.

## Step 3: Build your pipeline

Add nodes by clicking **+** on the canvas or dragging nodes from the left sidebar. Each node has a settings panel that opens on the right: configure inputs, reference variables from earlier steps, and label the output so it's available to subsequent nodes.

Common starting patterns:

* [Visibility Score](/agents/workflows/nodes/data/visibility-score) node → [Prompt LLM](/agents/workflows/nodes/prompt-llm) (generate brief) → [Slack](/integrations/slack/connect-slack-to-profound) message
* Google Search node → [Prompt LLM](/agents/workflows/nodes/prompt-llm) (analyze results) → [Generate Article](/agents/workflows/nodes/content-creation/generate-article) → [WordPress](/integrations/wordpress/overview) (publish to website)

Add [Conditionals](/agents/how-tos/add-conditionals-to-workflows) to branch your workflow based on variable values. Add [Iterations](/agents/how-tos/add-iterations-to-workflows) to loop a sequence of steps over every item in a list.

## Step 4: Test, publish, and run

New Workflows start as drafts. Before publishing, run a test by supplying sample inputs. Each step's output will be visible so you can verify the pipeline behaves as expected.

When the Workflow is ready, click **Publish** to make it live. You can then run it manually, trigger it on a schedule, or supply a spreadsheet of inputs via [Sheets](/sheets/getting-started) to process multiple items at once.


# Integrations and tools in Workflows
Source: https://docs.tryprofound.com/agents/workflows/integrations-and-tools-in-workflows

The tools and integrations Workflows can connect to, configured once per organization

Workflows connect to the tools your team already uses, so workflow output lands exactly where it needs to go. Integrations are configured once per organization and are then available to any Workflow.

> [!TIP]
>
> Learn more in [Get started with integrations](/integrations/overview).

## CMS integrations

A content management system (CMS) is the platform your team uses to create, store, and publish web content: articles, landing pages, product descriptions, and more. Connecting your CMS to Profound Workflows means content generated inside a workflow can be pushed directly into your CMS without any manual copy-and-paste step.

The CMS platforms supported by Profound Workflows include:

* Basehub
* Contentful
* Framer
* Notion
* Payload
* Sanity
* Webflow
* WordPress

All CMS integrations follow a common flow:

* **Find the integration**. It's usually named after the service you want to integrate: WordPress, Framer, and so on.
* **Choose the action** you want to take: list articles, get content, publish content, and so on.
* **Add a Workflow node** with the action and fill in the required fields: your site or workspace identifier and any necessary credentials. Follow the instructions on the node setup screen.

CMS integrations close the loop between AEO insight and action: use Workflows to identify content gaps, generate optimized articles, and publish them in a single automated pipeline.

> [!TIP]
>
> Multiple CMS accounts can be connected to the same Profound organization. If your team manages more than one site or content environment, configure each separately and select it per Workflow.

## SEO tools

Profound Workflows can pull data from traditional SEO platforms, letting you combine search engine performance signals with AEO metrics in a single workflow. This is especially useful for teams that want to prioritize content efforts using both classic keyword data and AI search visibility.

### DataforSEO

DataforSEO nodes give you access to a broad set of SEO data signals, including SERP results, keyword data, backlink metrics, and on-page data. Use the DataforSEO integration to:

* Retrieve live SERP data for specific queries as part of a content research workflow.
* Pull keyword metrics to supplement Profound's AEO data when building content strategies.
* Automate competitive research by fetching ranking data and passing it into an AI analysis step.

### Google Search

The Google Search node retrieves live search results for a given query and makes that data available to other nodes in your Workflow. It's a lightweight way to bring real-world search context into your workflows without leaving Profound.

Use the Google Search node to:

* Research what currently ranks for a target keyword before generating a content brief.
* Identify the top-ranking pages on a topic so an LLM node can analyze their content and suggest differentiated angles.
* Cross-reference Google search results with Profound's AEO data to spot gaps between what ranks in traditional search and what appears in AI responses.

To use the node, add it to your Workflow, enter a static query or pass a variable from a previous step, and configure the number of results to return. The output is a structured list of URLs and page titles that can be passed to any subsequent node.

## Communications and productivity

Workflows can deliver outputs directly to the tools your team communicates and works in, so results don't sit in a dashboard waiting to be found.

Supported integrations:

* **Slack** — Notify your team when a Workflow run completes, share AEO performance snapshots, flag content that needs attention, or deliver automated reports on a schedule.
* **Gamma** — Generate slide decks, visibility reports, and client-ready presentations in your workspace based on the output of previous Workflow steps.
* **Google Drive** — Save content generated by Workflow steps as files in Google Drive, making them accessible to your team.

## Website builders

Connecting a website builder to Profound Workflows lets you take AEO insights all the way to publishing new pages, without any manual development step. Use these integrations when your workflow needs to go beyond updating existing content and create or deploy new pages.

Supported integrations:

* **Vercel** — Generate and deploy new web pages directly from within a workflow, using Profound data and AI-generated content as inputs.

## Custom API and more

The **Call API** node lets you connect Workflows to any service that exposes a REST or GraphQL API, including internal tools, data platforms, analytics systems, and third-party services not covered by native integrations. This makes Workflows highly extensible: if a platform has an API, Profound can integrate with it.

Additional integrations available through the Call API node include Shopify for e-commerce workflows and Looker Studio for reporting pipelines.


# Article Title Suggestion node
Source: https://docs.tryprofound.com/agents/workflows/nodes/content-creation/article-title-suggestion

Generate an AEO-optimized article title from your topic, audience, and content goals

The **Article Title Suggestion** node generates an AEO-optimized, high-quality article title based on your topic, audience, and content goals. Unlike a simple LLM prompt, this node applies a structured, multi-step workflow to evaluate your inputs, apply editorial and SEO best practices, and output a single, strong title crafted for visibility, trustworthiness, and user intent match.

Add this node to your Workflows when you need reliable, polished titles that set up your content for success—whether for editorial planning, SEO, or Answer Engine Optimization.

## When to use this node

Use Article Title Suggestion for tasks such as:

* Quickly generating compelling, search-aligned titles before content production
* Developing titles that are optimized for AI answer engines and user prompts
* Testing variations of messaging for different audiences or company contexts
* Feeding optimized titles into downstream steps (e.g., content briefs or full articles)

This node is designed to be used early in a content strategy or creation workflow.

## Node configuration

Selecting the Article Title Suggestion node opens its configuration panel on the right side of the Workflow Builder.

### Topic (required)

Provide the subject or theme the article should cover, such as:

* “Emerging trends in healthcare technology”
* “How to optimize your Shopify store for conversion”
* “AI visibility strategy for enterprise brands”

This is the primary driver of the generated title.

### Content Type

Select the type of content the title should be optimized for, such as **General** (default), **Listcle, Blog Post,** and more.

The node adjusts tone, structure, and style based on this setting.

### Company Name

If provided, the title will account for brand context when relevant. This is useful when creating articles for a specific company blog or brand voice.

### Audience Segment

Specify the target audience for the content—e.g., “B2B marketing leaders,” “healthcare professionals,” or “startup founders.”

The node uses this to improve alignment, terminology, and specificity.

### Target Prompt

Enter the user prompt (as used in AI search or on-platform interaction) that your content should optimize for. This helps ensure that the title is directly tied to the way users pose questions or seek information, for example:

* “What is AI visibility and why does it matter for enterprises?”

### Output Label (required)

Assign a label for the generated title. This label becomes available to downstream steps, such as:

* [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief)
* [Generate Article](/agents/workflows/nodes/content-creation/generate-article)
* [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard)
* Custom Prompt LLM steps

## How the node works behind the scenes

Article Title Suggestion runs a multi-step workflow that evaluates your inputs and produces one optimized title. While this process is abstracted into a single step, the internal logic includes:

1. **Input evaluation**
   * The node checks whether inputs like company name, audience segment, or target prompts are provided.
   * It adjusts title generation logic accordingly (e.g., more generic if fields are blank).
2. **Context and metadata preparation**
   * Current-year or temporal context may be added when relevant for accuracy or search alignment.
   * Input variables are transformed or normalized for consistent title generation.
3. **Title generation logic**
   * A structured set of editorial, SEO, and AEO rules is applied, such as:
     * Ensuring clarity and authority
     * Avoiding exaggerated or misleading phrasing
     * Matching user intent and search demand
     * Maintaining strong keyword alignment
     * Producing titles appropriate for your audience segment
4. **Optimized variant selection**
   * If multiple internal candidates are produced, the node selects the strongest one based on clarity, engagement potential, and AEO alignment.
   * Only **one title** is returned to maintain consistency and support downstream workflows.

This workflow ensures titles are more consistent and high-quality than ad-hoc LLM prompt outputs.

## Output

The output is a **single AEO-optimized article title**, suitable for

* Direct use in content production
* Passing into a brief or article generation workflow
* Use in editorial calendars
* SEO or topic clustering workflows

## Example usage

### 1. Create an optimized title before generating a content brief

1. Add the **Article Title Suggestion** node.
2. Provide topic, target prompt, and optional audience and company name.
3. Set an output label (e.g., `title`).
4. Pass `title` into the [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) node.

### 2. Build a content pipeline

1. Article Title Suggestion → [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) → [Generate Article](/agents/workflows/nodes/content-creation/generate-article)
2. Add [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard) afterward to evaluate the final article.

### 3. Generate multiple titles for testing

Duplicate the node with changes to Topic, Audience Segment, or Target Prompt to compare results.

## Best practices

* Include a **Target Prompt** when writing for AI-driven search; this increases intent alignment.
* Use **Audience Segment** for B2B or specialized verticals to ensure relevancy.
* Keep the **Topic** specific enough to guide the node but broad enough to allow creative generation.
* Downstream steps should reference the output label directly (e.g., for content brief creation).


# Create Content Brief node
Source: https://docs.tryprofound.com/agents/workflows/nodes/content-creation/create-content-brief

Generate a structured content brief tailored to your topic, audience, and brand

The **Create Content Brief** node generates a complete, structured content brief tailored to your topic, audience, and brand. This brief includes recommended angles, headers, research insights, platform considerations, and editorial guidance—built using Profound’s full content research and analysis pipeline.

Behind the scenes, this node performs multi-step research, citation analysis, AI-answer pattern detection, and outline generation to produce a brief that is both strategically sound and optimized for AEO.

## When to use this node

Use Create Content Brief when you need:

* A ready-to-use content brief for writers or downstream article generation
* A deep understanding of how AI systems currently answer a topic or prompt
* Strategic recommendations informed by real citation patterns
* Structured guidance for content teams or LLM-driven article generation
* AEO-optimized outlines that map to user intent and on-platform behavior

This node is ideal for launching a new article, planning an editorial strategy, or building automated content pipelines.

## Node configuration

Selecting the Create Content Brief node opens its configuration panel on the right side of the Workflow Builder.

### Topic (required)

The primary subject of the article, such as:

* “AI-driven supply chain optimization”
* “What is programmatic SEO?”
* “How to create an AEO content strategy”

This topic anchors the research and shapes the outline.

### Company Name (required)

Your company or client name. Used to contextualize the brief in terms of tone, positioning, brand relevance, and so on.

### Company Domain (required)

Your company’s website URL (e.g., `https://example.com`). This is used for internal link recommendations and context alignment.

### Article Title

If you’ve already generated a title—often via the [Article Title Suggestion](/agents/workflows/nodes/content-creation/article-title-suggestion) node—you may provide it here as a variable. If not provided, the node will still generate a fully informed brief.

### Audience Segment

The target audience for the content. Examples:

* “Marketing managers at B2B SaaS companies”
* “Healthcare procurement teams”
* “Small business owners”

This influences tone, examples, and depth.

### Content Type

The category or format of content to produce. Default is **General**, but other options include **Listicle, Blog Post, How-to**, and so on.

### Target Platforms

Select which AI platforms the content should optimize for. Examples include ChatGPT, Perplexity, Google AI Overviews, and others.

This ensures the analysis aligns with how specific platforms cite, structure, and present information.

### Target Prompts

These are user prompts your content should aim to answer or optimize for, for example:

* “What is AI visibility?”
* “How does supply chain automation work?”

This field enhances intent alignment and platform-specific optimization.

### Citations

List any manually provided sources you want the node to incorporate into the research process. This is helpful when working with proprietary or must-use sources.

### Additional Instructions

Anything else that should inform the content brief creation, such as special formatting or delivery notes.

### Output Label (required)

Assign a descriptive label for the brief, such as:

* `content_brief`
* `structured_brief`
* `brief_output`

## How the node works behind the scenes

Although it runs as a single step, the Create Content Brief executes a sophisticated, multi-stage workflow similar to the full Profound Content Creation pipeline. The internal process includes:

### 1. Input normalization and context setting

The node prepares all user-provided inputs, including topic, audience, company context, title, prompts, and platforms. It adapts the workflow based on what is or isn’t provided.

### 2. Target prompt and platform mapping

If target prompts or platforms are included, the node tailors research and outline strategy toward:

* How AI systems currently respond
* Which sources they cite
* Which answer patterns are most common
* Where gaps or opportunities exist

### 3. Citation retrieval and aggregation

The node fetches top citations across relevant platforms based on your topic, prompts, or keywords, including:

* Frequently cited sources
* High-authority domains
* Content types used in AI answers
* Structural patterns (lists, definitions, frameworks)

These citations form the core research dataset.

### 4. Deep research analysis

Using the aggregated citations, the node:

* Extracts themes
* Identifies recurring angles, frameworks, and definitions
* Detects gaps between competitor content and user intent
* Surfaces authoritative signals and trusted sources
* Highlights common misconceptions or ambiguous areas

This ensures the brief is grounded in real-world informational ecosystems.

### 5. Content brief assembly

The node synthesizes the research into a structured content brief, which typically includes:

* **Content summary** — what the article should aim to communicate
* **Key angle recommendations** — strategic viewpoints based on research
* **Suggested headers & structure** — optimized for readability and AEO
* **Research highlights** — distilled insights from citations
* **Target audience notes** — tone, complexity, and messaging guidelines
* **Platform considerations** — how to align with specific AI platforms
* **Target prompt alignment** — ensuring user-intent match
* **Internal linking strategy** — recommended internal pages from your domain
* **Call-to-action or conversion notes** (when appropriate)

### 6. Final formatting and delivery

The node formats the final deliverable as a clean, ready-to-use content brief.

## Output

The output is a **complete, polished, and AEO-ready content brief**.\
Writers, editors, and downstream workflow steps (such as [Generate Article](/agents/workflows/nodes/content-creation/generate-article)) can use it immediately.

The brief is designed to be:

* Structurally sound
* Research-backed
* Optimized for AI answer engines
* Tailored to your audience and brand
* Ready for production workflows

## Example usage

### 1. Full content creation pipeline

1. Generate a title using [Article Title Suggestion](/agents/workflows/nodes/content-creation/article-title-suggestion)
2. Pass title into **Create Content Brief**
3. Feed the resulting brief into [Generate Article](/agents/workflows/nodes/content-creation/generate-article)
4. Score the output with [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard)

### 2. Helpful for editorial workshops

Use the content brief to guide collaboration between strategists, writers, and SMEs.

### 3. Automated content operations

Run recurring workflows that generate briefs for:

* Product categories
* New topics in your SEO roadmap
* Common user prompts
* Competitor queries

## Best practices

* Provide a **Company Domain** to enable internal linking insights.
* Use **Target Prompts** to boost alignment with AI-answer behavior.
* Add **Target Platforms** to tailor the brief to specific citation ecosystems.
* Pair with **Answer Engine Insights** for a data-driven pre-brief stage.
* Use clear, consistent output labels to chain into downstream nodes.


# Generate Article node
Source: https://docs.tryprofound.com/agents/workflows/nodes/content-creation/generate-article

Produce a publication-ready article tailored to your topic, audience, and brand

The **Generate Article** node produces a complete, publication-ready article tailored to your topic, audience, and brand. It is the final step in Profound’s content creation pipeline and is designed to work seamlessly with upstream Workflow nodes such as [Article Title Suggestion](/agents/workflows/nodes/content-creation/article-title-suggestion) and [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief).

Unlike a single LLM call, this node runs a structured, multi-stage workflow that drafts, evaluates, refines, and brand-aligns the final article. The result is a polished, AEO-optimized piece that is consistent, trustworthy, and ready for editing or direct publication.

## When to use this node

Use Generate Article when you need:

* A full-length, coherent, well-structured article
* Consistent voice and tone that aligns with your company or audience
* AEO-aligned phrasing and structure optimized for AI-generated answers
* Articles informed by a full content brief or standalone inputs
* Automated content pipelines for SEO, editorial teams, or content operations

This node is ideal as the final stage in automated or semi-automated content production workflows.

## Node configuration

Selecting the Generate Article node opens its configuration panel on the right side of the Workflow Builder.

### Article Title (required)

Provide the final working title of the article. This is the anchor for narrative structure, tone, and topical focus.

It's most common to pass the output from [Article Title Suggestion](/agents/workflows/nodes/content-creation/article-title-suggestion) into this field.

### Audience Segment

Specify who the article is for, for example:

* “Retail operations managers”
* “Enterprise IT leaders”
* “Founders of early-stage startups”

This helps adjust tone, vocabulary, and level of explanation.

### Company Name

Adds brand context to the writing. Useful when the company should be referenced or when tone alignment matters.

### Content Brief

Paste or reference the output of the [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) node. It is used to:

* Structure the article based on the brief outline
* Incorporate key points, angles, and metadata
* Align with recommended headers, insights, and prompts

If you don't provide a brief, the node still generates a high-quality article using the title and audience alone, but without the advanced research backing.

### Additional Instructions

Any special requirements such as:

* Formatting preferences
* Markdown or HTML structure
* Reading-level constraints
* Examples to incorporate
* Style guidelines (e.g., “conversational but authoritative”)

These instructions override or augment the default behavior.

### Target Prompt

If your article must optimize for a specific user query used in AI search, include it here.

Example:

* “How is AI transforming retail supply chains?”

This helps align the introduction, framing, and definitions with answer engine behavior.

### Output Label

Name the output variable. Example: `generated_article`.

## How the node works behind the scenes

Generate Article executes a multi-step workflow designed for quality, stability, and alignment:

### 1. Input validation and branching

The workflow determines whether a **Content Brief** was provided.

* **If brief exists:** The node follows the outline closely, using it as the blueprint for structure, sections, and insights.
* **If brief is empty:** The node generates an internal outline based on topic and title, ensuring readability and cohesion.

### 2. First-pass drafting

Using a high-quality generative model, the node produces the initial draft.\
The draft incorporates:

* Title framing
* Target audience considerations
* Brand context
* Brief-based structure (if provided)
* Any additional instructions

This is the most substantial stage of article creation.

### 3. Draft consolidation

If the workflow generates multiple variants or sections, these are merged and processed into a single cohesive article. Variable aggregation ensures:

* Consistent narrative voice
* Smooth transitions
* Unified structure

### 4. Optional refinement and QA logic

If certain fields are provided (e.g., extensive instructions or company name), the workflow may run additional logic to adjust tone, expand examples, or reinforce brand positioning.

### 5. Brand alignment pass

A final refinement stage ensures the article:

* Reads as brand-safe
* Maintains proper terminology
* Uses the desired voice and perspective
* Aligns with AEO-based clarity and trust guidelines

This step often employs a lighter model to polish and finalize the text.

### 6. Final output formatting

The node outputs the complete article as a single, clean block of text.

## Output

The output is a **fully drafted, publication-ready article** that can be:

* Published directly
* Edited by a human writer
* Scored by the [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard)
* Used in automated content distribution workflows

The output generally includes:

* A compelling introduction
* Clear sections and headers
* Well-structured explanations
* Actionable insights
* A conclusion that reinforces the article’s value

## Example usage

### 1. Full content pipeline

1. Generate title
2. Create brief
3. Generate article
4. Score and optimize using AEO Content Scorecard

### 2. Standalone article creation

Provide a title + audience segment + additional instructions for a quick, single-step content generation workflow.

### 3. Automated multi-topic content series

[Iterate](/agents/how-tos/add-iterations-to-workflows) over a list of topics, feeding each into this node to produce a series of consistent, high-quality articles.

## Best practices

* Always provide a **Content Brief** for the strongest results—this unlocks the research-driven outline.
* Use **Audience Segment** to ensure the article resonates with the right people.
* Add **Additional Instructions** for custom formatting or editorial rules.
* Use **Target Prompt** when optimizing for AI search queries.
* Pair with the [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard) to evaluate and refine output.


# AEO Content Scorecard node
Source: https://docs.tryprofound.com/agents/workflows/nodes/content-optimization/aeo-content-scorecard

Score how well an article performs for Answer Engine Optimization

The **AEO Content Scorecard** node evaluates how well a given article performs for Answer Engine Optimization (AEO). It analyzes the URL you provide, retrieves and inspects the content, and produces a structured scorecard with category-level scoring, weighted factors, and specific, actionable recommendations.

This node uses the same scoring logic that powers Profound's Content Optimization feature, ensuring consistency between your Workflow outputs and the optimization experience in the Profound platform.

Use this node when you want to diagnose how well an article aligns with the way AI systems evaluate relevance, clarity, structure, and answerability—and when you want to programmatically trigger optimization workflows based on these results.

## When to use this node

Use the AEO Content Scorecard to:

* Audit existing content for AEO readiness
* Prioritize content refresh opportunities
* Trigger automated improvement workflows
* Benchmark your content against top-ranking competitors
* Provide structured optimization recommendations to writers or LLMs
* Standardize content scoring across your organization

This node is highly effective in both human-driven editorial processes and fully automated content optimization pipelines.

## Node configuration

Selecting the AEO Content Scorecard node opens its configuration panel on the right side of the Workflow Builder.

### URL (required)

Enter the URL of the article you want to evaluate.

This can be:

* A published content page
* A staging URL
* A temporary hosting URL
* Any accessible webpage containing the article text

The node retrieves the content and applies a full analysis across readability, structure, answerability, machine readability, and more.

### Target Prompt (required)

Enter the user prompt or question your content should be optimized for.

Example:

* “How do keyword hierarchies work in AI search?”
* “What is predictive maintenance in manufacturing?”

This helps evaluate whether the article directly matches user intent as seen in AI-generated answers.

### Output Label

Assign a descriptive label to access the scorecard output in downstream steps.

Examples:

* `aeo_scorecard`
* `content_score`
* `scorecard_output`

## How the node works behind the scenes

AEO Content Scorecard runs a multi-step evaluation pipeline:

### 1. Content retrieval

The node fetches the webpage content from the provided URL and prepares it for analysis.

### 2. AEO-focused content parsing

The underlying models identify:

* Text structure
* Header hierarchy
* Answerability signals
* Entity coverage
* URL patterns
* Schema presence
* Internal linking
* Machine readability cues

### 3. Scoring across AEO categories

The node produces a weighted score across categories, commonly including:

* **Readability**
* **Content Freshness**
* **Content Structure**
* **Answerability Signals**
* **Machine Readability**
* **Information Density**

Each category is scored and flagged (e.g., green, yellow, red).

### 4. Top recommendations

The node identifies the highest-impact improvements based on:

* AEO principles
* Competitor benchmarks
* Structural inconsistencies
* Opportunities to simplify, clarify, or enhance the content

Recommendations include:

* Before/after examples
* URL improvements
* Schema suggestions
* Title enhancements
* Subheading updates
* Opportunities to increase topical relevance

### 5. Final score assembly

The node compiles a structured scorecard containing:

* Final Score
* Target Score Range (based on top competitor performance)
* Detailed category breakdown
* Actionable recommendations

## Output

The resulting output is a structured report similar to the following example:

>

> [!NOTE]
>
> ### AEO Content Scorecard
>
> **URL:**\
> ` https://www.tryprofound.com/blog/introducing-keyword-hierarchies`
>
> **Final Score:** 72/100\
> **Target Zone (Top Competitors):** 62–72
>
> #### Category Breakdown
>
> | **Category** | **Score** | **Status** |
> | - | - | - |
> | Readability | 64 | 🟡 |
> | Content Freshness | 91 | 🟢 |
> | Content Structure | 67 | 🟡 |
> | Answerability Signals | 81 | 🟢 |
> | Machine Readability | 63 | 🟡 |
> | Information Density | 21 | 🔴 |
>
> #### Top Recommendations
>
> 1. **Enhance Subheading Relevance**\
>    Improve clarity by explicitly mentioning “Keyword Hierarchies in AI Conversations.”\
>    *Before:* “The technical innovation”\
>    *After:* “How Keyword Hierarchies Enhance AI Conversations”
> 2. **Simplify the URL Structure**\
>    Shorten the URL to make it more topical and easier for AI systems to parse.
> 3. **Implement FAQ Schema Markup**\
>    Add FAQ entries such as “What are keyword hierarchies?” to better support answer engines.
> 4. **Add Superlatives to the Title**\
>    Example improvement:\
>    *Before:* “Introducing Keyword Hierarchies”\
>    *After:* “Discover the Most Effective Keyword Hierarchies for AI Conversations”

## Example Workflow: AEO Content Refresh

Here is how you might use this node inside a real optimization Workflow aimed at automatically improving existing articles and republishing them.

**Steps:**

1. **AEO Content Scorecard**
   * Input the article URL
   * Retrieve a full AEO scorecard and improvement recommendations
2. **[Prompt LLM](/agents/workflows/nodes/prompt-llm): generate improved article**
   * Provide the scorecard and instruct the model to apply the recommended improvements
   * Rewrite the article with updated structure, headings, clarity, and answerability
   * Maintain original brand voice
   * Example prompt:

     ```text wrap theme={null}
     Here is the AEO Content Scorecard for this article.
     Apply all recommendations, rewriting the content where necessary.
     Produce an improved article that would score at least 10 points higher
     while maintaining factual accuracy and brand voice.

     {{aeo_scorecard}}
     ```
3. **[Prompt LLM](/agents/workflows/nodes/prompt-llm): validate against scorecard criteria**
   * Ensure the updated article meets or exceeds recommended category scores
4. **Publish or Update Step**
   * Use [Call API](/agents/workflows/nodes/call-api) or CMS publishing logic to update the article on your site

This workflow allows teams to create scalable, repeatable content optimization systems powered by Profound.

## Best Practices

* Use the **Target Prompt** field whenever the article exists to answer a specific question.
* Pair this workflow with the [Generate Article node](/agents/workflows/nodes/content-creation/generate-article) for iterative optimization loops.
* Use clear output labels when chaining scorecards into rewrite steps.
* Run this node periodically in recurring Workflow runs to detect content decay.


# Query Fanout Estimator node
Source: https://docs.tryprofound.com/agents/workflows/nodes/content-optimization/query-fanout-estimator

Predict how answer engines expand a single prompt into multiple web search queries

The **Query Fanout Estimator** node predicts how answer engines (like ChatGPT, Claude, Gemini, or Perplexity) are likely to expand a single user prompt into multiple **web search queries**. These expanded searches—called [**query fanouts**](https://www.tryprofound.com/blog/introducing-query-fanouts)—are what actually drive retrieval and citations in AI-generated answers.

This node helps you understand the search layer behind the answer: which queries an answer engine might run, what facets of intent it will explore, and where your content needs coverage to be cited more often.

## When to use this node

Use the Query Fanout Estimator when you want to:

* See how a specific **user prompt** is likely to be decomposed into multiple web searches
* Identify the **high-intent queries** that matter most for a given prompt
* Plan content that covers the full **fanout space** behind important prompts
* Feed likely fanout queries into AEO data nodes, content briefs, or keyword clustering workflows
* Model how answer engines translate prompts into retrieval behavior for AEO strategy

## Node configuration

Selecting the Query Fanout Estimator node opens its configuration panel on the right side of the Workflow Builder.

### Target Prompt (required)

The user prompt that your content should optimize for, such as:

* “How do keyword hierarchies work in AI search?”
* “Best tools for monitoring AI citations for my brand”
* “How does query fan-out impact SEO strategy?”

The node will estimate which web search queries an answer engine is likely to run when answering this prompt.

### Output Label (required)

A descriptive name for the node’s output, for example:

* `query_fanout_estimate`
* `predicted_fanout_queries`

You will use this label to reference the fanout data in downstream steps (e.g., content briefs, research steps, reports).

## How the node works behind the scenes

This node runs a two-step pipeline:

### 1. Retrieve real query fanout examples

Profound maintains a knowledge base of **historical query fanouts**: real pairs of user prompts and web search queries that answer engines actually ran behind the scenes (the “fanout” set).

Given your **Target Prompt**, the node:

1. Performs a semantic search over this fanout dataset
2. Retrieves the most similar prompt–fanout examples
3. Passes those examples forward as in-context demonstrations

These examples are not restricted to the same topic; they illustrate how Answer Engines in general expand prompts into multiple searches.

### 2. Estimate fanout for your prompt using an LLM

Next, the node uses a model configured specifically to “think like” an Answer Engine’s retrieval layer:

* It receives:
  * Your **Target Prompt**
  * A set of historical prompt + fanout examples from Profound’s knowledge base
* It is instructed to:
  * Infer how an answer engine would break your prompt into multiple web search queries
  * Generate **realistic, high-intent, semantically diverse** search queries
  * Focus on **sub-queries** that would retrieve relevant sources (not just phrasing variants)
  * Capture different facets of the original intent (definitions, comparisons, how-to steps, evaluation, etc.) ([Profound](https://www.tryprofound.com/blog/expanding-analysis-for-query-fanouts-in-profound?utm_source=chatgpt.com))

The model returns a list of predicted web search queries—your **estimated query fanout**.

## Output

The node returns a structured text output containing likely search queries that answer engines would fan out from your prompt. For example:

```text wrap theme={null}
what is query fan-out in ai search, how do llms expand user prompts into multiple web searches, impact of query fan-out on brand visibility, examples of query fan-out in google ai overviews,how to optimize content for query fan-out
```

You can parse this text into a list in downstream steps or feed it directly to other nodes.

## Example workflow: Fanout-informed content planning

**Goal:** Build a content plan that covers the full set of queries an answer engine might run for a critical prompt.

**Steps:**

1. **Query Fanout Estimator**
   * Input: Target Prompt (e.g., “How do AI-generated answers choose citations?”)
   * Output: Predicted fanout queries
2. **Research and insights**
   * For each fanout query:
     * Use [Perplexity Search](/agents/workflows/nodes/perplexity-search) or [Web Page Scrape](/agents/workflows/nodes/web-page-scrape) to gather citations
     * Use [Research Snippets Generator](/agents/workflows/nodes/research/research-snippets-generator) to extract facts and quotes
3. **Create content brief**
   * Feed the fanout queries + research into the [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) node
   * Ensure the brief mandates coverage for each high-intent sub-query
4. **Generate article**
   * Use the brief to [Generate Article](/agents/workflows/nodes/content-creation/generate-article) that explicitly addresses the main prompt *and* its fanouts
5. Score the final article
   * Use [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard) to checking whether the article covers enough of the fanout space implied by the prompt

This workflow ensures your content isn’t just optimized for a single phrasing—but for the **full set of searches** answer engines are likely to run behind that prompt.

## Best practices

* Use natural language prompts that mirror real user questions; the fanout will be more realistic.
* Store the fanout output in a reusable label (e.g., query\_fanout\_estimate) so it can power multiple downstream steps.
* Combine this node with Profound AEO data nodes to see how often your site appears across the predicted fanout queries.
* Use fanout results to design FAQs, headings, and section structure that map directly to the likely sub-queries answer engines care about.


# Citation Domains node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/citation-domains

Retrieve hostname or root domain citation data from Profound

The [Citation Domains](/answer-engine-insights/interpret#citations) node retrieves hostname or root domain-level citation data.

Use this step when you want to understand which domains are most frequently cited in AI answers referencing your brand. It can be used when:

* Identifying top citing domains
* Understanding media or partner visibility
* Prioritizing digital PR outreach
* Analyzing domain-level trends

## Node configuration

Selecting the Citation Domains node opens its configuration panel on the right side of the Workflow Builder.

### Metric

* Citation Share
* Citation Count

### Date Range

Select the time window.

### Filters

Use filters to narrow the dataset.

Fields may include:

* Analysis Types
* Citation Categories
* Hostnames
* Regions
* Personas
* Platforms
* Prompts
* Tags
* Topics

You can paste a static value or insert a value dynamically from earlier Workflow steps by typing `/`.

When filtering on a field, you must also include that field as a dimension in the field settings.

### Output Format

Examples:

* JSON
* Markdown Table
* Domain List

### Output Label (required)

Examples:

* `citation_pages`
* `cited_urls`

## Advanced settings

### Date Interval

Group the date interval by day, week, month, or year.

### Limit

Set the maximum number of rows returned by the query. This is helpful when using multiple dimensions or sending data into LLM steps.

### Sort Direction

Sort the output in ascending (Asc) or descending (Desc) order.

## Output

Returns a list of domains with associated citation metrics.


# Citation Pages node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/citation-pages

Pull cited pages with share of voice and citation counts from Profound

The [Citation Pages](/answer-engine-insights/interpret#all-citations) node pulls cited pages from Profound. Retrieve pages with share of voice and citation counts, with optional filters such as personas and domain.

Use this step when you want:

* Page-level visibility into citations

* Identifying which specific URLs are cited

* Prioritizing SEO or content updates

* Auditing citation coverage

* Building outreach workflows

## Node configuration

Selecting the Citation Pages node opens its configuration panel on the right side of the Workflow Builder.

### Metric

* Citation Share
* Citation Count

### Date Range

Select the time window.

### Filters (optional)

Use filters to narrow the dataset.

Fields may include:

* Analysis Types
* Citation Categories
* Hostnames
* Regions
* Personas
* Platforms
* Prompts
* Tags
* Topics

You can paste a static value or insert a value dynamically from earlier Workflow steps by typing `/`.

When filtering on a field, you must also include that field as a dimension in the field settings.

### Output Format

Examples:

* JSON
* Markdown Table
* URL List

### Output Label

Examples:

* `citation_pages`
* `cited_urls`

## Advanced settings

### Date Interval

Group the date interval by day, week, month, or year.

### Limit

Set the maximum number of rows returned by the query. This is helpful when using multiple dimensions or sending data into LLM steps.

### Sort Direction

Sort the output in ascending (Asc) or descending (Desc) order.

## Output

Returns a list of URLs with associated citation metrics.


# Citation Share node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/citation-share

Retrieve share-of-voice citation data from Profound

The [Citation Share](/answer-engine-insights/interpret#citation-share-and-citation-coverage) node retrieves share-of-voice citation data from your Profound account. Use this step when you want to understand how often your brand is cited relative to competitors within AI-generated answers.

* Measuring competitive share of citations
* Tracking citation share trends over time
* Comparing citation presence by region or model
* Feeding competitive data into LLM analysis

## Node configuration

Selecting the Citation Share node opens its configuration panel on the right side of the Workflow Builder.

### Metric

* Citation Share
* Citation Count

### Date Range

Select the reporting window.

### Filters

Use filters to narrow the dataset.

Fields may include:

* Analysis Types
* Citation Categories
* Hostnames
* Regions
* Personas
* Platforms
* Prompts
* Tags
* Topics

You can paste a static value or insert a value dynamically from earlier Workflow steps by typing `/`.

When filtering on a field, you must also include that field as a dimension in the field settings.

### Output Label (required)

Examples:

* `citation_share`
* `weekly_citation_share`

## Advanced settings

### Date Interval

Group the date interval by day, week, month or year.

### Limit

Set the maximum number of rows returned by the query. This is helpful when using multiple dimensions or sending data into LLM steps.

## Output

Returns citation share values grouped by selected dimensions.


# Citation Watched Pages node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/citation-watched-pages

Retrieve citation data for the pages you watch in Profound

The **Citation Watched Pages** node retrieves citation data specifically for pages you have designated as watched pages in Profound.

Use this step when you want focused monitoring of priority URLs. Use this node when:

* Monitoring high-priority product pages
* Tracking citations for strategic landing pages
* Alerting when watched pages lose citation coverage
* Evaluating performance of key content assets

## Node configuration

Selecting the Citation Watched Pages node opens its configuration panel on the right side of the Workflow Builder.

### Metric

* Citation Share
* Citation Count

### Date Range

Select the time window.

### Filters

Use filters to narrow the dataset.

Fields may include:

* Analysis Types
* Citation Categories
* Hostnames
* Personas
* Platforms
* Prompts
* Regions
* Tags
* Topics

You can paste a static value or insert a value dynamically from earlier Workflow steps by typing `/`.

When filtering on a field, you must also include that field as a dimension in the field settings.

### Output Format

Examples:

* JSON
* Markdown Table
* URL List

### Output Label

Examples:

* `citation_pages`
* `cited_urls`

## Advanced settings

### Date Interval

Group the date interval by day, week, month or year.

### Limit

Set the maximum number of rows returned by the query. This is helpful when using multiple dimensions or sending data into LLM steps.

## Output

Returns a list of URLs with associated citation metrics.


# Sentiment node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/sentiment

Pull positive and negative sentiment scores from Profound into a Workflow

> [!TIP]
>
> To learn more about Sentiment metrics, see the [About Sentiment](/answer-engine-insights/sentiment/about) section.

The Sentiment node pulls current Sentiment data from your Profound account into a Workflow. It returns a positive or negative sentiment score for one or more brands or competitors, on a scale from 0 to 1. The score is model-balanced: it is the average across AI platforms, so no single platform controls the score.

Use the Sentiment node for tasks such as:

* Tracking positive and negative sentiment toward your brand over time
* Comparing sentiment between your brand and competitors
* Finding sentiment differences across platforms, regions, topics, or themes

> [!NOTE]
>
> The Sentiment node is the newer version of the [Sentiment Score node](/agents/workflows/nodes/data/sentiment-score), which uses an earlier version of Sentiment data.

## Node configuration

Selecting the Sentiment node opens its configuration panel on the right side of the Workflow Builder.

### Assets

Select one or more brands or competitors to query. The node queries each asset separately and labels each row with the asset name.

### Metric

Choose the sentiment metric to return:

* **Positive Sentiment** – The share of sentiment toward the asset that is positive.
* **Negative Sentiment** – The share of sentiment toward the asset that is negative.

The node returns each score as a ratio from 0 to 1. For example, `0.72` means 72%.

### Date range

Specify the time window to query. Options include:

* Last 2 Days
* Last 7 Days
* Last 14 Days
* Last 4 Weeks
* Last 3 Months
* Custom date range

If you select **Custom date range**, enter a start date and an end date.

### Filters

Use filters to narrow the dataset. Fields include:

* Tag
* Topics
* Platforms
* Personas
* Regions
* Theme

Select values from the list, or switch the filter to **Variable** to use a value from an earlier Workflow step. The Theme list comes from the first asset you selected.

> [!NOTE]
>
> When a filter value comes from a variable, Tag, Platforms, Personas, and Regions expect IDs. Topics and Theme accept names.

To group results by a filter field, open the filter's settings and select **Add \[field] as a dimension**. You can group by up to two dimensions.

### Output label

Examples:

* `sentiment`
* `weekly_positive_sentiment`
* `negative_sentiment_by_platform`

## Advanced settings

* **Date interval**: Group the results by day, week, or month.
* **Limit**: Set the maximum number of rows the node returns. The default is 100 and the maximum is 10,000. The limit applies to the combined results for all selected assets.
* **Sort by**: Sort the output by date or by the selected metric.
* **Sort direction**: Sort the output in ascending or descending order.

## Output

Returns a list of rows. Each row includes:

* `asset_name` – The brand or competitor.
* `date` – The date interval that the score covers.
* `metric_value` – The selected sentiment score, from 0 to 1.
* One field for each dimension you group by, such as `platform` or `region`.

This output is commonly used in:

* LLM narrative summaries
* Reputation monitoring Workflows
* Competitive sentiment reports

> [!TIP]
>
> Check out [Create a Workflow](/agents/workflows/create-workflow) to learn how to add this node to a Workflow.


# Sentiment Score node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/sentiment-score

Pull brand sentiment data from Profound into a Workflow

> [!WARNING]
>
> The Sentiment Score node uses an earlier version of Sentiment data and will be deprecated soon. For current Sentiment data, use the [Sentiment node](/agents/workflows/nodes/data/sentiment).

The Sentiment Score node pulls brand sentiment data from your Profound account into a Workflow. Use this step when you want to analyze positive, negative, or neutral sentiment toward your brand in AI answers.

This node is optimized for understanding perception trends and tone shifts over time.

Use Sentiment Score for tasks such as:

* Tracking positive vs. negative AI answer sentiment
* Identifying sentiment changes after product updates
* Comparing sentiment across models or regions

## Node configuration

Selecting the Sentiment Score node opens its configuration panel on the right side of the Workflow Builder.

### Metric

Choose the sentiment metric to return. Options include:

* Positive Count
* Negative Count
* Total Occurrences
* Positive Ratio (%)
* Negative Ratio (%)

### Date Range

Specify the time window to query (for example, Last 7 Days).

The node will return sentiment data only for answers within this range.

### Filters

Use filters to narrow the dataset.

Fields may include:

* Analysis Types
* Asset
* Citation Categories
* Hostnames
* Regions
* Personas
* Platforms
* Prompts
* Tags
* Topics

You can paste a static value or insert a value dynamically from earlier Workflow steps by typing `/`.

When filtering on a field, you must also include that field as a dimension in the field settings.

### Output Label

Examples:

* `sentiment_score`
* `weekly_sentiment`
* `sentiment_by_model`

## Advanced settings

### Date Interval

Group the date interval by day, week, month or year.

### Limit

Set the maximum number of rows returned by the query. This is helpful when using multiple dimensions or sending data into LLM steps.

### Sort By

Sort the output by metric or date.

### Sort Direction

Sort the output in ascending (ASC) or descending (Desc) order.

## Output

Returns structured sentiment metrics grouped by any selected dimensions.

This output is commonly used in:

* LLM narrative summaries
* Reputation monitoring Workflows
* Executive dashboards


# Visibility Score node
Source: https://docs.tryprofound.com/agents/workflows/nodes/data/visibility-score

Pull brand visibility data from Profound into a Workflow

The **Visibility Score node** pulls brand visibility data from your Profound account and into a Workflow. Use this node as a step when you want to measure how prominently your brand appears in AI-generated answers. You can then pass the data into downstream steps such as content generation.

This node is built for tracking overall brand presence across models, regions, and topics.

Use the Visibility Score node for tasks such as:

* Monitoring how often your brand appears in AI answers
* Measuring share of voice over time
* Comparing visibility across regions or topics
* Tracking performance before and after campaigns
* Triggering alerts when visibility drops below a threshold

## Node configuration

Selecting the Visibility Score node opens its configuration panel on the right side of the Workflow Builder.

### Metric

Choose the visibility-related metric to return. Options include:

* Visibility Score
* Share of Voice

### Date Range

Specify the time window to query (for example, Last 7 Days).

The node will return visibility data only for answers within this range.

### Filters

Use filters to narrow the dataset. Options include:

* Asset
* Personas
* Platforms
* Prompts
* Regions
* Topics
* Tags

You can paste a static value or insert a value dynamically from earlier Workflow steps by typing `/`.

When filtering on a field, you must also add that field as a dimension in the field settings.

### Output Label

Provide a descriptive label for the node’s result. This becomes the variable name used in downstream steps.

Examples:

* `visibility_score`
* `weekly_visibility`
* `regional_share_of_voice`

## Advanced settings

### Date Interval

Group the date interval by day, week, month, or year.

### Limit

Set the maximum number of rows returned by the query. This is helpful when using multiple dimensions or sending data into LLM steps.

### Sort By

Sort the output by metric or date.

### Sort Direction

Sort the output in ascending (**Asc**) or descending (**Desc**) order.

## Output

The Visibility Score node returns structured data containing:

* Selected visibility metric
* Any grouping dimensions
* Values constrained by filters and date range

This output can be consumed by downstream nodes including:

* Prompt LLM nodes
* Reporting Workflows
* Alerting workflows


# Research Snippets Generator node
Source: https://docs.tryprofound.com/agents/workflows/nodes/research/research-snippets-generator

Produce citation-backed research snippets to enrich articles and briefs

The **Research Snippets Generator** node produces a curated set of factual, citation-backed research snippets that you can use to enrich articles, briefs, and other content assets. These snippets are extracted directly from high-authority sources using structured research workflows—dramatically reducing hallucinations and ensuring accuracy.

Each snippet is short, quotable, and includes its source URL, making it ready for direct insertion into content or for use in downstream analysis or generation steps.

Unlike broad summarization nodes, this one focuses on evidence, not narrative: facts, statistics, quotes, and verifiable claims.

## When to use this node

Use the Research Snippets Generator when you need:

* Accurate facts and citations to support an article
* Quickly sourced statistics, definitions, quotes, or authoritative statements
* Verifiable evidence for AEO-friendly content
* Study materials for briefs, outlines, or LLM-driven content creation
* A clean set of structured research snippets for editorial workflows

This node is ideal for writers, strategists, and automated content systems that require reliability and transparency.

## Node configuration

Selecting the Research Snippets Generator node opens its configuration panel on the right side of the Workflow Builder.

### Article Title (required)

The topic or title that the node should research, for example:

* “Emerging trends in healthcare technology”
* “How generative AI improves supply chain forecasting”
* “What is AI visibility and why does it matter?”

This title directs the research tools and extraction models.

### Audience Segment

Adds context so the node can select sources and snippets that better match the intended reader. For example:

* “Enterprise CIOs”
* “Healthcare operations managers”
* “B2B SaaS marketers”

### Company Name

Used for contextual alignment when sourcing or selecting certain snippet styles (e.g., B2B tone vs. consumer tone).

### Content Type

Defaults to **General**, but may support variations depending on your workspace configuration. This setting helps the node adjust how granular or technical the snippets may be.

### Target Prompt

If you want the research snippets to specifically support a user prompt used in AI search or AEO workflows, enter it here. For example:

* “How does AI reduce logistics costs?”

### Output Label (required)

Provide a variable name for the resulting JSON structure containing the snippets.

Examples:

* `research_snippets`
* `evidence_blocks`
* `snippet_list`

## How the node works behind the scenes

The underlying workflow of this node includes:

### 1. Topic-focused research

The node queries multiple research tools and models, including systems like Perplexity, to gather:

* High-authority webpages
* Government or academic reports
* Industry data
* News articles
* Trusted organizational publications

This process surfaces a large pool of raw research material.

### 2. Aggregation and de-duplication

All retrieved research is consolidated into a structured dataset. The workflow filters:

* Low-authority sources
* Redundant information
* Irrelevant tangents
* Non-citable content

Only reliable materials move forward.

### 3. Evidence extraction

An LLM configured as an AEO Snippet Extractor processes the structured research and identifies extractable facts. Snippets must meet strict criteria:

* At least **20 snippets** must be produced (goal: 20–40)
* Each snippet must be **factually accurate**
* Each snippet must be **directly supported by a cited source**
* Snippets must be **under 160 characters**
* Avoid hallucinations by extracting text, *not inventing it*
* Prefer .gov, .edu, associations, reports, or major media
* Verbatim text is allowed when accuracy is critical
* Every snippet includes a **URL**

This is designed for high trustworthiness, especially in enterprise or regulated contexts.

### 4. JSON formatting

The node packages the results into a clean JSON structure, ideal for downstream workflows such as:

* Article generation
* Snippet clustering
* Fact-checking
* Scorecards
* Human editorial review

## Output

You receive a structured JSON object containing **20–40 evidence snippets**, each with a corresponding URL.

These snippets can be used to:

* Enrich articles
* Support claims in content briefs
* Provide trustworthy evidence for LLM prompts
* Generate research summaries
* Accelerate editorial workflows
* Power automated article creation pipelines

## Example usage

### 1. Fact-enriched article generation

1. Generate a title with [Article Title Suggestion](/agents/workflows/nodes/content-creation/article-title-suggestion)
2. Run **Research Snippets Generator**
3. Feed snippets + brief into [Generate Article](/agents/workflows/nodes/content-creation/generate-article)
4. Produce a deeply factual article grounded in authoritative sources

### 2. Human editorial workflows

Editors can pull snippets directly into outlines, paragraphs, or fact boxes.

### 3. Automated content QA

You can use snippets to validate whether article content aligns with cited facts.

## Best practices

* Use a **clear and specific Article Title** to ensure relevant research.
* Pair with [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) for the strongest content strategy alignment.
* Include a **Target Prompt** if writing for AI-answer funnels.
* Snippets work extremely well as grounding data in prompt LLM steps.


# Workflow runs
Source: https://docs.tryprofound.com/agents/workflows/workflow-runs

What a Workflow run is, how to start one, and how to review its results

A Workflow run is a single execution of a Workflow. It starts when an input is submitted and ends when the final output is generated. While a Workflow provides the reusable structure (its steps, nodes, and logic), each run represents a unique instance of that Workflow in action.

Workflow runs let you automate analysis, generate insights, and execute connected tasks consistently and repeatedly, without needing to reconfigure the Workflow each time.

## Test Workflow runs

Test runs let you track the inputs, outputs, and variables of each step more closely. From the Workflow Builder, you can view the results of a test run, including:

* The steps executed.
* The variables produced at each stage.
* The final output returned by the Workflow.

This makes it easy to troubleshoot, refine prompts, or verify that your Workflow logic behaves as expected.

Trigger a test run by clicking **Run Test** in the upper right-hand corner of the Workflow Builder.

![Screen Shot 2025-12-09 at 4.27.56 PM.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/agent-runs-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=27ea6305a2f0ff475e892c0de8f16c84)

> [!TIP]
>
> You can run tests for both published and draft Workflows. To run a test from a published Workflow, go to the Workflow page, select **Edit** on the top right of the page, and test the Workflow in the Workflow Builder.

## Trigger a Workflow run

A Workflow run begins when you:

* Select **Run** for the desired Workflow
* Provide input

Each run processes the values you provide, transforms them through each step, and produces a final output.

## How Workflow runs work

During a run, the Workflow's steps execute in a linear order:

1. **Inputs are captured**. The Workflow records the values you provide and stores them as variables.
2. **Steps execute sequentially**. Each node performs its action, such as running Answer Engine Insights, prompting a model, or calling an external tool.
3. **Variables pass between steps**. Outputs from each step become new variables that can be reused or combined later in the run.
4. **A final output is generated**. The run ends at the End step, producing a result or simply marking the run’s completion.

Because each run is independent, the same Workflow can produce different results based on the inputs you provide or data retrieved during the process.

## View run results

Workflow outputs from the latest run appear on the Workflow's page. Select **Past runs** to see the outputs for all runs of the Worklow.


# About Profound's AI Marketer (Aim)
Source: https://docs.tryprofound.com/aim/about

What AI Marketer (Aim) is and how it analyzes your data, suggests work, and gets it done

## What is Profound's AI Marketer

AI Marketer, also known as Aim, is your AI marketing teammate in Profound. It works across your AI search data, helps your team analyze it, proactively suggests the work worth doing, and helps you get it done. It also collects and uses the business context you add to Profound.

Aim powers [Projects](/projects/about), and every chat surface across the platform, including [Profound Docs](/documents/about), [Sheets](/sheets/getting-started), and Slack. Whenever you chat with Profound, you're talking to Aim.

## What AI Marketer can do

Aim handles analysis, execution, and everything in between: it analyzes your data in the background, suggests Projects and tasks, chats with you wherever you work, and can use your Agents to do the work.

### Answer questions about your data

Start a conversation wherever you see an Aim icon: inside an Agent, a Project, or a Sheet, in the Answer Engine Insights charts, or in Slack.

![Visibility tab with the AI Marketer button highlighted in the top right corner](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/aim/about-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=e7d9530d1e91f3d70d5d079eba227e7c)

![Visibility Score chart tooltip with an arrow pointing to the Click to investigate option](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/aim/about-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=6065c2f9f7cc3a08a93ea4e66851be0f)

Ask why a metric moved and what changed, for example:

* "Why did our visibility on ChatGPT drop last week?"
* "Which competitor is outperforming us on the pricing topic, and why?"
* "What's driving negative sentiment for our brand this month?"

Then watch Aim scan for peaks and valleys across Visibility, Citation Share, Sentiment, and Prompt Volumes data, and explain the drivers.

Beyond questions, ask Aim to build charts, tables, and [dashboards](/custom-dashboards/about), or to edit Agent outputs in Profound Docs without leaving the conversation.

### Work on your goals in Projects

A Project gives your Aim chats shared purpose, instructions, and context, such as files and Knowledge Base documents. Create a Project from the **Projects** page, or ask Aim in the home page chat box to start one. Learn more in [About Projects](/projects/about).

### Act on Suggestions

Aim analyzes your data and context and suggests new Projects and tasks. Select **Start task** on a Suggestion, and Aim starts working on a plan, such as refreshing a declining page, drafting a head-to-head comparison article, or writing outreach to a top-cited publication. Learn more in [Suggestions from Aim](/projects/suggestions-from-aim).

## AI Marketer's context

Aim is grounded in your organization's context. This grounding is what separates its recommendations from generic AI output: the more Profound knows about your business, the more relevant and accurate Aim's suggestions, plans, and content become.

> [!TIP]
>
> Set up and review your brand context in [Context Manager](/context-manager/about).

### The more knowledge, the better the results

A well-stocked [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits#how-knowledge-bases-work) lets Aim connect changes in your data to what's true about your business and recommend actions that fit your strategy.

Essentials to upload to your Knowledge Base:

* **Brand and positioning documents**: messaging guides, tone of voice, style guides, and personas, so content sounds like you.
* **Product documentation**: fact sheets, support docs, and help articles. These also serve as the source of truth for FactCheck accuracy analysis.
* **Strategy and planning materials**: content calendars, campaign briefs, and target audience research, so Aim prioritizes work that matches your goals.
* **Anything you'd hand a new marketing hire**: if a teammate would need it for context, Aim does too.

Uploads inform Aim's suggestions and recommendations as soon as they're processed. You can add documents in the platform or push them directly from your LLM tool through the [Profound MCP](/mcp/overview) connector.

## Tips for working with AI Marketer

### Keep your Knowledge Base current

Outdated documents produce outdated recommendations. Set up auto-sync for your Knowledge Base documents so that changes to your positioning, pricing, or product facts flow in automatically.

### Start with a conversation

The fastest way to see Aim at work is to chat with it. If you're not sure what to build or what it can do, go into the platform and ask it.

### Bring AI Marketer into Slack

Once you connect your Slack workspace to Profound, you can ask Aim questions and follow its work in real time without leaving the thread.

> [!TIP]
>
> To learn how to set up a Slack connector, see [Connect Slack to Profound](/integrations/slack/connect-slack-to-profound).


# Writing prompts for AI Marketer (Aim): best practices and examples
Source: https://docs.tryprofound.com/aim/writing-prompts

How to tell AI Marketer (Aim) what to look at, what to measure, and how to structure the answer

## Building blocks of a good prompt

Tell [AI Marketer (Aim)](/aim/about) what to look at, what to measure, how deep to go, and how to structure the answer. A good prompt covers:

1. **Input** — The thing to act on: a brand, a URL, a competitor, a list of topics. Use `[brand]`, `[competitor]`, and `[topic]` as placeholders so a prompt works for any team.
2. **Data source** — Where to look and how deep: "top 20 cited pages from the last 30 days", "owned URLs cited on ChatGPT this month".
3. **Task** — The specific analysis or decision you want made, not just a summary: "classify each source as press, affiliate, or UGC".
4. **Purpose** — The business goal behind the task. It helps Aim prioritize useful findings over noise: "to find partnership opportunities we're missing".
5. **Output format** — The deliverable structure: a table with named columns, a ranked list, flagged items. This helps make answers readable and useful.

For example:

> **(Input)** Create an exec-ready gap analysis for `[brand]` vs `[competitor]` **(Source)** across all tracked topics from the past 30 days. **(Task)** Compare visibility share and citation share by topic, then inspect the top citation pages behind the biggest gaps to identify why `[competitor]` is winning, **(Purpose)** so we know where to focus content efforts. **(Output)** Output a table with: topic, owned and competitor visibility share, owned and competitor citation share. Add a second table with: top citation page, likely driver, confidence, recommended action. Under the tables, add concise bullets on the biggest gaps and priority next steps.

## Prompt best practices

1. **One prompt = one job.** Keep the prompt focused to a single task. Prompts with multiple tasks like "Review the top 20 cited pages, and then also check whether any affiliate pages have outdated pricing claims" won't return a good result for either.
2. **Give detailed instructions**. Explain a task to Aim the way you'd explain it to a less experienced colleague. Vague instructions like "Analyze my data" gives Aim nothing to work with.
3. **Start with a verb**: go through, scan, review, identify, compare, flag, output.
4. **Be specific about scope**: top 20 pages, last 30 days, top three competitors.
5. **Define the output format**: a table, a ranked list, flagged items with columns. Otherwise results become unreadable and inconsistent from run to run.
6. **Include a confidence or rationale column**. It makes the output actionable.
7. **Chain the logic**: first do X, then for each result do Y, then output Z.

### Scale a prompt with Sheets

When you want the same analysis across many topics, URLs, or competitors, run it in a [Sheet](/sheets/getting-started). Put one input value per row in column A, point the prompt at that column, and Aim runs the prompt for each row in parallel. This works well for topic reports (a gap analysis across 80 topics), content audits (an accuracy check across 200 URLs), and user-generated content sweeps (a Reddit scan across 100 topics).

> [!TIP]
>
> Learn more in [Get started with Sheets](/sheets/getting-started)​.​

## Example prompts

Below is the catalog of reusable prompt examples. Copy the prompt text, fill in the placeholders and adjust the scope to your needs.

> **See where competitors are beating you in AI answers.**
>
> ```txt wrap theme={null}
> Compare [brand] and [competitor] across all tracked topics from the past 30 days on visibility share and citation share. For the topics with the biggest gaps, inspect the top citation pages and explain why [competitor] is winning, so we can prioritize content work. Output a table with: topic, owned visibility share, competitor visibility share, owned citation share, competitor citation share, likely driver, confidence, recommended action. Finish with bullets on the biggest gaps and priority next steps.
> ```

> **Discover which content types earn your citations.**
>
> ```txt wrap theme={null}
> Review the top cited owned pages for [brand] from the past 30 days. Classify each page as blog, product page, FAQ, landing page, comparison, docs, guide, case study, glossary, or other, so we can see which content formats are working. Output an exec summary, key findings, and a table with: URL, content type, citation count, evidence, confidence, recommended action.
> ```

> **Find pages where AI cites you but doesn't name you.**
>
> ```txt wrap theme={null}
> Go through the owned URLs for [brand] that appeared in AI answers this month. Check whether [brand] appears prominently in the title, H1, intro, first paragraph, or key summary sections, so we can fix pages that get cited without clearly saying our name. Output an exec summary, key findings, and a table with: URL, prominence status, evidence, confidence, recommended fix.
> ```

> **Diagnose which owned pages are losing citation share, and to whom.**
>
> ```txt wrap theme={null}
> Scan the top 20 owned cited pages for [brand] from the past 60 days across ChatGPT, Google AI Overviews, Gemini, and Google AI Mode. Identify pages with a month-over-month drop in citation share. For each drop, check which engines and prompts are affected and which replacement sources gained visibility, so we can diagnose lost AI visibility. Output an exec summary, key findings, and a table with: URL, citation share, MoM delta, replacement source, likely cause, confidence, recommended action.
> ```

> **Catch AI misrepresentations of your owned pages.**
>
> ```txt wrap theme={null}
> Take the top 20 cited unique owned pages for [brand] from the past 30 days and scrape each page. Compare the AI responses that cite them against the page content and flag meaningful omissions or contradictions, so we catch brand accuracy issues before they spread. Output an exec summary, key findings, and a table (max 20 priority items) with: URL, engine, prompt, AI claim, page claim, mismatch type, severity, recommended action.
> ```

> **Classify every earned source driving your AI visibility.**
>
> ```txt wrap theme={null}
> Review the top 20 cited pages for [brand] from the past 30 days. Classify each source as press, analyst, review site, affiliate, social/UGC, owned, or other, so we understand which earned channels drive visibility. Output an exec summary, key findings, and a table with: URL, source category, brand sentiment, evidence, confidence, recommended action.
> ```

> **Catch outdated affiliate content that can misrepresent your brand.**
>
> ```txt wrap theme={null}
> Using the [brand] Knowledge Base as the source of truth, go through affiliate and partner pages cited in the past 60 days. Flag outdated or inaccurate claims about positioning, features, pricing, audience, screenshots, or competitors, so partner content doesn't misrepresent the brand in AI answers. Output an exec summary, key findings, and a table with: URL, partner, issue, evidence, severity, confidence, recommended fix.
> ```

> **Monitor reputational risk in AI-cited community content.**
>
> ```txt wrap theme={null}
> Scan social-origin citations for [brand] from the past 30 days and scrape the body text of each page. Flag UGC that mentions [brand] negatively, inaccurately, incompletely, or with reputational risk, so we can monitor brand risk in AI-cited community content. Output an exec summary, key findings, and a table with: URL, UGC type, mention type, risk, evidence, severity, confidence, recommended action.
> ```

> **Find Reddit threads worth engaging.**
>
> ```txt wrap theme={null}
> Review the top cited Reddit threads for [brand] on [topic] from the past 60 days. For each thread, check recency, [brand] mentions, competitor mentions, sentiment, and whether [brand] has a credible opportunity to enter the conversation, so we know which threads to monitor, engage, or support with owned content. Output an exec summary, key findings, and a table with: prompt, subreddit, URL, brand mentioned, competitors mentioned, sentiment, user need, opportunity level, evidence, recommended action.
> ```


# About Answer Engine Insights (AEI)
Source: https://docs.tryprofound.com/answer-engine-insights/about

How Answer Engine Insights (AEI) shows your brand's performance across AI answer engines

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Analytics > Answer Engine Insights

Answer Engine Insights is a collection of analytical tools that show how your brand performs across AI answer engines, such as ChatGPT, Perplexity, or Gemini. Profound queries these answer engines and builds a dataset from the responses that shows how your brand is represented.

Below are the core concepts that Answer Engine Insights is built on.

## Prompt-driven analysis

Prompts are the queries Profound sends to answer engines daily. Profound aggregates the responses into a dataset you can filter to measure your brand's performance in answer engine results.

For example, Profound sends a prompt such as "What is the best tool for answer engine analysis" to each answer engine, then captures every response as a data point. Together, these data points show how your brand is represented across all responses.

### Topics

A topic is a group of related prompts organized around a common theme, product, or business objective. Topics structure your full collection of prompts.

### Tags

Tags add another layer of classification to prompts. Use them to group or filter prompts in Answer Engine Insights.

> [!TIP]
>
> Learn more in [Create, manage and tag prompts](/answer-engine-insights/prompt-designer).

## Visibility score

Profound tracks whether your brand appears in an answer engine response. If your brand is present, your Visibility score increases for that response; if it's not, the score decreases.

The Visibility score is a percentage: the number of responses that include your brand divided by the total number of responses that include at least one brand. For example, if your brand appears in 5 out of 10 responses, your Visibility score is 50% (5/10).

## Citations

A citation is a reference or link to a specific webpage, article, or resource within an answer engine response. Profound tracks how often answer engines cite your brand, which shows how they use your site's content.

## Sentiment

[Sentiment](/answer-engine-insights/sentiment/about#extracting-themes-from-ai-responses) is Profound's evaluation of how your brand is portrayed in AI-generated responses. For example, a response mentioning "high cost" lowers the Sentiment score because it reflects a negative association with your brand. A response stating "proven results" raises the score because it indicates a positive association.

## FactCheck

FactCheck is Profound's accuracy layer. It measures how AI engines describe your brand and pinpoints which sources are responsible when they get it wrong. FactCheck extracts the specific claims AI models make about your brand, checks each one against your own source of truth, and shows which citation URLs fuel any inaccurate information.

FactCheck has its own set of metrics and concepts, such as Accuracy, Claims, and Themes. Learn more in [About FactCheck](/answer-engine-insights/factcheck/about).

## Share of Voice

Share of Voice measures how often answer engines mention your brand relative to competitors, showing how prominent your brand is against other brands.

Share of Voice is a percentage: the number of responses that mention your brand divided by the total number of brand mentions across all responses. For example, if your brand is mentioned 20 times out of 100 total brand mentions, your Share of Voice is 20% (20/100).

## Average Position

Average Position ranks where your brand is mentioned in the AI answers against competitors in your market. It aggregates performance indicators, such as Share of Voice, into a ranked score. A score of 1 means that your brand is typically the first one to be mentioned in an AI response.

## Charts

Charts are the graphs, tables, and visualizations that help you interpret how your brand performs in answer engine results.

Each chart has its own configuration options depending on what it displays. To learn how each chart works, see [Interpret Answer Engine Insights](/answer-engine-insights/interpret).​


# Chart annotations in AEI
Source: https://docs.tryprofound.com/answer-engine-insights/chart-annotations

Mark events on Answer Engine Insights and Shopping charts to explain changes in your metrics over time

> [!NOTE]
>
> **Available on:** the Enterprise plan
>
> **Find it in:** Analytics > Answer Engine Insights, Analytics > Shopping
>
> **Required role:** all roles can view, Admin and Editor can create and manage. Seeing who made a prompt update requires the View Activity Logs permission.

Chart Annotations are notes pinned to a date on a chart. Use them to record events, such as a product launch, a content refresh, or a change to your prompts, so anyone looking at a trend can see the context behind it.

## View annotations

Annotations appear on all time-series line charts in [Answer Engine Insights](/answer-engine-insights/about) and on the line charts in [Shopping](/shopping/about). AI Marketer (Aim) can also show annotations on the line charts it generates.

1. Find the annotation markers along the bottom of a line chart.
2. Select a marker to see every annotation for that date, newest first. When dates are close together, their markers merge into one.

To hide annotations, open the chart’s settings (the sliders icon) and turn off **Annotations**.

![Line chart with annotation markers along the bottom.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=e31d9ae4da273d25be9f857d1621ddba)

## Types of annotations

### User annotations

Created by people in your organization. Shows anything your team wants to record.

![User annotation details shown in an annotation panel.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=eca27e4e7b867786ddb32f534d4e23a5)

### Platform updates

Created automatically by Profound for a model change on an AI platform that it tracks, such as a new ChatGPT model. Platform updates are hidden when that platform isn’t selected in your model filter.

![Platform update annotation details shown in an annotation panel.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-3.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=95570e8d767b1cd1559aac27fd021f0a)

### Prompt updates

Created automatically by Profound for a daily summary of prompt changes in a category, for example “12 prompts added, 3 disabled.” If you have the View Activity Logs permission, prompt updates also list the people who made the changes, and you can select **View activity logs** to see each change in Activity Logs.

![Prompt update annotation details with a View activity logs option.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-4.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=2cdd67e16b98dd531a13c5206225bc4e)

## Add an annotation

Add an annotation in any of the following ways:

* Select **Add annotation** at the top right of a chart. The date defaults to the end of your selected date range.

  ![Add annotation button at the top right of a chart.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-5.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=6bd8de2c0ce7f1aebc0748809389ef15)

* Select a single date on a chart to pin it, then select **Annotate** in the tooltip.

  ![Annotate option shown after selecting a date on a chart.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-6.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=c59f6916996a8bec3cb5dbf5989dc542)

* In an open annotation list, select **Add annotation** next to a date.

  ![Add annotation option shown next to a date in an annotation list,](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-7.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=351ea08437fbb2d2ae544eb6704ff81e)

Fill in the **Add chart annotation** form:

![Add chart annotation form with date, description, and category fields.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-8.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=b42d5eebf310f052a01129a8c1e33eaf)

* **Date**: any date within the chart’s selected date range. To annotate a date outside the range, widen the date range first with the top-level filters.

* **Description**: what happened.

* **Categories**: where the annotation shows. By default it’s **All categories**, which applies to your whole organization, including categories you add later. To narrow it, select specific categories. The category you’re viewing is always included.

Select **Save**. Everyone in your organization who can view annotations sees it.

## Edit or delete an annotation

You can edit or delete only the annotations you added.

1. Open the annotation list and select **…** next to your annotation.
2. Select **Edit** to change the annotation, or select **Delete**. Deleting an annotation is permanent.

Platform updates and prompt updates can’t be edited or deleted.

![Menu for editing or deleting a user-created annotation.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-9.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=96f3d121ad8d9ca37f46713581618b21)

## Investigate an annotation with Aim

If you have access to Aim, select **Investigate with Aim** next to any annotation. Aim opens with the annotation and the chart you were viewing, and starts looking into how the event relates to that metric.

![Investigate with Aim option shown next to an annotation.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/chart-annotations-10.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=fb0aca6701a25806a353b38b1c1190ca)

Aim can also read your annotations when you ask about your Answer Engine Insights data.

In Shopping, you can also ask Aim to explain a rise or drop in a metric, such as the Shopping Visibility Score, merchant score, or product visibility score. Select **Investigate** at the top right of a chart, or select a date range on the chart and then select **Investigate with Aim** in the tooltip.

## Permissions

> [!TIP]
>
> Learn more in the [Roles and permissions](/platform-config/people/roles-and-permissions) documentation.

| Role | View annotations and prompt updates | Add, edit, and delete their own annotations | See who made prompt updates |
| - | - | - | - |
| Admin | ✅ | ✅ | ✅ |
| Editor | ✅ | ✅ | ❌ |
| Member | ✅ | ❌ | ❌ |
| Viewer - AEI Only | ✅ | ❌ | ❌ |
| Viewer - AEI + AI Marketer | ✅ | ❌ | ❌ |
| Agent Operator | ✅ | ❌ | ❌ |

Prompt updates come from [Activity Logs](/platform-config/activity-logs/about), which are available on the Enterprise plan. Everyone who can view annotations sees them; seeing who made the changes and opening Activity Logs needs the **View Activity Logs** permission. Admins on the Enterprise plan can change these defaults with a custom role using the **View chart annotations** and **Create, edit, and delete custom chart annotations** permissions.

Annotations belong to your organization, and category-scoped annotations only show when viewing that category.


# Export prompts and topics
Source: https://docs.tryprofound.com/answer-engine-insights/export-prompts-and-topics

Download or copy prompts and topics from Answer Engine Insights

You can export prompts and topics in two places: [**Answer Engine Insights > Prompts**](/answer-engine-insights/interpret#prompts) tab, and in [Prompt Designer](/answer-engine-insights/prompt-designer). Exports support downloading files or copying data in JSON and CSV formats.

## Export in Prompt Designer

### Export prompts

In Prompt Designer table, select the prompts you want to export. Then select **Export prompts** at the top of the table.

![Prompt Designer Export prompts button highlighted with CSV and JSON export options](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/export-prompts-and-topics-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=8474cf985430043610b4d6d384bd7075)

### Export topics

Prompt Designer doesn’t support topic exports, but it does support exporting all prompts from a topic:

* Select a topic in the topics panel, this selects all prompts in that topic.
* Select **Export prompts** at the top-right of the table.

![Prompt Designer with a topic selected and Export 92 prompts button highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/export-prompts-and-topics-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=e04f3526c036e7603e2c776c4550a9cd)

## Export in Answer Engine Insights

### Export prompts

In **Answer Engine Insights > Prompts**, select the download icon next to the search field. To export the full prompt list, select a time period and make sure that no **Group By** filter is applied.

![Prompts tab with date range, Group by menu, and download icon highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/export-prompts-and-topics-3.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=72f8a2ef17384d2043e006c558119847)

Prompt exports include tracked prompts and their key performance indicators.

Prompt export fields include:

* Prompt
* Topic
* Tags
* Visibility
* Average Position
* Citation Share
* Citation Ranks
* Executions

### Export topics

In **Answer Engine Insights > Prompts**, apply the **Group by: Topic** filter to the prompts table. Then, copy or download the CSV or JSON the same way.

![Prompts tab with settings icon and Group by Topic option highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/export-prompts-and-topics-4.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=55eee746c9c6cca5ccd198ff1af678d5)

> [!TIP]
>
> Select **Open in Sheets** to feed your exported prompts directly to your [Workflows](/agents/workflows/about-workflows). Learn more in [Get started with Sheets](/sheets/getting-started).


# About FactCheck
Source: https://docs.tryprofound.com/answer-engine-insights/factcheck/about

How FactCheck catches what AI engines get wrong about your brand

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Analytics > Answer Engine Insights > FactCheck tab

FactCheck is Profound's accuracy layer. It measures how AI engines describe your brand and pinpoints which sources are responsible when they get it wrong. It extracts the specific claims AI models make about your brand, checks each one against your own source of truth, and shows you which citation URLs fuel any inaccurate information. The result is a clear, actionable picture of where AI-generated misinformation comes from, so your content team knows exactly what to fix.

[Profound Factcheck walkthrough](https://player.mux.com/x34671MQlYaTJ02yS5MYjTI2N7H1hi86EvfbSqBLaL7U?metadata-video-title=Profound+Factcheck+walkthrough\&video-title=Profound+Factcheck+walkthrough)

## When AI gets your brand wrong

AI search engines like ChatGPT, Perplexity, and Gemini don't just surface links, they make direct claims about your brand. They describe your product features, pricing, policies, and service terms as facts. When those claims are wrong, your customers are misled before they ever reach your website. And in many cases, inaccuracies don't come from AI fabricating information. They come from a brand's own published content being out of date.

The problem is that monitoring this at scale is nearly impossible without specialized tooling. Brands may know anecdotally that AI sometimes gets things wrong, but they have no systematic way to measure how often it happens, which prompts are affected, or which sources are responsible. Without that information, there's no clear path to fixing it. Content teams can't update a page they don't know is the problem, and brand teams can't report on accuracy without data to back it up.

For brands where facts matter most, such as those with regulated products, complex pricing, or frequently updated policies, this gap is especially costly.

## How FactCheck works

> [!TIP]
>
> FactCheck requires a Knowledge Base to verify claims. Learn more in [Knowledge Bases and Brand Kits](/context-manager/knowledge-bases-and-brand-kits).

FactCheck brings together three things that were previously separate: claim extraction, source-of-truth verification, and citation attribution.

Profound pulls the claims AI engines make about your brand across all your FactCheck-tagged prompts, compares each claim against your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits), and classifies it as accurate, inaccurate, or not relevant. Only factual, actionable, and atomic claims are verified: each claim must assert a single thing that your source of truth can confirm or contradict. Subjective statements are routed to [Sentiment](/answer-engine-insights/sentiment/about) instead.

For every inaccurate claim, FactCheck traces the finding back to the specific citation URLs in the AI response that appear to drive it. The citation URL is what makes the data actionable: instead of knowing only that AI is saying something wrong, you know which page or document in your Knowledge Base is the likely source of the problem.

Once you've identified the inaccuracies, run a [Profound Agent](/agents/about) directly from the FactCheck view to draft the outreach, update your Knowledge Base, create content, or otherwise help answer engines discover the correct information about your brand.

## Key features

### Claim extraction and verification

FactCheck automatically extracts the specific claims AI engines make about your brand in response to your FactCheck-tagged prompts. Claims are filtered to those that are factual, actionable, and atomic, each asserting one verifiable thing. Subjective claims are routed to Sentiment instead.

Each claim is then checked against your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) and classified as accurate, inaccurate, or not relevant. This gives you a structured, consistent view of where AI gets things right and where it doesn't, with no manual review of individual responses.

### Claim clustering by theme

Rather than presenting thousands of individual claims, FactCheck groups similar claims into themes. This makes the data much more digestible and helps you understand which topic areas have the most accuracy problems. Instead of sifting through raw claim lists, you see at a glance that, for example, claims about your cancellation policy are a bigger problem than claims about your pricing.

### Citation-level accuracy breakdown

For every inaccurate claim, FactCheck shows you which citation URLs in the AI response are associated with that inaccuracy. This is the feature that turns FactCheck data into a direct action item for content teams. You know which page to update, not just that something is wrong.

### Accuracy overview and metrics

The **Accuracy Overview** section shows what changed from last week to the current week (the time period is customizable): Accuracy Score and Share, top inaccurate claims, movement across themes, and a comparison across reporting periods. Click into any inaccurate claim to see why it's wrong, the ground truth from your Knowledge Base, and every location in the [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) where the correct information is defined.

### User feedback mechanism

If a claim is marked inaccurate but is actually correct, report it directly in the FactCheck UI, and the claim is removed from your view. If a claim isn't something you want to track, mark it as not relevant. Profound uses this feedback to keep improving claim verification.

### Run Agents directly from claim views

Take action directly from any inaccurate claim view: run an [Agent](/agents/about) to draft outreach to the citation source, update your own content, or create new content to address the inaccuracy.


# FactCheck FAQ
Source: https://docs.tryprofound.com/answer-engine-insights/factcheck/faq

Frequently asked questions about FactCheck in Answer Engine Insights

> **What’s the difference between FactCheck and Sentiment?**
>
> FactCheck evaluates the factual accuracy of what's said about your brand, while [Sentiment](/answer-engine-insights/sentiment/about) measures the tone and perception of how your brand is described. The two features are complementary. Use them together for the best results.

> **How often does the FactCheck run?**
>
> FactCheck runs daily across all your FactCheck prompts. When set up for the first time, it takes 7 days for FactCheck results to appear in Profound.

> **Is FactCheck automatically enabled for every new workspace?**
>
> No. FactCheck is set up separately in each workspace. Learn more about how to set it up in [Get started with FactCheck](/answer-engine-insights/factcheck/getting-started).

> **Will FactCheck automatically show older historical data if I set it up later?**
>
> No. FactCheck starts producing results after setup is complete, and first-time results typically take 7 days to appear.

> **What is the maximum number of FactCheck prompts I can configure?**
>
> The limit is 200 prompts. If you need more, reach out to your Profound team.

> **What types of content can I use as a source of truth?**
>
> Support documentation and help center pages work well because they're factual, structured, and ideally regularly maintained. Internal product specs, policy documents, and Q\&A documents also work well. Learn more in [How to get the best results working with FactCheck](/answer-engine-insights/how-tos/factcheck-best-results).

> **What happens if there is nothing in the Knowledge Base?**
>
> FactCheck requires a [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) to function. Without one, there's no source of truth to verify claims against, and the feature won't produce results. Even a simple Q\&A document covering your brand's most important facts is enough to get started.

> **Can I track how often AI responses align with our official brand facts?**
>
> Yes. FactCheck surfaces a representation accuracy metric: the percentage of times AI engines state correct information about your brand across your FactCheck prompts. Track this over time to measure progress as you update citation sources, expand your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits), or add new prompts.

> **Does FactCheck work only for brand-specific facts, or can it verify other types of claims?**
>
> FactCheck is designed for brand-specific, fact-based claims: product specifications, pricing, policies, safety information, features, and service terms. It isn't designed to verify general legal or medical information that isn't specific to your brand, and it doesn't verify visual content such as logos or images in AI-generated media.

> **Why can some claims not be verified?**
>
> FactCheck verifies only claims it can confidently check against your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits). If the relevant information isn't in the Knowledge Base, the claim is left unverified. Expanding and updating your Knowledge Base directly increases verification coverage. Learn more in [How to get the best results working with FactCheck](/answer-engine-insights/how-tos/factcheck-best-results).

> **Can FactCheck help us get ahead of misinformation before a product launch?**
>
> Yes. FactCheck works best when used proactively. Before a launch, upload updated product documentation to your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) and tag prompts that cover the facts most likely to be misrepresented. Once FactCheck identifies inaccurate citations, your content team can contact those sources or submit corrections before the misinformation spreads. The citation-level breakdown shows which third-party pages to prioritize.

> **Can FactCheck help recover brand reputation after a negative event?**
>
> FactCheck gives you visibility into which AI responses state incorrect or outdated information about your brand and which sources are responsible. After a negative event, use this data to identify the specific pages driving harmful claims, take action to update or correct those sources, and track whether representation accuracy improves over time. Used alongside [Sentiment](/answer-engine-insights/sentiment/about), it gives you a complete picture of both what AI says and whether it's factually correct.


# Get started with FactCheck
Source: https://docs.tryprofound.com/answer-engine-insights/factcheck/getting-started

Set up FactCheck by assigning prompts and connecting a Knowledge Base

## Step 1: Open the FactCheck tab

Go to [**Answer Engine Insights**](/answer-engine-insights/about) **> FactCheck**. The first time you open the tab, an onboarding screen guides you through setup: assigning the FactCheck analysis type to your prompts and connecting a [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits).

## Step 2: Choose your prompts

1. From the onboarding screen, open the [Prompt Designer.](/answer-engine-insights/prompt-designer)

   ![Set up FactCheck screen with the Open Prompt Designer button highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/factcheck/getting-started-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=affdd859cf37d906bed4b88ef3f6bd86)

2. Select 100 to 200 fact-based prompts that are likely to surface verifiable claims about your brand. Good candidates include prompts about product specifications, pricing, policies, safety claims, features, and service terms. Avoid opinion-based or subjective prompts.

   > [!TIP]
   >
   > To learn more about FactCheck prompt best practices, see [How to get the best results working with FactCheck](/answer-engine-insights/how-tos/factcheck-best-results).

3. After you select the prompts, go to **Edit > Analysis Types** and set their analysis type to **FactCheck**.

4. If you're already tracking prompts for Visibility or Sentiment, you can add FactCheck to those or create new prompts specifically for FactCheck.

## Step 3: Connect a Knowledge Base

On the onboarding screen, select the **+ Connect Knowledge Base** and select the Knowledge Base you want FactCheck to use as its source of truth.

![Set up FactCheck screen with the Connect Knowledge Base button highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/factcheck/getting-started-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=d821988b66e9706cd9236c6e46c8361c)

FactCheck uses the entire Knowledge Base, so there's no need for a special file structure. Organize your documents however works best.

If you don't yet have a Knowledge Base you want to set up with FactCheck, go to **Knowledge Bases** in the left navigation menu, create a Knowledge Base, and upload your documentation as described in [Create and manage Knowledge Bases](/context-manager/manage-knowledge-bases).

> [!NOTE]
>
> You can connect only one Knowledge Base to FactCheck. Learn more about Knowledge Base best practices and setup limitations in [How to get the best results working with FactCheck](/answer-engine-insights/how-tos/factcheck-best-results).

## Step 4: Complete setup

Once your prompts carry the FactCheck analysis type and a Knowledge Base is connected, select **Complete setup**.

## Step 5: Explore the results

FactCheck results typically take 7 days to appear. Once available, go to **Answer Engine Insights** > **FactCheck** to review them:

* Browse **Inaccurate Claims** by theme and expand themes to view the claims and the ground truths they were evaluated against.
* Review the **Accuracy Breakdown** per citation, platform or prompt. Select citations or prompts to open a detailed analysis view.

## Step 6: Take action

Review the **Inaccuracy Drivers** table. It lists the pages that drive the most inaccurate claims about your brand. Take action directly from the table: select **Draft Outreach** in the **Suggested Action** column and have [AI Marketer](/aim/about) build an [Agent](/agents/about) to draft outreach for that source.


# How to create effective prompts
Source: https://docs.tryprofound.com/answer-engine-insights/how-tos/create-effective-prompts

Use Prompt Volumes and Intent data to write prompts that match how real users talk to AI tools

Creating effective prompts ensures your brand aligns with how real users engage with AI tools. Profound’s Prompt Volumes and Intent data help confirm that your topics and phrasing match the language people use when talking to LLMs.

> [!TIP]
>
> Learn more about creating and validating prompts with Prompt Volumes data in the [Validating Your Tracking with Volume Data](https://university.tryprofound.com/courses/profound-101/modules/setup/lessons/validating-your-tracking-with-volume-data) lesson, part of Profound University's [Profound 101](https://university.tryprofound.com/courses/profound-101) course.

## Map Prompt Volumes (demand signal)

[Prompt Volumes](/prompt-volumes/about) reveals what people are already asking AI about your product or category. Use these volumes to validate that your topics and phrasing align with real audience demand.

To map Prompt Volumes:

1. Open **Prompt Volumes** in Profound.
2. Enter a seed term (for example, "corporate card rewards").
3. Review related prompts: check volume, region, and LLM breakdown.
4. Identify variations that map best to your audience's natural phrasing.

> [!TIP]
>
> Start with your existing SEO key phrases, then rephrase them into natural, question-style prompts that reflect how people actually ask in ChatGPT, Claude, or Perplexity. For example, turn “best corporate card small business” into **“What’s the best corporate credit card for small teams?”**

## Balance by intent (purpose of query)

Every prompt has an underlying intent, whether users are learning, comparing, or acting. Use the **Intent** tab in **Prompt Research Report** view (opens when you select a prompt in Prompt Volumes) to balance your set across the full funnel.

| **Intent Type** | **Example Prompt** | **Best For** |
| - | - | - |
| **Informational** | “What is Answer Engine Optimization?” | Awareness and education |
| **Commercial** | “Best AEO platforms for enterprise marketing” | Mid-funnel comparison |
| **Generative / how-to** | “Create a blog outline optimized for AI search” | Task execution |

> [!TIP]
>
> Maintain a healthy mix of intents to capture demand across every stage of discovery.

## Anchor to customer problem areas

Pull from CX, Sales, and Support insights to find recurring questions or friction points, and translate those into prompt form using the same phrasing your customers use.

For example, if your support team often gets “I don’t understand how visibility scores work,” turn it into:

* “How do AI visibility scores work?”
* “How can I improve visibility score in ChatGPT answers?”

These authentic, problem-driven prompts perform well because they reflect real customer language and context.

## Combine and iterate

Once you've mapped volumes, intent, and customer needs:

* [Tag and cluster your prompts](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags) by topic or funnel stage.
* Prioritize those with high AI conversation volume and actionable intent.
* Revisit every month: user phrasing evolves as AI adoption grows.

Feed your refined prompts into [Profound Agents](/agents/about) or [Answer Engine Insights](/answer-engine-insights/about) to measure visibility gains over time. This creates a continuous feedback loop: Analyze → Act → Measure.


# How to create topics and prompts to track in Profound
Source: https://docs.tryprofound.com/answer-engine-insights/how-tos/create-topics-and-prompts

Build a structured prompt set by defining topics, generating prompts, and uploading them to Profound

This guide explains how to build a structured prompt set for tracking in the Profound platform. The process covers defining overarching topics, generating prompts, and uploading them into the platform.

> [!TIP]
>
> Learn more about creating topics and prompts in the [Setting Up Your Topics and Prompts](https://university.tryprofound.com/courses/profound-101/modules/setup/lessons/setting-up-your-topics-and-prompts) lesson, part of Profound University's [Profound 101](https://university.tryprofound.com/courses/profound-101) course.

## How to create topics

### Step 1: Define overarching topics

Topics are the high-level themes that shape your monitoring strategy. They should cover the range of areas relevant to your brand and industry.

Examples of overarching topics:

* **Product or service topics**: project management software, online grocery delivery, skincare products
* **Customer use cases or needs**: "tools to collaborate with remote teams," "where to order groceries in under an hour," "face cream for sensitive skin"
* **Industry themes or trends**: AI in workplace productivity, sustainable packaging, rise of quick commerce
* **Competitors or alternatives**: "Company vs Company," "Instacart alternatives," "Alternatives to Sephora"
* **Regions and languages**: think about where you want visibility. For example, prompts for the US market should be in English, while prompts for Spain should be written in Spanish. Consider differences in phrasing across regions.

Aim for 5–10 topics to start. Keep them broad enough that each one can branch into multiple prompts.

### Step 2: Generate prompt ideas

Once you've defined your topics, expand them into prompts. Use multiple sources to create a well-rounded set.

**Use the Profound prompt generator**

* Use Profound's generator or other LLMs to create 5–10 prompts for each topic.
* Include different types of prompts:
  * Informational: "What is project management software?"
  * Comparative: "Best project management software vs Trello"
  * Problem-solving: "How do I track deadlines in project management software?"
  * Transactional: "Where can I buy affordable project management tools?"
* Generate variations to reflect how users naturally ask similar questions.

**Internal FAQs and support data**

* Pull frequent customer questions from support or sales teams.
* Reframe them into natural prompts. Example: FAQ = "Does the tool integrate with Slack?" → Prompt = "Does project management software integrate with Slack?"

Each prompt runs daily by default, and Profound scores it across all selected platforms (ChatGPT, Gemini, Perplexity, and others) and regions.

### Step 3: Upload prompts into Profound

After generating and refining prompts, you can upload them into the platform in three ways:

**Generate prompts in the platform (best for scaling quickly)**

1. Go to **Answer Engine Insights** → **Prompts** → **Modify Prompts.**
2. From the **Generate** option, create new topics and prompts directly in Profound.
3. Select region and language (for example, US/English or Brazil/Portuguese).
4. Profound generates prompts for you.
5. Review and add them directly to your configuration for tracking.
6. Prompts created here are automatically associated with the selected topic and region, and begin running on the next daily visibility cycle.
7. Best for: exploring new topics or quickly expanding coverage.

**Bulk upload (best for large prompt sets)**

* Go to **Answer Engine Insights** → **Prompts** → **Modify Prompts**.
* Select **Bulk Upload** > **Download template** to download the prompts sheet template.
* Fill in the template with your prompts, regions, tags, and languages. Make sure each field (Topic, Prompt, Tags, Regions, Language) appears in a separate column. If all values appear in a single column when you open the file, use "Split text to columns" in Excel or Google Sheets with comma as the delimiter.
* Upload the completed file to add all prompts at once.
* Best for: adding many prompts across multiple topics and geographies.

> [!TIP]
>
> If you receive an error that "Topic and Prompt are missing," verify that your CSV has each field in its own column. Open the file in Excel or Google Sheets and confirm the headers (Topic, Prompt, Tags, Regions, Language) are separated across the top row, not combined in a single cell.

**Manual add/edit (best for smaller updates and refinements)**

* Go to **Modify Prompts**, where you already have topics and prompts set up.
* Use **Add Topic** or **Add Prompt** to write in new entries individually.
* Best for: testing a few new prompts, adding FAQs, or refining specific topics.

Visibility, Share of Voice, and Citations metrics populate after the first full 24–48 hour cycle; an initial "No Data" status is normal.

After uploading through any method, prompts run daily, with results visible within about 48 hours. Aim to create approximately 15 strategic topics, with around 10 prompts per topic. This balanced approach helps maintain quality while ensuring comprehensive coverage.

### Best practices

1. **Think in topics, not keywords**

Group related questions into broader, strategic topics rather than tracking individual keywords. Each topic should represent a clear business theme.

**Example**: instead of tracking variations of "cheap flights," create a topic called "Flight pricing and deals" with relevant prompts.

2. **Focus on user intent**

Write prompts that reflect natural conversation patterns. Use complete questions or statements rather than short phrases.

**Example**:

* Less effective: "Cheap flights Europe"
* Better: "What's the best way to find affordable flights to Europe?"

3. **Prioritize relevance over volume**

Choose fewer, high-impact prompts that align with your strategic goals rather than tracking thousands of low-value variations.

**Example**:

* Less effective: "Hotel ratings good sites"
* Better: "Which travel site has the most accurate hotel ratings?"

## Categorize Your Prompts With Tags

### How to tag prompts

Tagging is a powerful way to organize and filter prompts inside Profound. You can add tags in two places:

1. **Excel template (Bulk Upload)** → add tags in the “Tags” column before uploading.
2. **In the platform (Modify Prompts)** → edit an individual prompt or topic and assign tags directly.

### Why tagging matters

* **Organization**: Tags help you quickly filter large prompt sets into manageable views.
* **Analysis by theme**: See how you perform in a specific area (e.g., competitor comparisons vs. product features).
* **Cross-topic tracking**: Some prompts may overlap topics; tags let you track them across multiple dimensions.
* **Team collaboration**: Tags make it easier for different teams (e.g., product, PR, marketing) to focus on the prompts that matter to them.

### Tagging guidelines

* **Use clear, descriptive labels**
  * Good: “Competitor Comparison,” “Feature Request,” “Purchase Intent”
  * Not good: “Misc,” “Random,” “Test”

* **A single prompt can have multiple tags**

  Prompts often belong in more than one topic. Assigning multiple tags to one prompt lets you analyze prompts from various angles without duplicating them.

  * In the Excel template: put multiple tags in the same cell, separated by commas. Example: Competitor Comparison, Purchase Intent, Project Management
  * In the Profound platform: assign multiple tags directly to the same prompt when editing.

> [!TIP]
>
> Tag prompts by topic (e.g., “Competitor comparisons,” “Use cases,” “Trends”) so you can group and analyze results easily later.


# How to get the best results working with FactCheck
Source: https://docs.tryprofound.com/answer-engine-insights/how-tos/factcheck-best-results

Best practices for Knowledge Bases, prompts, and review when working with FactCheck

## What makes good FactCheck prompts

FactCheck prompts should have a realistic chance of triggering an LLM response that mentions your brand and makes a verifiable, objective claim: a price, a feature, a spec, or a policy. Vague and subjective prompts ("Is Brand X good?") belong in [Sentiment](/answer-engine-insights/sentiment/about). Comparisons ("Brand X vs Y") can work as long as they include verifiable claims.

### Branded vs unbranded prompts

> [!NOTE]
>
> **Branded prompts** — prompts that mention a brand directly. For example: "Does Profound track brand mentions across AI search engines?"
>
> **Unbranded prompts** — prompts that ask broader category questions and may surface a brand naturally. For example: "What are the best tools for tracking brand visibility in AI search and why?"

Branded prompts are recommended but not required. They make it easier to filter for relevant responses and reduce noise. Unbranded prompts are fine when the category is specific enough that the brand is likely to surface anyway (for example, "What is the best AEO tool?").

### Examples

|  **Good: factual, specific** |  **Bad: vague, subjective** |
| - | - |
| Can Profound track brand mentions across ChatGPT and Perplexity? | Is Profound a good company? |
| How much does Profound's Pro plan cost? | Tell me about Profound. |
| What is Profound's refund policy? | What do people think of Profound's customer service? |
| Is Profound available in Australia? | What is the best CRM tool? |

## What makes a good Knowledge Base for FactCheck

### Content best practices

* Make sure **the information in your Knowledge Base answers the prompts you're tracking** with FactCheck. For example, if you're tracking prompts about pricing, include a pricing page or FAQ that covers it.

* Make sure the information in your Knowledge Base **covers all the content** relevant to the prompts you're tracking. If coverage is partial, results degrade in two ways:
  * Claims about topics your Knowledge Base doesn't cover return as "not relevant" because FactCheck has nothing to verify them against.
  * Claims about topics it covers only partially may be misclassified. For example, a claim about an old pricing tier might be marked accurate if the Knowledge Base doesn't include the updated pricing page.

* **Add your public-facing website** to your Knowledge Base. Knowledge Bases support recrawling on a daily, weekly, or monthly cadence, keeping your source of truth current automatically. Learn more in the [Create and manage Knowledge Bases](/context-manager/manage-knowledge-bases) section.

* **Use live, publicly accessible pages you own**, such as help center articles, policy pages, and product pages. Set up an [auto-sync schedule](/context-manager/knowledge-bases-and-brand-kits#content-auto-sync) to keep the content current.

### Limitations

* Each FactCheck configuration uses exactly one Knowledge Base, and you can't split verification across several.
* You can't change which Knowledge Base is connected to FactCheck after the initial setup. However, its contents stay fully editable: add, remove, or update files at any time, and the next time FactCheck verifies claims, it will check them against the updated content.


# Interpret Answer Engine Insights
Source: https://docs.tryprofound.com/answer-engine-insights/interpret

Read the Visibility, Sentiment, FactCheck, Citations, and Prompts tabs in AEI

> [!NOTE]
>
> This article describes the redesigned Answer Engine Insights. To switch to the original design, go to [Answer Engine Insights settings](/answer-engine-insights/settings) > **Use legacy UI.**

Answer Engine Insights (AEI) breaks down your brand's performance across several analytical dimensions. Each dimension has its own tab in the top navigation: Visibility, Sentiment, FactCheck, Citations, and Prompts.

[https://player.mux.com/pkFN9Dcdn02027vfw00tAp01oaaemR246A2tfteOSEIgxFk?metadata-video-title=Profound+Answer+Engine+Insights+Walkthrough\&video-title=Profound+Answer+Engine+Insights+Walkthrough](https://player.mux.com/pkFN9Dcdn02027vfw00tAp01oaaemR246A2tfteOSEIgxFk?metadata-video-title=Profound+Answer+Engine+Insights+Walkthrough\&video-title=Profound+Answer+Engine+Insights+Walkthrough)

## Filters

Filters are available throughout Answer Engine Insights. They allow you to adjust the analytical time frame, narrow the data view, and save views you return to often. Filters appear as dropdowns beneath the top navigation bar.

![Visibility tab with date range, comparison period, and Topics to Personas filters highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=e35b7ab73897e226005177f47225a0a4)

### Date range and comparison period

Use the date range filter to select the time window of the answer engine response collection. Choose a preset window (such as **Last 7 Days**) or define a custom range using the calendar selector.

After you select a time frame, only answer engine responses generated during that period appear in the charts.

Charts show change by comparing the selected time frame to a comparison period. Open the **Prev. Period** dropdown to choose how the comparison period is calculated:

* **Prev. Period**: the window immediately before the selected date range, of the same length.
* **Prev. Period Last Month**: the same window one month earlier.
* **Custom range**: any date range you define.

### Attribute filters

Attribute filters narrow the data shown on every chart in the current tab. To apply one, open the dropdown for the attribute (**Topics**, **Platforms**, **Tags**, **Regions**, **Competitors**, or **Languages**) and select the values to include. If your configuration uses [Personas](/answer-engine-insights/personas), a **Personas** filter is also available.

Use the **Competitors** filter on the Visibility tab to choose which of the brands you track appear in its charts and rank lists.

The Citations tab adds two filters of its own: **Prompt Type** and **Citation Categories**. Select **+** to add more filters to the bar.

### Save view

Select **Save view** to store the current combination of date range, comparison period, and filters. Saved views let you return to a specific analysis without rebuilding it each time.

### Export

Select **Export** in the top right of the tab to download the underlying data for the current filters. The Visibility, Sentiment, FactCheck, and Prompts tabs export answers. The Citations tab exports citations.

### Settings

Select the gear icon in the top right to open Answer Engine Insights settings. See the [Answer Engine Insights settings](/answer-engine-insights/settings) for details.

## The Compare dropdown

Most charts in Answer Engine Insights include a **Compare** dropdown in their top right corner.

![Visibility Score chart and rank list by platform with Compare Platforms dropdown highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=cc50775cf3da35ab45de015accec67b5)

Use it to change how the chart breaks down data. The options are:

* **Competitors**: your brand and the brands you track, one line or row per brand. This is the default for the main Visibility chart.
* **Platforms**: one line or row per answer engine, such as ChatGPT, Perplexity, or Google AI Overviews.
* **Topics**: one line or row per topic.
* **Tags**: one line or row per prompt tag.
* **Regions**: one line or row per region you track.
* **Personas**: one line or row per Persona, if your configuration uses [Personas](/answer-engine-insights/personas).

For example, to see how your Visibility Score differs by platform, open the Visibility tab and set the **Compare** dropdown on the Visibility Score chart to **Platforms**. The rank list beside the chart then ranks platforms instead of brands.

In the Sentiment and FactCheck tabs, the **Compare** dropdown has additional options: **Themes** and previous period comparison.

![Sentiment Summary chart with Compare dropdown open listing Prev. Period through Personas](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-3.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=af066ec2670b3ab2b575d4a6732d47fc)

Each chart has a chart settings icon next to the **Compare** dropdown. Use it to change how the chart visualizes data.

![Chart settings icon highlighted showing Bar Chart, Line Chart, Auto Y-Axis Range, and Table options](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-4.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=efc1ad2fb736014452948976701cde24)

## Visibility

The **Visibility** tab breaks down your brand's presence in answer engine responses.

### Visibility Score, Share of Voice, and Average Position

The top chart has three sub-tabs that switch the metric: **Visibility Score**, **Share of Voice**, and **Avg Position**.

![Visibility tab with Visibility Score, Share of Voice, and Avg Position metric tabs highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-5.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=a0582e6965d57d8938e28ed45dad2f12)

Each metric appears as a line chart and a rank list.

* The line chart shows the metric over the selected date range. Select **Current Period** or **Previous Period** at the bottom of the chart to show or hide either line, and toggle the **Show ...** option to overlay the brands you track.
* The rank list shows your brand's rank by the selected metric relative to competitors, with the change versus the comparison period. Select **Expand** to see the full list. When the chart compares topics or tags, select a row to open that topic or tag's subpage.

Use the **Compare** dropdown to pivot the chart and rank list by platforms, topics, tags, regions, or Personas.

### Visibility Trends

**Visibility Trends** shows the largest movements in Visibility Score during the selected period, split into **Biggest Spikes** and **Biggest Drops**. Each item shows a sparkline, the current score, and the change versus the comparison period.

![Visibility Trends panel listing Biggest Spikes and Biggest Drops by topic with sparklines](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-6.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=fb8078bd9f74dbca1ef22862416b091a)

By default the trends compare topics. Use the **Compare** dropdown to see the biggest movers across competitors, platforms, tags, regions, or Personas instead.

### Visibility Ranking

Visibility Ranking is a customizable heat map that shows how two dimensions intersect, for example how each platform performs in each topic. Configure it with the controls above the table:

* **Value**: show the Visibility Score as a percentage, or show rank instead.
* **Rows**: the dimension shown as rows, such as topics, tags, regions, or platforms.
* **Columns**: the dimension shown as columns, such as competitors or platforms.

Darker cells indicate higher values.

![Visibility Ranking heatmap with Matrix and Table toggle and Value, Rows, Columns dropdowns highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-7.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=59aa289c2cefd5caa3b7a8e745251a8f)

Select **Table** to view the same data in a standard rank table. Expand the topics to see the ranking of their prompts. Select a prompt to open it's Prompt Analysis view with visibility metrics, executions and citations.

![Visibility Ranking table view with AI Agents topic expanded to show prompts ranked by platform](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-8.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=74fb40ffe584c3af5d6bc5f2ff70dce3)

### Topic Prioritization

Topic Prioritization plots each topic by its Visibility Score and Citation Share so you can decide where to focus.

![Topic Prioritization scatter plot of citation share versus visibility with four quadrants](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-9.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=d1fa72ea965ab6926301e28b69e4859b)

The matrix splits topics into four groups:

| Group | What it means |
| - | - |
| Strongholds | High visibility and high citation share. Maintain your position. |
| Needs attention | Low visibility and low citation share. Content and citation work are both needed. |
| Cited, Not Visible | Your content is cited, but your brand isn't mentioned in the answer. Review how the cited pages present your brand. |
| Visible, Not Cited | Your brand is mentioned, but answer engines cite other sources. Build content that answer engines can cite. |

Select **Table** above the plot to view the same data as a ranked list.

## Citations

The **Citations** tab assesses how often answer engine responses cite your domains, and which other sources they rely on.

### Citation Share and Citation Coverage

The top chart has two sub-tabs that switch the metric: **Citation Share** and **Citation Coverage**.

![Citations tab with Citation Share and Citation Coverage tabs highlighted above chart and rank list](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-10.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=c48fa730fdfd8404e90a793063eb4e2a)

Each metric is displayed as a line chart and a rank list.

* The line chart shows the metric over the selected date range. Select **Current Period** or **Previous Period** at the bottom of the chart to show or hide either line, and toggle the **Show ...** option to overlay other domains.
* The **Citation Share Rank** list ranks the most cited domains, with your owned domains marked **Owned**. Select **Expand** below the table to see the full list.

By default the chart compares websites. Use the **Compare** dropdown to pivot by platforms, topics, tags, regions, or Personas.

### Citation Categories

Citation Categories classify the sources AI platforms cite, helping you understand where your visibility comes from and identify opportunities.

The **Citation Share by Category** chart shows the share of citations for each category. Use the chart settings on the top right of the chart to switch to a line chart and see how each category trends over time.

![Citation Categories bar with chart settings icon highlighted showing Bar Chart and Line Chart options](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-11.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=73ad3dda9fab5fbf3a242672c81bc9e7)

**Owned**, **Competition**, and **Custom** categories are defined in [Answer Engine Insights settings](/answer-engine-insights/settings). All other categories are assigned automatically but can be overridden in settings.

| **Category** | **Definition** | **Includes** |
| - | - | - |
| Owned | Websites and digital properties directly owned, controlled, or managed by your brand or its related entities. | Primary domains, subdomains, product microsites, help centers, documentation sites |
| Competition | Websites belonging to or controlled by competitors your brand tracks. | Competitor homepages, blogs, press sections, resource pages |
| Earned Media | Sites whose primary purpose is publishing content. A marketer can identify a site owner, editor, or publisher to partner with. Content is the product, monetized via ads, affiliate revenue, or subscriptions. Does not include corporate blogs, company newsrooms, or lead generation sites. | News outlets, magazines, trade publications, editorial review sites (Wirecutter, PCMag), affiliate and comparison sites (NerdWallet, Investopedia), broadcast media, contributor-model publications (Forbes, HuffPost), personal blogs |
| PR Wire | Press release distribution services that syndicate company announcements to media outlets. | PR Newswire, Business Wire, GlobeNewswire, Accesswire, and other major PR wires |
| Institution | Government, educational, research, or public-interest nonprofit organizations. Institutional TLDs are auto-classified: .gov, .edu, .mil, .int, .gov.*, .ac.*, .nhs.*, .gouv.*, .gob. | Government sites, universities, research institutions, academic publishers (Nature, arXiv), international bodies (UN, WHO), professional associations (ABA, AMA), mission-driven nonprofits (Red Cross, UNICEF), Wikipedia |
| Social | Platforms where users publish freely and visibility is determined by votes, algorithms, or follower graphs. Does not include vendor-owned support forums or e-commerce product reviews. | Social networks (Facebook, Instagram, X/Twitter, TikTok, LinkedIn), video platforms (YouTube), discussion platforms (Reddit, Discord), Q\&A platforms (Quora, Stack Overflow), open publishing platforms (Medium, Substack), consumer review aggregators (Yelp, TripAdvisor), B2B review aggregators (G2, Capterra, TrustRadius) |
| Other | Sources that don't fit the above categories, typically where content exists to support selling products or services. Domains with fewer than 100 total citations are also classified as Other. | Corporate websites, SaaS, e-commerce, service businesses, company blogs and newsrooms, vendor-owned forums, marketplaces (Airbnb, Etsy), course platforms (Coursera, Udemy), job boards, directories, industry trade groups, lead generation sites |
| Custom | Categories you create to track specific domains or pages. | Any domains or pages you specify |

### All Citations

All Citations is a ranked table of every cited source in the selected period.

Use the search box above the table to find a domain or page, the settings icon to adjust the table, and the download icon to export it.

![All Citations table with Domains, Pages, Watchlist tabs and search domains field highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-12.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=196262c754cb88bf1cb3fa2d0664602d)

Switch between three views:

* **Domains**: the most frequently cited domains. Each row shows the domain, its category, its citation share, and any citation tags. Domains with several cited subdomains show the subdomain count.
* **Pages**: the most referenced individual URLs. Multiple pages from the same domain can appear in this list.
* **Watchlist**: specific URLs you choose to track for citation performance. While Pages lists the most cited pages overall, the Watchlist lets you define and monitor a custom set of URLs.

### Citation Decay

Citation Decay shows how long sources keep getting cited after they first appear. Switch between **Domains** and **Pages**. The summary at the top shows the **Average Rise** (days from first citation to peak citation count) and **Average Half-life** (days for citations to fall to half of their peak and stay there) across all sources.

![Citation Decay panel with Domains and Pages tabs, Average Rise and Average Half-life highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-13.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=4eac831e50633d260738812b527c7ba6)

Each row shows the share trend, the date the source was first cited, its rise in days, peak share, peak date, half-life in days, and the date it was last cited. Use this table to spot content that is losing traction and content that answer engines have recently picked up.

### Citation Ranking

Citation Ranking is a customizable heat map of citation share across two dimensions, for example how much each domain is cited within each topic. Configure it with the **Value**, **Rows**, and **Columns** controls, the same way as Visibility Ranking.

![Citation Ranking matrix with Value, Rows, and Columns selectors highlighted above the share heatmap](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-14.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=44d34e4a4df37c82cad5244ca72538ba)

Select **Table** to switch to a standard rank table.

![Citation Ranking with the Table tab highlighted, showing ranked domain icons per topic](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-15.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=1d642a05d278a9d3b05a6a50f0d62c35)

### Citation Relationships

The Citation Relationships chart visualizes how citations are connected across answer engines and topics. Select **3D** or **2D**, then select **Show Graph** to render it.

![Citation Relationships graph with the 3D and 2D view toggle highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-16.webp?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=3e0554d4362b07ad615526d4260be660)

Clusters represent groups of cited sources that share mentions for a particular topic and platform. For example, if your brand is mentioned in two different publications that are both cited by AI answer engine platforms, those publications appear in the same cluster.

## Prompts

The **Prompts** tab shows performance for every prompt you track, and the queries answer engines generate from them.

![Prompts tab, Tracked prompts table grouped by topic with Visibility Score and Share of Voice columns](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-17.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=3d4b1ea0c79744aa0a55f322a3d28ff0)

### Tracked prompts

The **Tracked prompts** table groups prompts by topic. Expand a topic to see the prompts within it. Each row shows Visibility Score, Visibility Rank, Share of Voice, Share of Voice Rank, Average Position, Citation Share, Citation Rank, Executions, and Prompt Volume, each with its change versus the comparison period. Sort the table by a metric's current value or by its change versus the comparison period.

Controls above the table:

* **Prompt type**: filter to prompts with a given analysis type, such as Visibility, Sentiment, or FactCheck.
* **Search topics**: find a topic or prompt by name.
* **Download icon**: export the table.
* **Modify prompts**: open the Prompt Designer to add, edit, or tag prompts.
* **Settings icon**: change **Group by** (for example, topic) and **Customize columns**.

Learn more about managing prompts in [Create, manage and tag prompts](/answer-engine-insights/prompt-designer).

### Query fanouts

Answer engines often turn a single prompt into several search queries before they answer. The **Query fanouts** view shows those queries for each prompt you track.

Each row is a prompt, with its **Query Count** (the number of distinct queries generated in the period) and **Average Queries Per Execution**. Expand a prompt to see the individual queries. Select **View all** to view all queries for a prompt.

![Query fanouts tab with Query Count and Avg Queries Per Execution columns and View all button highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-18.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=ff1b9a620611fd540aed0db08661b35e)

Use this view to understand how answer engines interpret your prompts and which underlying searches your content needs to rank for.

To see the more detailed analysis of a prompt, hover over it and select **Analyze**.

![Query fanouts row with the Analyze button highlighted next to a prompt](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-19.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=600cac5f8e2ea278fbbba43bb8e211c7)

This opens the **Query Fanout Analysis** view for the prompt. Select Show Prompt Diffs to compare how queries changed across executions.

![Query Fanout Analysis panel with View Executions button and Show Prompt Diffs toggle highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-20.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=95974ab052582d8d14962faece85be42)

Select **View Executions** for a more detailed prompt executions analysis. The metrics you see depend on your prompt type. For example, for a FactCheck prompt, it's the accuracy-related views and claim analysis, and for a Visibility prompt it's Visibility Score and visibility and citations breakdown.

![Prompt Analysis panel with Accuracy Score tab, Accuracy Score Rank, and Inaccurate Claims highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-21.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=296590081141ccedc8425a3c992a47d2)

## Sentiment

The **Sentiment** tab tracks how answer engines describe your brand. Learn more in [About Sentiment](/answer-engine-insights/sentiment/about) and ​the [Get started with Sentiment](/answer-engine-insights/sentiment/getting-started) guide.

## FactCheck

The **FactCheck** tab tracks the accuracy of what answer engines say about your brand. Learn more in [About FactCheck](/answer-engine-insights/factcheck/about) and the [Get started with FactCheck](/answer-engine-insights/factcheck/getting-started) guide.

## Investigate charts with AI Marketer

To ask [AI Marketer](/aim/about) to interpret a chart, select a point or a date range on it, then select **Investigate**:

* To investigate a single date, click a point on a line chart to pin that date.
* To investigate a period, click and drag across points on a line chart to select a date range.

![Citation Share bar chart with the Investigate option highlighted in a bar tooltip](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-22.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=71a34565997edf70bb5b95c02ff48067)

Profound opens AI Marketer in the side panel and sends a prompt that includes the chart and the date or date range you selected, so you can ask what drove the change. Investigate is available in most charts, including the Visibility, Sentiment, FactCheck, Citations, and Query fanouts charts.

> [!TIP]
>
> Learn how to prompt AI Marketer in [Writing prompts for AI Marketer (Aim): best practices and examples](/aim/writing-prompts).

Select **View citations** (or **View executions** in Visibility tab) to view the underlying data for that point in the chart.

![Citation Share bar chart with the View citations option highlighted in a bar tooltip](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-23.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=ca491a5abdf86904ea3fb9f58c0ad709)

Filter and group the data in the **All data** view using the options above the table.

![All data citations view with Citations tab, filters, Aggregate by, and Search urls highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/interpret-24.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=42633ad2f81ddfce01e7788725cabc03)

## Annotate charts

Chart annotations are notes pinned to dates on time-series line charts. They show what happened, such as a launch, a model change on an AI platform, or a prompt update, next to the trend it may have affected. Learn more in [Chart annotations in AEI](/answer-engine-insights/chart-annotations).


# Personas in Answer Engine Insights
Source: https://docs.tryprofound.com/answer-engine-insights/personas

Use audience profiles to see how answers change by who is asking

A Persona is an audience profile: the demographics, employment, and behavior of the person asking an answer engine a question.

Add Personas to prompts to simulate the context an answer engine has about a user, so you can see how brand mentions, citations, and sentiment change by audience.

## Quick facts

* When you add a Persona to a prompt, Profound adds the Persona's details to the prompt text and **runs it as a separate prompt**. Each Persona on a prompt adds a prompt run. This affects your prompt limits.

* The baseline for a prompt is **No Persona**: it runs without any Persona context.

* Personas belong to a category. By default, you can create **up to three Personas per category**. Contact your Profound team to raise the limit or to add Personas to your plan.

* New Persona runs start with the next daily prompt run, so expect the first Persona data in [Answer Engine Insights](/answer-engine-insights/about) to appear after the next processing cycle.

## Add a new Persona

![Visibility tab Personas filter dropdown with Manage personas option highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=eb2cce203e577a12d9ccb4fe7020c374)

1. In [Answer Engine Insights](/answer-engine-insights/about), select **Personas** in the filter bar, then select **Manage Personas**.
2. Select **+ Create New**.

   > [!TIP]
   >
   > If **+ Create New** is disabled, you may have reached your Personas limit. Contact your Profound team to explore your options.
3. Enter a **Name** (up to 50 characters).
4. Fill in the fields that shape how an answer engine reads the prompt: **Demographics** (Gender and Age Range), **Employment** (Company Size, Industry, Role Seniority, and Job Title), and **Behavior** (Motivations and Pain Points, up to 500 characters each).
5. Select **Save**.

Fill in only the details that change how an answer engine interprets the prompt. Motivations and Pain Points carry the most weight.

## Manage and delete Personas

**Manage Personas** lists every Persona in the category with its details, the number of prompts it runs on, and when it was last edited. Open the **Actions** menu (**…**) on a Persona to:

* **Edit**: change any field, then select **Save**.
* **Duplicate**: create a copy in the same category.
* **Export**: download the Persona's details.
* **Delete**: remove the Persona and stop its prompt runs.

Personas are specific to a category, so to use the same Persona elsewhere, create it again in that category.

## Add Personas to prompts

Assign Personas in the [Prompt Designer](/answer-engine-insights/prompt-designer).

### One prompt

Select the prompt's **Personas** cell, check or uncheck the Personas you want to apply, then select **Save**.

![Prompt Designer with Personas column dropdown and Save button highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-2.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=9968d1e50f324e511279df49eee2b356)

### Several prompts

1. Check the prompts you want to change.
2. In the bar at the bottom, select **Edit** > **Personas**.

![Prompt Designer with four prompts selected and Edit menu showing Personas highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-3.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=71c62b66efce118cc36ba666cb332f64)

3. Select the Personas, then choose **Add Personas**, **Replace Existing**, or **Remove**.

![Persona picker with Add Persona, Replace Existing, and Remove buttons highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-4.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=b8967ba8bfebab513a55eb673fae88d7)

3. Select **Save**.

Removing a Persona from a prompt stops that Persona's runs for the prompt. The data collected for that prompt and persona is no longer visible in Answer Engine Insights.

## Analyze topics and prompts by Persona

To analyze prompts by Persona, use the **Personas** [filter](/answer-engine-insights/interpret#filters) in [**Answer Engine Insights**](/answer-engine-insights/about) **> Prompts**. Then, select **Analyze** on any topic.

![Prompts tab with Personas filter highlighted and dropdown listing five personas](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-5.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=73e470e9ede9cb585602f986b39de557)

To view the visibility and citations data for Personas, select [**Compare**](/answer-engine-insights/interpret#the-compare-dropdown) > **Personas** in the **Visibility** and **Citations** tabs.

![Visibility Score chart with Compare Personas dropdown highlighted and rank list by persona](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-6.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=75d65b5175a98e80c1307d1bec8265ae)

Filtering by Persona applies to [Sentiment](/answer-engine-insights/sentiment/about) and [FactCheck](/answer-engine-insights/factcheck/about) data as well, if you have at least one Persona configured. Access the filter in the [top filters bar](/answer-engine-insights/interpret#filters) in the **Sentiment** and **Factcheck** tabs.

![FactCheck tab with Personas filter highlighted and Accuracy Score chart](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/personas-7.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=c50822e6896489f032f39477f5bdf7ca)

> [!TIP]
>
> Learn more about filtering and interpreting data in [Interpret Answer Engine Insights](/answer-engine-insights/interpret).


# Create, manage, and tag prompts in Prompt Designer
Source: https://docs.tryprofound.com/answer-engine-insights/prompt-designer

Add, edit, tag, and manage prompts in Prompt Designer

> [!NOTE]
>
> **Available on:** the Enterprise plan
>
> **Find it in:** Answer Engine Insights > Prompts > Modify prompts
>
> **Required role:** Admin, Editor or custom

Prompts are the queries that Profound sends to answer engines to create your Profound dataset. [Answer Engine Insights (AEI)](/answer-engine-insights/about) data comes from these prompts, so it is important to closely maintain them.

Create, modify, tag, and manage prompts in the Prompt Designer. Access it from the **Prompts** tab in Answer Engine Insights or [Answer Engine Insights settings](/answer-engine-insights/settings).

> [!NOTE]
>
> If you don't see the **Modify prompts** option, you may not have permission to work with prompts. Learn more in [Roles and Permissions](/platform-config/people/roles-and-permissions).

## Create prompts

There are three ways to add prompts to Profound: upload, generate, or add the prompts manually.

![Prompt Designer prompt list with Generate, Upload, and Add prompt buttons highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/prompt-designer-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=1fb6a66821113a7539a2a36f8ba8071e)

New prompts typically need 24–48 hours to accumulate data after being added. Prompts refresh once per day on Eastern Time, so you should expect to see initial visibility data in [AEI](/answer-engine-insights/about) after the next daily processing cycle completes.

### Upload

Select  **Upload > Add new prompts** on the top right of the prompts table to upload a CSV file containing your prompts.

Format the CSV with these columns:

| **Column** | **Required** | **Definition** |
| - | - | - |
| Topic | Yes | Key concept or area of focus that you want to rank for in AI answers |
| Prompt | Yes | An individual query to be sent to answer engines |
| Tags | No | One or more tags to assign to the prompt for filtering within the platform |
| Regions | No | One or more regions where the prompt should be run |
| Language | No | A two-letter [ISO language code](https://www.loc.gov/standards/iso639-2/php/langcodes-search.php) followed by a two-letter [country code](https://www.iso.org/obp/ui/#search) (`en-US` for US English) |

Download the template prompts file with all necessary columns from the CSV upload window.

![Upload CSV step with Download template button highlighted and Upload file drop area](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/prompt-designer-2.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=35908272bba6a63cb60caa9ac0b72697)

**CSV upload troubleshooting**

If you see a "Topic and Prompt missing" error during bulk upload, verify that your CSV has data properly separated into columns rather than all combined in a single column. Open the file in Excel or Google Sheets to check the column structure. Make sure:

* Each field (Topic, Prompt, and so on) appears in its own column
* If all data appears in one column, use "Split text to columns" with comma as the delimiter
* Alternatively, re-save the file ensuring it uses proper comma delimiters between fields

### Generate prompts

Use Prompt Generator to create prompts from existing topics: select **Generate** on the top-right of the prompts table.

In the generator, specify the topics that you want to create prompts for and configure regions, languages, and platforms you want the prompts to be run in. Specify any additional instructions for the prompt generator in the **Additional Instructions** section.

> [!TIP]
>
> Review the autogenerated prompt preview, then click **Upload and run** to finish the setup and make sure the prompts don't get stuck in an incomplete state.

### Add prompts manually

Select **+ Add Prompt** on the top-right of the prompts table, or type in the prompt text and configure the prompt attributes directly in the first row of the prompts table (*“Add a new prompt…”*).

![Prompt Designer with Add prompt button and Add a new prompt row highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/prompt-designer-3.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=a9ddd70365888f3e0c56d045656b302f)

## Modify or delete prompts

To modify existing prompts, select the cell containing the data you want to update in the prompts table.

To edit or delete prompts in bulk, select the prompts, and then select the action you want to perform from the options at the bottom of the table.

![Prompt list with four prompts selected and Edit, Translate, Disable, Duplicate, Delete bar highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/prompt-designer-4.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=f32ef52e6d80106bdf0d53d08dd53f2b)

To edit the prompts by uploading a CSV file, select **Upload > Edit existing prompts** at the top right of the prompts table and upload the file as instructed in the Upload section above.

## Tag prompts

Tags create an additional layer of organization for prompts inside Profound. Learn more in [creating topics and prompts guide](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

To add tags to prompts, in the Prompt Designer table, select the **Tags** cell for a prompt, then select the tags you want to add or remove, or type a name for a new tag to add it to the tag list.

When editing the prompts in bulk with the CSV file, include a `Tags` column into the file to add and remove the tags for all prompts in the file.

## Use Personas with prompts

Personas let you test how an answer engine responds when the same prompt is run with additional audience context, such as a job role or decision-maker profile.

To assign Personas to a prompt, select the **Personas** cell for that prompt in the Prompt Designer table and select the Personas to apply.

To create or edit Personas, select **Manage Personas** at the bottom of the same menu. Learn more in [Personas in Answer Engine Insights](/answer-engine-insights/personas).

> [!NOTE]
>
> Adding Personas to prompts affects your prompt limits. If you assign one Persona to all prompts, you are effectively creating an additional run for each of those prompts.

## How prompt changes affect reporting

Prompt reporting is based on the prompts that are currently in your configuration. If you add, remove, or substantially rewrite prompts, the totals you see in Profound can change as well:

* **Adding prompts** increases the number of tracked questions and can raise aggregate counts after the next processing cycle.
* **Removing prompts** excludes those questions from current totals, which can make counts drop compared with an older export or snapshot.
* **Rewriting prompts** can change what Profound matches and measures, so results may no longer be directly comparable with earlier versions.
* **Changing tags** does not change the underlying prompt data by itself, but it can change which subset of prompts you are viewing if you filter by tag.
* **Adding Personas** increases the number of prompt executions because Profound runs the base prompt and the Persona-context version separately. Personas help you compare how responses differ for a given audience, but topics and tags still act as organization and filtering tools rather than prompt context.

If you need stable period-over-period comparisons, keep the prompt set consistent for the reporting window or export a snapshot before making large prompt changes.

## Topics

Topics are groups of prompts organized around a common theme. For example, a collection of different prompts all inquiring about data accuracy can be organized within a "Data Accuracy" topic.

Learn more about topics in [How to create topics and prompts to track in Profound](/answer-engine-insights/how-tos/create-topics-and-prompts).

To manually add a new topic, select the **+** option next to **All Topics** in the Prompt Designer.

![Prompt Designer topics panel with Add topic plus button highlighted next to All Topics](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/prompt-designer-5.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=7c37bc3b2892cf73d8dafcebed35b435)

After you enter the topic name, it's available to populate with prompts. Select the **…** action menu next to the topic name to rename or delete the topic.

![Topic context menu with Rename and Delete options highlighted next to LLM Optimization](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/prompt-designer-6.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=325c108b3c28b2829972f82932efda28)

> [!TIP]
>
> Topics and prompts in Profound can be exported in CSV or JSON format. Learn more in [Export prompts and topics](/answer-engine-insights/export-prompts-and-topics).​


# About Sentiment
Source: https://docs.tryprofound.com/answer-engine-insights/sentiment/about

How Sentiment tracks the narrative AI engines build around your brand

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Analytics > Answer Engine Insights > Sentiment tab

Sentiment is a part of [Answer Engine Insights](/answer-engine-insights/about) that goes beyond tracking whether your brand appears in AI responses to tracking the tone and content of how AI platforms talk about your brand.

The feature surfaces how AI platforms rate and describe your brand: an overall sentiment score, themes that AI engines consistently associate with your brand, and the attributes within each theme that carry positive or negative signals. It also scans the citations behind AI responses so you can see which sources are shaping the narrative.

## What Sentiment measures

When an AI engine mentions a brand, it does more than cite it. It describes strengths, limitations, and use cases based on the sources it draws from. That framing shapes how potential customers perceive your brand before they ever visit your site.

Sentiment helps you understand:

* Whether the tone is positive or negative, and how that changes over time
* Which specific themes AI engines surface about your brand, such as pricing, customer support, or product quality, and the attributes within each theme that carry positive or negative signals, such as “opaque pricing” (negative) or “responsive support” (positive)
* Whether negative themes come from the content of a single outdated cited page or reflect a broader pattern, so you can address the right root cause and take action immediately
* How your brand's sentiment compares to competitors, broken down by AI platform and topic

## How it works

### Which prompts get analyzed

Profound gathers Sentiment data from prompts assigned the **Sentiment type**.

The Sentiment type can be applied to any prompt in your account, including prompts you configure manually. When a prompt is marked with this prompt type, Profound analyzes the response for any brands mentioned and extracts sentiment from it.

> [!TIP]
>
> To assign the Sentiment prompt type to a prompt, select a prompt in the [Prompt Designer](/answer-engine-insights/prompt-designer#use-personas-with-prompts), click on **Edit > Analysis Types**, and select **Sentiment** in the dropup menu.

This means Sentiment data can come from any prompt. For example, a prompt like "What are the best corporate credit cards for small businesses?" might return a response that mentions a competitor but notes that their foreign transaction fees are too high. Profound picks that up and counts it as a negative attribute under a pricing theme.

### Prompt limits and timing for Sentiment prompts

Sentiment prompts count toward the same total prompt allowance as other tracked prompts in your account. New prompts and prompt configuration changes usually appear in reporting after the next daily processing cycle, and initial data may take 24-48 hours to accumulate.

### Extracting themes from AI responses

AI responses are processed by extracting raw claims, semantically grouping them into clusters determined by the data, and surfacing the themes and attributes that are statistically significant.

The overall sentiment score is the average of positive sentiment across each tracked AI model (ChatGPT, Gemini, and others). The claims extracted from AI responses are weighted by mention frequency: a positive claim mentioned 100 times has 10 times the influence on a theme's sentiment compared to a negative one mentioned 10 times. Because themes are built from aggregated claims, a theme that appears more frequently across responses naturally carries more weight in the overall score.

## Key Features

### Themes and theme attributes

A theme is a neutral, high-level topic area that AI engines associate with a brand, such as "Customer Support" or "Pricing." Under each theme, attributes capture the specific claims and patterns that make up that theme, such as "long response times" or "opaque pricing," each with a positive or negative polarity.

This hierarchy is determined by the data: raw claims from AI responses are extracted, semantically grouped into clusters, and only statistically significant groupings are surfaced. The result is a clean, actionable structure that identifies exactly what is driving a positive or negative signal.

### Exact text surfacing

When a sentiment signal is detected, Profound surfaces the exact text from the AI engine's response that the sentiment originates from. This removes ambiguity: instead of inferring why a theme is positive or negative, you can read the precise language the AI used.

### Citation-level attribution

For each sentiment signal, Profound identifies which specific citations are driving it, whether for an overall score or for a particular theme. The platform shows not just which source was cited, but where on that page the sentiment originates. This is the primary tool for distinguishing a content problem (one article driving a narrative) from a product or service problem (multiple independent sources reflecting the same issue).

### Negative Sentiment Drivers

The Negative Sentiment Drivers table shows the most cited web pages containing your top negative claims. It's useful as an actionable priority list of citation sources driving your negative sentiment. Drill down to investigate each claim, or use the Suggested Actions column to have an Agent built automatically to address negative sentiment in each citation source.

### Competitive comparison

Because Sentiment prompts run across your full competitor set, you can compare your brand's sentiment themes directly against competitors within the same view. This includes both the overall positive or negative split and the specific attributes AI engines surface for each brand.

### Time series by theme and competitor

Visualize sentiment trends over time, grouped by theme or by competitor. This lets you spot when a particular theme started trending negatively, correlate shifts with external events or content changes, and track whether corrective actions are working.

## Language and region support

Sentiment analysis supports prompts and themes across multiple languages and regions: it runs in the user's native language. Theme names in the **Themes** table appear in English, as these are labels generated by the Profound platform, but this does not affect the sentiment analysis itself. Prompts grouped under a theme are displayed in the language in which they were run and analyzed.


# Sentiment FAQ
Source: https://docs.tryprofound.com/answer-engine-insights/sentiment/faq

Frequently asked questions about Sentiment in Answer Engine Insights

> **Can the platform track sentiment for specific user personas or audience segments?**
>
> Yes. Profound supports Personas as a filter in the Sentiment tab. Learn more in [Personas in Answer Engine Insights](/answer-engine-insights/personas).

> **Can I track specific negative themes like cost concerns or safety issues, and see which AI models surface them?**
>
> Yes. Click into any theme in the Sentiment tab and add a Platforms filter with the specific AI platforms you want to track.

> **Does the platform monitor sentiment from third-party review sites like Trustpilot?**
>
> Yes. Profound crawls Trustpilot and similar review sites directly. When those pages are cited in AI engine responses, they appear in the source attribution for the relevant theme, and you can see exactly where on the page the sentiment originates.

> **Is there a limit to how many competitors I can track for sentiment analysis?**
>
> There's no hard limit, but the standard maximum is 10 competitors. In most cases, tracking three to five competitors is recommended.

> **How can I influence what AI says about my brand?**
>
> AI engines synthesize brand reputation from their training data and real-time web citations. The most effective ways to influence sentiment are earning positive coverage in frequently cited domains, creating content that directly addresses known negative themes, and updating outdated third-party content that is driving negative narratives. Brand-owned content is less effective when a negative theme is driven by many independent cited sources.


# Get started with Sentiment
Source: https://docs.tryprofound.com/answer-engine-insights/sentiment/getting-started

Verify your Sentiment setup and start using the data

Profound automatically classifies eligible prompts for Sentiment analysis. Follow these steps to verify your setup and start using the data.

## Step 1: Set the date range and establish a baseline

Go to [**Answer Engine Insights**](/answer-engine-insights/about) and select the **Sentiment** tab in the top navigation. Use the date range filter to select your monitoring window. The **Prev. Period** dropdown controls your comparison baseline.

![Sentiment Summary with the date range and comparison period selectors highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/sentiment/getting-started-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=671fbaa994f3bc4e1129985f928dd9c8)

Before optimizing, note your current positive sentiment percentage and your top positive and negative themes. Use the available filters (Topics, Platforms, Tags, and so on) to slice the data and identify where sentiment varies most across your setup.

## Step 2: Investigate and diagnose

Use the drill-down flow to investigate: start from your overall sentiment score, navigate into the themes driving it, then into the attributes within each theme, then into the specific claims behind those attributes included in the content of the cited pages.

When investigating a theme, open the cited pages to see the exact content Profound surfaced and determine whether a content fix or a product change is the right response.

[Sentiment drilldown view](https://player.mux.com/ZSBfaaKfMdjfbR2TSsF31Rf01t8YndQmFnBW31mvlNZE?metadata-video-title=Sentiment+drilldown+view\&video-title=Sentiment+drilldown+view\&autoplay=muted)

## General diagnosis advice

* Negative themes with rising mention counts: **investigate first**
* Positive themes with rising counts: **note and reinforce**
* Themes trending down: **identify what changed**
* A single negative mention with no broader trend: likely noise, **do not act on it**

## Step 3: Take action to improve your sentiment score

Review the **Negative Sentiment Drivers** table. It lists the most frequently cited web pages that include top negative claims about your brand. Take action directly from the table: select **Draft Outreach** in the **Suggested Action** column for a citation source and have [AI Marketer](/aim/about) build an [Agent](/agents/about) to draft outreach for that source.


# Answer Engine Insights settings
Source: https://docs.tryprofound.com/answer-engine-insights/settings

Configure competitors, citation categories, name matching, and entity parsing in AEI

Settings in Answer Engine Insights (AEI) is your control center for defining how Profound processes, classifies, and analyzes AI engine responses to generate insights. Think of it as a translation layer: it maps the raw text AI engines produce to the clean objects you see across your dashboards.

Open settings from the top-right corner of AEI view, next to **Export answers.**

> [!NOTE]
>
> AEI settings are specific to visualizations and data generation. To open organization and account settings, select your organization name at the top left of the platform, then select **Settings**.

## Prompts

To open the Prompt Designer, go to **Answer Engine Insights** > **Prompts** > **Modify prompts**. From there, select the prompts you want to tag or modify. Learn more in [Create, manage and tag prompts](/answer-engine-insights/prompt-designer).

## Configure Competitors

> [!NOTE]
>
> If this section is blank in your AEI settings, that's expected. Configure Competitors isn't a setup tool, it's a filter for edge cases.

During onboarding, your Profound team configures your primary competitive set. **Configure Competitors** exists for the exceptional cases where you need to manually filter which brands appear in your AEI analysis. It has two modes: **Whitelist** (include) and **Blacklist** (exclude).

| **Mode** | **What it does** | **When to use** |
| - | - | - |
| Blacklist | Removes specific brands from all AEI analysis | Rarely, and only for genuinely irrelevant or restricted brands |
| Whitelist | Shows only listed brands in the AEI analysis | Not recommended: distorts and narrows data |

### When to blacklist

Use **Blacklist** to remove a brand from your results entirely. Apply it when a brand appears in your analysis but isn't a relevant competitor, or is a competitor you aren't permitted to mention.

Common examples:

* Large platform companies (Google, Amazon, Apple) appearing in a niche competitive set
* Brands from adjacent industries that aren't true competitors
* Affiliate companies you don't want in your competitive benchmarking

To exclude a brand, switch to **Blacklist** mode, select **+ Blacklist Competitors**, and type the brand name into the list.

![Configure Competitors section with Blacklist Competitors button and Whitelist/Blacklist toggle highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=511b46f8f3a73172792c47d30ca93b5a)

### When to whitelist

> [!WARNING]
>
> Whitelisting can artificially inflate your visibility score, so we don't recommend using it.

In **Whitelist** mode, Profound only shows results from AI responses where one of the brands on your list appears. Profound excludes any response that mentions none of your listed brands from calculations entirely. That creates a narrow, less realistic view of your competitive landscape.

If relevant competitors aren't showing up in your results, refine your prompts instead so they reflect your broader industry and both direct and indirect competitors.

## Citation Categories

Citation Categories classify the sources AI engines cite when generating responses. Knowing where your citations come from, and controlling how they're grouped, is one of the most useful ways to work with citation data in Profound.

### Default categories

Profound assigns every cited domain to one of the default categories. Learn more in [Interpret Answer Engine Insights](/answer-engine-insights/interpret#citation-categories).

### Custom categories

Custom categories track any grouping that matters to your strategy: partner sites, PR placements, online travel agency sites, comparison engines, or anything else.

Custom categories are useful when you want to:

* **Define indirect competitors:** bulk-paste the 10–20 competitors outside your primary competitive set into a custom category so they're pulled out of Other
* **Run content experiments:** categorize a set of URLs or a single subpath you've optimized, then measure the before and after citation share for that content
* **Apply earned media tiers:** separate top-tier placements from general earned media for more granular analysis

> [!TIP]
>
> The **Other** category catches everything that doesn't fit the default categories, so it grows large and vague over time. Use custom categories to split it for clearer citation analysis.

### Create or edit a category

You define the **Owned**, **Competition**, and custom categories during initial setup with your account team. Profound assigns the rest automatically. Override or delete those assignments at any time with the **+ Add Citation Category** option.

![Citation Categories table with Add Citation Category button highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-2.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=5723817f1f16d65b3ebc5973f7c8c1e8)

Create custom categories in the same **+ Add Citation Category** dialog. In the **Citation Category** dropdown, type the category name in the filter field, then select **+ Add new category** to save it.

![Add Citation Category dialog with Add new category option highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-3.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=cce03971f6d7fff1d15a2aff4b802c88)

Category assignment works at the domain and subpath level. Enable **Include Subpaths** to capture every page under a domain, so tagging `tryprofound.com` also covers `tryprofound.com/blog` and `tryprofound.com/resources`. Or disable it and paste a specific path, such as `yoursite.com/blog`, to categorize that section on its own.

## Name Matching

Name Matching connects the different names that answer engine responses use for your brand and your competitors' brands. For example, Profound may appear as "Try Profound," "Profound Metrics," or "Profound Search." Name Matching groups all variations under the same asset. Without it, each variation counts as a separate entity, which makes your visibility look lower than it is.

### How it works

Profound combines string matching, fuzzy matching, and an LLM review step to group similar phrases under a single asset. Profound surfaces candidates automatically, and you decide which mappings to accept.

Name Matching has three views:

* **Mapped** — Every asset with its currently mapped mentions, plus a total mention count. Child assets nest under their parent.
* **Unmapped** — Brand mentions Profound detected but hasn't matched to an asset yet, with the date first seen and a mention count.
* **Suggestions** — Automatic suggestions for mapping mentions to assets, based on text similarity.

![Name Matching page with Mapped, Unmapped, and Suggestions tabs highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-4.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=e195f114b7271c1e9081f073dec1f43d)

Profound aggregates responses and identifies unmapped listings automatically. Map these values to an existing asset, or create a new asset for future values.

### Manage existing mappings

To manage or add mapped mentions, select **Edit Mentions** in the asset's action menu.

![Name Matching row menu with Edit Mentions option highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-5.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=c24e3686d8acfe31ba357103a24946f5)

To create a new mention, type it into the search bar and select **Add "..."**.

![Edit Mentions dialog with search field and Add mention option highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-6.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=f4902327c20890a6cd5d049838ee2f57)

### Group sub-brands under a parent brand

Sub-brands and product lines appear as child assets nested under their parent in the **Mapped** view. To change a relationship, open the row's action menu:

* **Merge** combines two assets of the same type. You can't merge owned and competitor assets.

* **Unlink from parent** promotes a child asset back to the top level.

> [!TIP]
>
> When a sub-brand and parent brand are closely related, for example "L'Oréal" and "L'Oréal Hair," Profound can't always tell them apart. Keep the mentions mapped to each asset as mutually exclusive as possible.

### Create new mappings

To map assets surfaced in the **Unmapped** tab, select the mentions, then select **+ Map Mentions**.

To create mappings from the **Suggestions** tab, select the mentions, then select **+ Add Mapping** to accept the suggestion and keep any current mappings. Select **Remap** to accept the suggestion and remove all existing mappings.

Use the **Export** and **Upload Mappings** controls to export your mappings as a CSV file or upload mappings in bulk.

![Name Matching page with Export and Upload Mappings buttons highlighted](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/answer-engine-insights/settings-7.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=d000f55c6051fa34f91eb457d2b4a9a4)

## Entity Parser Configuration

> [!NOTE]
>
> Your Profound team configures the Entity Parser during onboarding, so most customers don't need to change these settings.

Entity Parser Configuration sets which entities Profound extracts from answer engine responses, and lets you add rules for edge cases.

### What it does

Entity Parser configuration has two settings:

* **Entities to Parse** — The entity type to focus on during parsing. Options include Companies, Products, People, Product Features, and Publication Sources.
* **Additional Instructions** — Free-text guidance for the extraction model, up to 500 characters. Use it for context-specific parsing rules.

### When it's needed

Entity Parser configuration becomes relevant when:

* Your brand name is also a common word or phrase, such as "Popsicle," "Comply," or "Good Hands."
* You need to extract product names or people, such as CEOs or celebrities, alongside company names.
* Default extraction consistently over- or under-counts mentions in a way Name Matching alone doesn't fix.

If your entity extraction behaves unexpectedly, contact your Profound team rather than changing these settings yourself.

## FAQ

### How often should I review Name Matching?

Check the **Unmapped** and **Suggestions** views at least once during your first two weeks on the platform, then periodically after major events: new product launches, rebrands, or an unexpected change in your visibility score.

### I pasted a list of domains into the wrong citation category. Can I fix it in bulk?

No. Each **Add Citation Category** batch applies one category to every URL in it, so check your category selection before you save a large paste. To correct existing rows, delete them individually with the trash icon and re-add them under the right category.

### The Configure Competitors section is blank. Did I set something up wrong?

No, blank is the correct default state. Your Profound team configures your competitive set, not this section.

### Does Name Matching apply to Shopping?

Yes, Name Matching works across both AEI and Shopping. Learn more about Shopping in [About Shopping](/shopping/about).


# About Dashboards
Source: https://docs.tryprofound.com/custom-dashboards/about

Turn your Profound data into live, shareable dashboards

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Analytics > Dashboards

Custom Dashboards let you build configurable views of your Profound data and return to them any time, set up as you left them. Pick the charts that matter to you, apply filters, arrange the layout, and your reporting is ready whenever you or your stakeholders need it.

![Custom dashboard with four charts for Visibility Score, Citation Share, and Sentiment with Compare categories toggles](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/custom-dashboards/about-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=09ec58d580cd2efb22a777fd2a8381f4)

## Key features

### Configurable visualizations

Choose from Profound metrics such as Visibility Score, Share of Voice, and Citation Share, with more available across Sentiment, FactCheck, Shopping, and Agent Analytics. Display each visualization as a chart or a table, and add as many as you need.

### Persistent filters

Narrow each visualization with filters such as Topics, Regions, and Competitors, with additional filters available for specific metrics. Filters are saved with each visualization, so the dashboard opens with them applied in your next session.

### Sharing and privacy

Every dashboard starts private, visible only to people you invite. Invite specific people by email, open access to everyone in your organization, or set access to anyone with the link and share it with stakeholders outside Profound.

People in your organization who open a shared dashboard can change the date range for their own session, export it as a PDF, and duplicate it. Only the owner can edit visualizations or rename, share, or delete the dashboard. Link viewers outside Profound see a read-only dashboard fixed to the date range the owner saved.

### PDF export

Export any dashboard as a PDF directly from Profound. One click produces a clean document you can attach to an email or drop into a monthly report.


# Custom Dashboards FAQ
Source: https://docs.tryprofound.com/custom-dashboards/faq

Frequently asked questions about Custom Dashboards

> **Can I share a dashboard with someone who doesn't have a Profound account?**
>
> Yes. Open your dashboard, select **Share** and set **General access** to **Anyone with the link**. Copy the link and share it with anyone to give them a view-only access to the dashboard.

> **Can I export the raw data behind a dashboard?**
>
> No, dashboards currently export only as PDF. To work with the underlying raw numbers, pull them with the [Profound API](/rest-api/introduction), or build a [Workflow](/agents/workflows/about-workflows) that uses the matching data node (for example, [Visibility Score](/agents/workflows/nodes/data/visibility-score) or [Citation Pages](/agents/workflows/nodes/data/citation-pages)) with the same filters as your chart. To give external AI tools access to your Profound data, connect them to [Profound's hosted MCP server](/mcp/overview).

> **Can a dashboard be emailed on a recurring schedule?**
>
> Not from Dashboards directly. To send a recurring report by email, build an [Agent](/agents/about) that generates the report and emails it on a schedule.

> **Someone shared a dashboard with me and I want to build on it. How can I edit it?**
>
> Only the dashboard owner can edit a dashboard. To build on someone else's dashboard, open the dashboard actions menu and select **Duplicate**, then edit your copy.

> **Which competitors appear in my charts?**
>
> For each chart, you choose which of the competitors tracked in your category to compare. Save that competitor comparison as the chart's default view to keep it.

> **Is there a limit on how many dashboards I can create?**
>
> No. Create as many dashboards as you need, and each one keeps its own charts, filters, and layout.


# Get started with Custom Dashboards
Source: https://docs.tryprofound.com/custom-dashboards/getting-started

Create your first dashboard and add widgets

> [!TIP]
>
> Dashboards read from your Profound categories, so set up your topics and prompts first. Learn more in [How to create topics and prompts to track in Profound](/answer-engine-insights/how-tos/create-topics-and-prompts).

## Setup steps

### Step 1: Create a dashboard

In Profound, go to **Dashboards** in the left navigation bar under **Analytics**.

When you create a new dashboard, Profound opens an Aim-first creation flow. Describe in the AI Marketer chat what you want to report on, and AI Marketer builds the dashboard and adds the visualizations for you. AI Marketer can query metrics across more than 25 categories, which helps when you need one dashboard that covers many categories.

You can still choose the manual route. To create the dashboard manually:

1. Select **+ Create new**.
2. Give the dashboard a clear name so stakeholders know what it shows, and add an optional description.

### Step 2: Add visualizations

1. Select **+ Add visualization** and pick the metric, such as **Visibility Score** or **Citation Share**, from the **Visualization** dropdown.
2. Toggle between **Chart** and **Table** to choose how the visualization displays.

A dashboard isn't limited to one category. Each visualization can read from its own category, so one dashboard can show different metrics for different categories, and a visualization can show a metric rolled up across several categories.

### Step 3: Apply filters and save

Narrow the visualization with filters such as **Topics**, **Regions**, or **Competitors**, then select **Save changes**. Repeat for each visualization you want on your dashboard.

When you’re ready to save your dashboard, select **Save view** on the top right of the screen. Your dashboard will appear on the **Dashboards** page.

### Step 4: Set the date range and arrange the layout

Open the dashboard and use the filters at the top of the screen to set the date range and the time period it compares against. Your selection saves automatically and becomes the default everyone sees. To rearrange the layout, select **Edit view**, drag and drop your charts, and select **Save changes** when done.

### Step 5: Share or export

Select **Share** to invite specific people to view the dashboard, open access to your entire organization, or allow anyone with the link to view. Select **Export as PDF** when you need a document instead of a live view.

## Tips and best practices

* **Build one dashboard per audience.** A regional lead, an executive, and a content team each read different metrics, and separate dashboards keep each view focused.
* **Name dashboards by audience and purpose.** A clear name like "Hotels: brand visibility" prevents confusion as your dashboard list grows, and the optional description can record who the dashboard is for.
* **Use prompt tags to track a specific set of prompts.** For example, tag the prompts your new content targets, then filter the dashboard by that tag to track visibility for just those prompts.
* **Set the date range before sharing a link.** Link viewers see the range you set, so pick the window your stakeholders should look at.
* **Use AI Marketer to build faster.** Describe the report you need in plain language, and AI Marketer creates the dashboard and adds the visualizations for you. Open an existing dashboard in an AI Marketer chat to ask questions about the data or add and rearrange charts.


# Cloudflare setup
Source: https://docs.tryprofound.com/dynamic-bot-rendering/cloudflare

Route AI assistant traffic to Profound Dynamic Bot Rendering with a single Cloudflare Worker, keeping human and search-crawler traffic on your origin.

This guide adds one Cloudflare Worker to your zone. The Worker inspects the `User-Agent`, forwards matching AI assistant requests to Profound, and passes everything else straight through to your origin.

The Worker calls Profound's general endpoint, `GET https://concierge.tryprofound.com/v1/concierge`, and supplies the original host and path as headers.

> [!WARNING]
>
> A Worker on a `/*` route runs on every request to that hostname, so it sits in your production request path. The code retries your origin for fetch failures, a response-header deadline, and statuses `400`, `401`, `403`, `500`, `502`, `503`, and `504`. It passes other statuses through, including redirects, `404`, and `429`. Adapt the public-page allowlist and latency budget, then test on a staging hostname first. A failure after response streaming starts cannot transparently replace a partially sent body.

## Prerequisites

* A Cloudflare account with Workers enabled on the zone
* An existing proxied DNS record for the hostname and an origin reachable through it. Use a [Worker Route](https://developers.cloudflare.com/workers/configuration/routing/routes/), not a Worker Custom Domain.
* An inventory of existing and overlapping Worker routes. Preserve their logic rather than replacing them. If composing them requires a design change, stop and review that separately.
* A recorded route failure mode and capacity check against [Workers limits](https://developers.cloudflare.com/workers/platform/limits/). Use fail-open for quota exhaustion only where bypassing the Worker preserves existing security. Fail-closed can affect human traffic too, before the JavaScript catch runs.
* Permission to add Worker routes for the hostname
* Your Dynamic Bot Rendering API key, a registered domain, serving policy, and a known published test page. See [Before you start](/dynamic-bot-rendering/overview#before-you-start).

## Setup

**1. Create the Worker**

In the Cloudflare dashboard, open **Workers & Pages → Create → Create Worker**. Name it `profound-bot-rendering` and deploy the placeholder so the Worker exists.

**2. Add the Worker code**

For the new Worker, open **Edit code** and add the following. Replace the four example paths with reviewed public pages. Do not overwrite existing Worker logic.

```js theme={null}
const CONCIERGE_ENDPOINT = 'https://concierge.tryprofound.com/v1/concierge';

// Keep this list in sync with the supported assistants in the Profound docs.
const BOT_RE = /duckassistbot|chatgpt-user|gemini-deep-research|perplexity-user|amzn-user|mistralai-user|claude-user|claude-code|codex/i;

// Explicit public pages only. Review each path before extending this list.
const PUBLIC_PATHS = new Set(['/', '/pricing', '/docs', '/docs/getting-started']);
const UPSTREAM_TIMEOUT_MS = 2000; // Example header deadline; leave time for origin fallback.
const FORWARDED_HEADERS = ['user-agent', 'accept', 'accept-encoding', 'accept-language'];

// Statuses that mean "Profound could not serve this" rather than "the origin
// said so". 404 is deliberately absent: Profound fails open internally and
// returns your origin's own status, so a 404 is your real 404.
const FAILOVER_STATUSES = new Set([400, 401, 403, 500, 502, 503, 504]);

async function failOpen(request) {
  // A subrequest to the same zone goes to the origin without re-running this Worker.
  const originResponse = await fetch(request);
  const response = new Response(originResponse.body, originResponse);
  response.headers.set('x-concierge-cdn-failover', '1');
  return response;
}

export default {
  async fetch(request, env) {
    const url = new URL(request.url);
    const userAgent = request.headers.get('user-agent') || '';

    // Loop protection: Profound stamps this on its own origin fetches.
    // Passing it through means those fetches reach the origin directly.
    if (request.headers.has('x-concierge-request')) return fetch(request);

    if (request.method !== 'GET') return fetch(request);
    if (request.headers.has('cookie') || request.headers.has('authorization')) return fetch(request);

    const isAssistant = BOT_RE.test(userAgent);
    if (!isAssistant || !PUBLIC_PATHS.has(url.pathname)) return fetch(request);

    // Copy only required negotiation headers, never viewer credentials or identity.
    const headers = new Headers();
    for (const name of FORWARDED_HEADERS) {
      const value = request.headers.get(name);
      if (value !== null) headers.set(name, value);
    }

    headers.set('x-concierge-api-key', env.PROFOUND_CONCIERGE_KEY);
    headers.set('x-concierge-host', url.hostname);
    headers.set('x-concierge-url', url.pathname + url.search);

    const controller = new AbortController();
    const timeout = setTimeout(() => controller.abort(), UPSTREAM_TIMEOUT_MS);
    try {
      const response = await fetch(CONCIERGE_ENDPOINT, {
        method: 'GET',
        headers,
        redirect: 'manual', // pass redirects through instead of following them
        cache: 'no-store', // also verify the endpoint owner's cache policy
        signal: controller.signal,
      });

      if (FAILOVER_STATUSES.has(response.status)) {
        if (response.body) void response.body.cancel().catch(() => {});
        controller.abort();
        return failOpen(request);
      }
      const result = new Response(response.body, response);
      result.headers.set('cache-control', 'no-store');
      return result;
    } catch {
      return failOpen(request);
    } finally {
      clearTimeout(timeout);
    }
  },
};
```

The example allows 2 seconds for Profound response headers. Choose a bounded value that leaves enough time for your origin response within the assistant's latency budget. The deadline ends when headers arrive; it does not bound response-body streaming. Workers supports [AbortController](https://developers.cloudflare.com/workers/runtime-apis/web-standards/).

**3. Store the API key as a secret**

Open **Settings → Variables and Secrets**, add a secret named `PROFOUND_CONCIERGE_KEY`, and paste your API key. Save and deploy.

Use a secret rather than a plaintext environment variable. Secrets are write-only once saved, so the key stays hidden from the dashboard.

To do the same with Wrangler:

```bash theme={null}
npx wrangler secret put PROFOUND_CONCIERGE_KEY
```

**4. Attach the Worker to your hostname**

Open **Settings → Domains & Routes → Add route**, enter the pattern for your page traffic, and select the zone.

```text theme={null}
www.example.com/*
```

If your site serves assets from the same hostname, the Worker still runs on them and returns them with a plain origin fetch. To keep the Worker out of that path entirely, narrow the route pattern instead of relying on the code's path check.

**5. Verify**

Check routing, then separately verify the known published page and excluded traffic:

```bash theme={null}
curl -sD - -o /dev/null https://www.example.com/pricing -A 'chatgpt-user'
```

Check the HTTP status, a valid `x-concierge-request-id`, and `x-concierge-bot-kind: chatgpt-user`. On your published test page, also require `x-concierge-cache: HIT` and the expected HTML body. A normal browser request should stay on your origin route. Test non-`GET`, API, asset, and credential-bearing requests too.

Full checks and header meanings are in [Verify and troubleshoot](/dynamic-bot-rendering/verify-and-troubleshoot).

## Caching

The Worker's upstream call must never be cached. Every page uses the same upstream URL, `https://concierge.tryprofound.com/v1/concierge`, and the host, path, and query that identify the page travel only in trusted headers. If that URL were cached without those headers in the cache key, different pages or hosts would collide, and adding `User-Agent` to the key would not fix it.

The example uses the [`cache: 'no-store'` fetch setting](https://developers.cloudflare.com/workers/runtime-apis/fetch/) and marks returned Profound responses `Cache-Control: no-store`. [Cloudflare evaluates caching on the fetch subrequest URL](https://developers.cloudflare.com/workers/reference/how-the-cache-works/), not the viewer URL, so cache rules on your own zone do not apply to the upstream call.

Before rollout, send repeated assistant requests to two published paths and a query variant of each, and confirm every response carries a fresh `x-concierge-request-id`. Also check any other cache layer in front of your site for assistant and human separation, and for what happens to cached entries after you publish a replacement, unpublish, or pause. Do not add a `User-Agent`-only cache key as a workaround. If you need to cache Profound's responses at your edge, contact Profound support first: the cache key has to include the host, path, and query identity and be invalidated on publication changes.

## How the Worker protects your site

* **Upstream headers are allowlisted.** For routed requests, the Worker copies only `User-Agent`, `Accept`, `Accept-Encoding`, and `Accept-Language`, then sets the trusted key, host, and path. It does not copy client-supplied `x-concierge-*` identity or credentials. Requests carrying cookies or authorization bypass Profound entirely. Ordinary bypass requests retain their headers for your origin.
* **Loop protection is not authentication.** The marker is spoofable; never use it to bypass WAF rules or access controls. See the [public-content prerequisites](/dynamic-bot-rendering/overview#before-you-start).
* **Loop protection is two-sided.** Profound stamps `x-concierge-request` on its own origin fetches, and the Worker passes those straight through. Profound also rejects any inbound request carrying that header, so even a misconfigured rule breaks a single request instead of looping.
* **The key never reaches the browser.** It lives in a Worker secret and is only attached on the upstream call.
* **Redirects are passed through.** `redirect: 'manual'` returns a redirect to the assistant instead of following it, so redirect handling stays your origin's decision.


# Amazon CloudFront setup
Source: https://docs.tryprofound.com/dynamic-bot-rendering/cloudfront

Route AI assistant traffic to Profound Dynamic Bot Rendering with a single CloudFront Function and an origin group, without Lambda@Edge.

This guide connects an existing CloudFront distribution to Dynamic Bot Rendering. This setup adds one CloudFront Function on viewer request to select an origin for eligible AI assistant page requests.

This setup uses no Lambda\@Edge. Two properties support this integration:

* With viewer `Host` forwarding disabled, CloudFront uses the custom origin domain for its `Host` header and TLS Server Name Indication (SNI). Preserve this setting in both policies in step 3.
* Profound's CloudFront endpoint returns `cache-control: no-store` itself, so CloudFront never stores a second, staler copy of content Profound already caches.

> [!NOTE]
>
> Requests routed to Profound use the CloudFront endpoint `GET https://concierge.tryprofound.com/v1/concierge/cloudfront/{path}`. You never construct that URL by hand. The origin's **Origin path** setting prepends the prefix, and the request URI passes through untouched.

## Why the design looks like this

Two CloudFront rules shape everything below:

1. A viewer-request function can't change which cache behavior a request matches. CloudFront selects the behavior before the function runs, and rewriting the URI doesn't re-match it. Assistant traffic therefore stays in the same behavior as human traffic, rather than a dedicated behavior with its own cache policy.
2. A viewer-request function can switch the origin, including building a per-request origin group with failover (`cf.createRequestOriginGroup`, runtime `cloudfront-js-2.0`).

Two consequences follow:

* Routing happens by swapping the origin, not the behavior. Because **Origin path** is a per-origin setting, a failover to your default origin automatically uses the original, unprefixed path. No path rewriting is needed.
* Assistant and human requests share the behavior's cache policy, so the cache key must be split with a marker header (step 3). Without it, CloudFront can serve a human-cached copy to an assistant, or the reverse.

## Prerequisites

* An existing CloudFront distribution

* AWS Identity and Access Management (IAM) permissions for CloudFront distributions, functions, and cache policies

* Your Dynamic Bot Rendering API key, a registered domain, an effective serving policy, and a known published test page. See [Before you start](/dynamic-bot-rendering/overview#before-you-start).

* Public, non-personalized pages only. The example allows `/`, `/pricing`, `/docs`, and `/docs/getting-started`. Replace these with reviewed public pages; do not include APIs, authentication, account, admin, or framework-resource paths.

* Record existing function associations, origins, and policies for rollback. Preserve existing security, authentication, and rewrite logic. If combining it requires a routing design change, stop and review that separately.

* Confirm that firewall rules, bot challenges, and origin allowlists permit both assistant traffic and Profound's origin/scraper fetches. The loop marker is not an authentication or firewall bypass.

## Setup

**1. Create the Profound origin**

Open **CloudFront → Distributions → your distribution → Origins → Create origin**.

| Setting | Value |
| - | - |
| Origin domain | `concierge.tryprofound.com` |
| Origin path | `/v1/concierge/cloudfront` |
| Name | `Profound_Origin` |
| Protocol | HTTPS only |

Add two custom origin headers:

| Header | Value |
| - | - |
| `x-concierge-api-key` | Your Dynamic Bot Rendering API key |
| `x-concierge-host` | Your public site host, for example `www.example.com` |

CloudFront overwrites any viewer-sent header of the same name when forwarding to this origin, so a client can't spoof the API key or the host. The function in step 2 also strips them, for defense in depth.

> [!NOTE]
>
> **Serving more than one hostname from this distribution?** A static `x-concierge-host` only works for a single public host, and static origin headers win over anything the function sets. For a multi-domain distribution, omit that custom header here, uncomment the marked line in the function below, and include `x-concierge-host` in the cache key in step 3 so it is also forwarded. Forwarding alone does not separate cached responses for different hosts. Review host identity for human and fallback responses too before enabling multiple hosts.

**2. Create the viewer request function**

Open **CloudFront → Functions → Create function**. Name it `profound-bot-rendering` and choose the runtime `cloudfront-js-2.0`. Origin groups require the 2.0 runtime.

```js theme={null}
import cf from 'cloudfront';

// Keep this list in sync with the supported assistants in the Profound docs.
var BOT_RE = /duckassistbot|chatgpt-user|gemini-deep-research|perplexity-user|amzn-user|mistralai-user|claude-user|claude-code|codex/i;

// Replace with reviewed public, non-personalized page paths.
var PUBLIC_PATHS = ['/', '/pricing', '/docs', '/docs/getting-started'];
var PUBLIC_HOSTS = ['www.example.com'];

function handler(event) {
    var request = event.request;
    var headers = request.headers;

    // Never trust these from a client. The API key and host come from the
    // origin's custom headers; the route marker feeds the cache key, so a
    // spoofed value could pollute the assistant cache entry.
    delete headers['x-concierge-api-key'];
    delete headers['x-concierge-host'];
    delete headers['x-concierge-url'];
    delete headers['x-concierge-route'];

    // Loop protection: Profound stamps x-concierge-request on its own origin
    // fetches. Never route those back to Profound.
    if (request.method !== 'GET' || headers['x-concierge-request']) {
        return request;
    }

    // CloudFront exposes cookies separately from request.headers.
    if (headers.authorization || Object.keys(request.cookies || {}).length > 0) {
        return request;
    }

    var ua = headers['user-agent'] ? headers['user-agent'].value : '';

    var isAssistant = BOT_RE.test(ua);
    var isPublicPage = PUBLIC_PATHS.indexOf(request.uri) !== -1;
    var host = headers.host ? headers.host.value : '';

    if (!isAssistant || !isPublicPage || PUBLIC_HOSTS.indexOf(host) === -1) {
        return request;
    }

    // Split the cache key. This header is in the behavior's cache policy,
    // so assistant responses never collide with human cache entries.
    headers['x-concierge-route'] = { value: 'bot' };

    // Multi-domain distributions only (see Step 1): derive the host from
    // the viewer Host header instead of a static origin custom header.
    // headers['x-concierge-host'] = { value: headers.host.value };

    // Try Profound first; fall back to your own origin if Profound is down
    // or misconfigured. CloudFront accepts 400, 403, 404, 416, 429, 500,
    // 502, 503 and 504 as failover criteria. 401 is not allowed, so a 401
    // from a bad API key surfaces to the assistant instead of failing over.
    // 404 is deliberately omitted: Profound fails open internally and
    // returns your origin's own status, so a 404 is your real 404 and
    // retrying would double-fetch. 429 is also omitted and passes through.
    cf.createRequestOriginGroup({
        originIds: [
            { originId: 'Profound_Origin' },
            { originId: 'YOUR_DEFAULT_ORIGIN' }
        ],
        failoverCriteria: {
            statusCodes: [400, 403, 500, 502, 503, 504]
        }
    });

    return request;
}
```

Replace `YOUR_DEFAULT_ORIGIN` with your distribution's existing default origin ID and `PUBLIC_HOSTS` with the registered hosts for this origin, then select **Publish**. For the static host header, list only that one host. Keep the exact public-path allowlist aligned with your site's routes. Requests with cookies or authorization keep their existing routing; the function does not send them to Profound.

The function routes only `GET`; all other methods return to existing routing. CloudFront origin groups support failover for `GET`, `HEAD`, and `OPTIONS`. The configured `503` criterion enables connection-failure failover; `504` enables response-timeout failover. Before deployment, set and record a bounded primary-origin connection-attempt, connection-timeout, and response-timeout budget that leaves time for the fallback request. CloudFront defaults can spend 30 seconds on connection attempts alone (three attempts of 10 seconds). Test the resulting total latency; do not assume failover is immediate. See [CloudFront origin failover](https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/high_availability_origin_failover.html).

**3. Update the cache and origin request policies**

CloudFront strips any viewer header or query string the behavior's policies don't include. This happens before the origin request, regardless of what the function decided. Three settings are required on the behavior serving your pages.

**1. Add `x-concierge-route` to the cache-key headers.** AWS managed cache policies can't be edited. If the behavior uses one, copy every setting into a new custom policy first, then add the header. Keep **Minimum TTL** (time to live) at `0`. In the multi-domain variant, include the trusted `x-concierge-host` in cache identity as well as forwarding it. The bot marker alone does not distinguish hosts.

**2. Include the query strings your pages vary on** (or "All") in the cache key. Cache-key query strings also reach the origin. An origin request policy can forward additional query strings without keying them, but responses that vary on those values can then share a cache entry. If neither policy includes a query string, CloudFront strips it. Verify both forwarding and cache identity for each supported variant.

**3. Forward the viewer `User-Agent` through the origin request policy.** By default, CloudFront replaces `User-Agent` with `Amazon CloudFront` on origin requests. Profound would then classify every routed request as an unknown agent and permanently fail open, with no renders served and none scheduled. Use a custom header allowlist for `User-Agent` and, when required, `Accept` and `Accept-Language` (plus `x-concierge-host` in the multi-domain case). Preserve required compression settings. Do not use **AllViewer**: viewer `Host` forwarding can override the service hostname and break origin routing or TLS. Exclude viewer `Host`, `Cookie`, and `Authorization` from both the cache and origin request policies; cache-key values are forwarded too. These policies apply to human traffic as well. If the existing behavior needs these headers or cookies, stop and review compatibility before changing it. CloudFront documents per-origin `originOverrides` for `hostHeader` and `sni` in [request origin groups](https://docs.aws.amazon.com/AmazonCloudFront/latest/DeveloperGuide/helper-functions-origin-modification.html); a customer origin requiring a different host needs that compatibility reviewed separately. This example does not choose or add an override. Profound's origin fetches do not forward viewer credentials, and published pages are not per-session or per-language variants.

> [!WARNING]
>
> For this setup, add `User-Agent` to the origin request policy rather than the cache key. An origin request policy applies to the whole behavior, so CloudFront now forwards the viewer `User-Agent` on human requests too. If your origin varies its HTML by `User-Agent` (separate mobile and desktop markup, for example) without corresponding cache-key variation, CloudFront can serve one cached variant to other viewers. Confirm that the existing cache key already accounts for your origin's variants. Locale, geography, and other header-dependent content also need review. If this requires a cache design change, stop before deployment. Raw `User-Agent` keying can substantially reduce cache reuse.

**4. Associate the function with the behavior**

Open **Behaviors → your page behavior → Edit**. Set the cache policy from step 3, then under **Function associations** set **Viewer request** to your `profound-bot-rendering` CloudFront Function. For a behavior without existing edge logic, no origin request or origin response association is needed. Do not remove existing associations. Merge and preserve existing viewer-request logic before associating this function; stop if that requires a design decision.

Save and wait for the distribution to finish deploying.

**5. Verify**

```bash theme={null}
curl -sD - -o /dev/null https://www.example.com/pricing -A 'chatgpt-user'
```

Check the HTTP status, a valid `x-concierge-request-id`, and the expected `x-concierge-bot-kind`. On your known published page, also require `x-concierge-cache: HIT` and the expected rendered body. Test normal browser traffic, non-GET methods, excluded paths, and credential-bearing requests against their existing behavior.

Test CDN-level failover only in staging or an isolated distribution. Record the original origin settings, configure an unreachable test origin, and repeat an uncached assistant request. Expect the default-origin response after the configured timeout, with no `x-concierge-request-id`. Restore the original settings and wait for propagation. A `401` does not trigger failover; the configured `400` and `403` statuses do, which can mask configuration errors. Cached fallback responses can persist until their TTL expires.

Full checks and header meanings are in [Verify and troubleshoot](/dynamic-bot-rendering/verify-and-troubleshoot).

## How the cache split behaves

* Human requests never carry the marker, because the function strips any client-supplied copy. The marker separates their entries from assistant entries, but query and header policy changes can affect human cache hit rates.
* Eligible assistant requests carry `x-concierge-route: bot` and get their own entries.
* Profound's responses set `cache-control: no-store`, so with Minimum TTL at `0` the assistant-keyed entries are only ever populated by failover responses from your own origin: correct content, expiring on your origin's own TTL.

> **Known gap: failover responses can linger after recovery**
>
> During a full Profound outage, failed-over origin responses can sit in the assistant-keyed cache until their TTL expires, delaying the return to rendered content after recovery.
>
> Validate that this recovery delay is acceptable for your existing TTLs. Changing how fallback responses are cached requires a separate design review; this guide does not add an edge function to change that behavior.

## Rolling back

Restore the recorded viewer-request association and any policies changed for this setup. Use **No association** only if there was no previous function. Allow the configuration to propagate, then verify the original routing and cache behavior. Previously cached fallback responses can remain until their TTL expires or they are invalidated through your normal process.


# Dynamic Bot Rendering
Source: https://docs.tryprofound.com/dynamic-bot-rendering/overview

Serve AI assistants a fully rendered version of your pages from the edge, while other traffic stays on your existing route.

Many AI assistants fetch your pages with a plain HTTP request and no JavaScript engine. On a client-rendered site, they may see an empty shell instead of your content.

Dynamic Bot Rendering fixes this at your content delivery network (CDN). You add one small routing rule to the CDN you already run. That rule recognizes AI assistant traffic by `User-Agent` and forwards it to Profound. For the public pages you have opted in, Profound returns a published, pre-rendered copy of the page when one exists. For everything else it passes your origin's live response straight through. Human visitors, search crawlers, API calls, and assets never leave your existing route.

> [!NOTE]
>
> Dynamic Bot Rendering is served by Profound's Concierge service, so its request and response headers use the prefix `x-concierge-`.

## How a request flows

**1. Your CDN classifies the request**

Your routing rule matches the `User-Agent` against the [supported assistant list](#supported-ai-assistants). Only matching requests go to Profound. The Cloudflare and CloudFront examples also restrict routing to a path allowlist; the Vercel rule forwards all assistant traffic and lets your Profound policy decide.

**2. Your CDN forwards the request to Profound**

The rule attaches your API key, your public host, and the original path and query as trusted headers. It sets those values from its own configuration and the parsed request, never from headers a client sent. Profound ignores client-supplied `x-concierge-*` identity headers.

**3. Profound serves a published render, or falls back to your origin**

Profound checks the API key, confirms the host is registered to your organization, and resolves the [policy](#how-rendering-and-serving-are-controlled) for the path. If the path has a published render and policy allows serving, Profound returns that render. Otherwise it fetches the page from your origin and returns the live response unchanged. Depending on policy, that miss can also schedule a background render of the page as a new candidate.

**4. The assistant receives your page**

A published render is fully rendered HTML: the same page a browser produces after JavaScript runs. Rendering makes your content readable to the assistant. It does not guarantee inclusion or citation in an answer.

## Supported AI assistants

Profound identifies the assistant from the original `User-Agent` on every request, using the same tokens your CDN rule matches on. Matching is a case-insensitive substring test.

| Token | Assistant |
| - | - |
| `chatgpt-user` | ChatGPT browsing |
| `codex` | OpenAI Codex |
| `claude-user` | Claude |
| `claude-code` | Claude Code |
| `perplexity-user` | Perplexity |
| `gemini-deep-research` | Gemini Deep Research |
| `duckassistbot` | DuckDuckGo DuckAssist |
| `amzn-user` | Amazon |
| `mistralai-user` | Mistral |

> [!WARNING]
>
> Keep your CDN rule's list in sync with this table. Profound classifies traffic independently, so a `User-Agent` your rule forwards but Profound doesn't recognize is treated as an unknown agent. It fails open to your origin and adds a round trip without serving or scheduling a render.

## Search crawler routing

Ordinary Googlebot and Bingbot requests fall outside the list above and stay on your existing route. Test your site's cache and edge rules to confirm this before rollout. Dynamic Bot Rendering doesn't cloak: the rendered page is the same page a browser produces after JavaScript runs.

## How rendering and serving are controlled

Two independent settings decide what an assistant gets for a given URL:

* **Policy** answers "may Profound serve stored content for this path, and to which assistants?"
* **Publication** answers "which exact render is live for this path?"

Both have to line up before a stored render is served. A published page under an `off` policy serves nothing. A `serve` policy on a path with no publication falls back to your origin. You manage both yourself in the Profound dashboard.

### Policy modes

| Mode | What the assistant receives | Renders new candidates in the background? |
| - | - | - |
| `off` | Your origin's live response | No |
| `warm` | Your origin's live response, even if a publication exists | Yes, when the request is eligible |
| `serve` | The published render when one is usable, otherwise your origin | Yes, on a miss when the request is eligible |

`off` is the default for any path no rule covers. A newly registered domain does nothing until a rule opts a path in, so onboarding is inert until you decide what to render. Use `warm` to build up candidates for review before you start serving. `serve` never publishes anything by itself; it only permits an existing publication to be delivered.

### Rules and precedence

Every rule belongs to one registered domain. A **domain default** sets the baseline for the whole host, and **page rules** override it for specific paths. A page rule matches an exact path (optionally pinned to a query string), a path prefix, or a glob such as `/blog/*`. Prefix and glob rules match on the path without its query string.

For each request Profound picks one winning rule:

1. An exact rule for the path and query string.
2. Otherwise, an exact rule for the path.
3. Otherwise, the best matching prefix or glob rule. Higher priority wins, then the more specific pattern.
4. Otherwise, the domain default.
5. With no rule at all, the effective mode is `off`.

Only the winning rule applies. Any setting it leaves unset is inherited from the domain default, never from a broader page rule that also matched. For example, with a domain default of `serve`, a `/docs/*` rule set to `warm`, and an exact `/docs/start` rule that only changes its allowed assistants, `/docs/start` resolves to `serve`. The `/docs/*` rule does not contribute.

A rule can also be deactivated. An inactive rule is skipped entirely, so the domain default or another matching rule takes over for that path. Deactivating an `off` rule that excluded a path can therefore turn serving back on. To keep a path on your origin, use an active rule with mode `off`.

### Per-rule settings

Each rule can set any of these and inherit the rest:

* **Mode.** `off`, `warm`, or `serve`, as above.
* **Allowed assistants.** Restrict which supported assistants receive treatment on the matched paths. Unset inherits the domain's list, and when nothing sets a list every supported assistant is eligible. A list replaces the inherited list rather than intersecting with it, and an empty list excludes every assistant. An excluded assistant receives your origin's response with an effective policy of `off`. The published render itself is shared: you cannot publish different versions for different assistants.
* **Automatic rendering.** Allow or block background candidate renders independently of mode. `serve` with automatic rendering off serves only what you publish by hand. `off` never renders automatically, whatever this setting says.

### Pausing a domain

Pausing a domain sends every assistant request for that host to your origin while keeping the registration, rules, renders, and publications intact. Resuming restores the saved behavior. Use pause when you want to stop serving quickly without touching your CDN or your rules. While paused, you can still render, review, and publish pages.

## From render to live page

Every render starts as a **candidate**: a stored snapshot of one URL, made either in the background by an eligible request or on demand. A URL can accumulate several candidates. None of them is served until you publish one.

**1. Review**

A candidate that differs sharply from the page currently published for that URL is held as **pending**. That happens when its visible text collapses to a fraction of the live page's, or when both the title and the first heading change. Preview the candidate, then approve or reject it. Approval marks it eligible for publication and nothing more. When a URL has no published page yet, there is nothing to diverge from and the candidate is stored as approved.

**2. Publish**

Publishing selects one exact render as the live page for that URL. Publishing a pending candidate approves it in the same step. A rejected candidate cannot be published.

**3. Roll back, unpublish, or replace**

Rolling back is publishing an earlier render for the same URL. Unpublishing removes the live page so assistants return to your origin, and keeps the render history. Rejecting a page that is currently published also unpublishes it. To refresh content, render a new candidate, review it, and publish it.

A published render stays live until you publish a replacement or unpublish it. Profound never replaces a publication automatically, and a render does not stop serving because it has aged. If a stored render can no longer be read, Profound falls back to your origin for that request rather than serving a different render.

Page identity is the public host plus the exact path and query string. Publishing `/pricing` does not publish `/pricing?plan=pro`, and each variant is rendered, reviewed, and published on its own.

## Fail-open by design

Fail-open happens at two layers, and neither covers every failure on its own.

> **CDN-level failover**
>
> The Cloudflare Worker retries your origin for fetch failures, a response-header deadline, and a configured list of statuses. CloudFront uses an origin group with configured failure criteria and timeouts. Vercel's routing rule has no CDN-level failover.

> **Service-level fail-open**
>
> Once a request is authenticated and valid, Profound fetches your origin whenever it cannot serve a stored render: no publication, policy set to `off` or `warm`, an unknown assistant, an unreadable render, or an internal lookup failure. Your origin must still be reachable.

> [!WARNING]
>
> On Vercel, a Profound outage or a hard rejection can reach the assistant as an error, because the external rewrite has no secondary origin to retry. Cloudflare and CloudFront customers get an extra layer of protection from their CDN failover.

A bad API key, an unregistered host, or a malformed path is rejected by Profound rather than proxied, so a misconfigured rule cannot leak requests to your origin under Profound's name. Cloudflare and CloudFront can mask some of these rejections through their CDN failover, which means a working page in a browser does not prove the integration is correct. Inspect the `x-concierge-*` response headers, as described in [Verify and troubleshoot](/dynamic-bot-rendering/verify-and-troubleshoot).

## Scope and limits

* **`GET` only.** Other methods, including `HEAD`, are not part of the contract. Use `curl -sD - -o /dev/null`, not `curl -I`, when testing.
* **Public, non-personalized pages only.** Opt in only reviewed public pages, whether through your policy rules in Profound or through a path allowlist in your CDN rule. An extensionless path is not proof of a page: keep APIs, GraphQL, authentication, administration, framework resources, and session-sensitive routes out. The Cloudflare and CloudFront examples allow only `/`, `/pricing`, `/docs`, and `/docs/getting-started`; replace these with your own paths.
* **Query strings are part of the page identity.** `/products?id=a` and `/products?id=b` are distinct pages, stored and rendered separately.
* **One host per rule.** The public host your rule sends must match a domain registered in Profound. Serve several hosts with one rule per host.
* **No credentials reach your origin through Profound.** Profound does not forward viewer cookies or authorization headers when it fetches your origin, and the example rules keep credential-bearing requests on your existing route. Protected pages may render as a login page or public content, which is a compatibility boundary rather than a leak.
* **One render per URL, not per visitor.** A stored render is not varied by session, language, geography, or other request headers. If the same URL serves different content by those signals, preview the render and confirm that one version is acceptable for every assistant before adding it to your allowlist.

## Before you start

Complete these steps in the Profound dashboard before configuring your CDN. The Dynamic Bot Rendering endpoint your CDN will call is `https://concierge.tryprofound.com`.

Also record your existing CDN routes, functions, cache policies, and rollback settings before you change anything, and check that web application firewall (WAF) rules, bot challenges, deployment protection, and origin allowlists let both assistant requests and Profound's origin fetches through. If your origin needs a way to identify Profound's fetches, contact Profound support.

**1. Register your domain**

Open **Dynamic Bot Rendering** in the Profound dashboard and add the exact public host you serve, such as `www.example.com`. The host must resolve publicly and answer over HTTPS. Register the host visitors reach: if your apex domain redirects to `www`, register the `www` host rather than the apex, and never an internal origin hostname. The hostname is fixed once registered.

**2. Create an API key**

Create a Dynamic Bot Rendering API key and copy it. The raw value is shown once, so store it in your CDN's secret store immediately.

One key covers every domain registered to your organization, including domains you add later, so most customers need only one. Create additional keys when you want to rotate without downtime, or to manage rotation and revocation separately across properties. Additional keys are still organization-wide, not restricted to one property.

**3. Set a policy and publish a test page**

Give the domain a policy that opts at least one public page in with mode `serve`, render a candidate for that page, review it, and publish it. Keep the published URL and its expected HTML for verification.

After configuring your CDN, an assistant request for that URL should return `x-concierge-cache: HIT` and the rendered HTML. Repeated requests never substitute for publication: a `MISS` stays a `MISS` until a render is published.

> [!WARNING]
>
> Revoking a key affects every site using it. If you share one key across properties, rotate deliberately: create the replacement, roll it out to each CDN, then revoke the old one.

## Choose your CDN

> **[Cloudflare](/dynamic-bot-rendering/cloudflare)**
>
> One Worker on your zone.

> **[Amazon CloudFront](/dynamic-bot-rendering/cloudfront)**
>
> One CloudFront Function, without Lambda\@Edge.

> **[Vercel](/dynamic-bot-rendering/vercel)**
>
> One routing rule in `vercel.json`.

Running something else, such as Akamai, Fastly, or a custom proxy? The underlying contract is three request headers on a `GET`. Contact Profound support and we'll help you map it.


# Vercel setup
Source: https://docs.tryprofound.com/dynamic-bot-rendering/vercel

Route AI assistant traffic to Profound Dynamic Bot Rendering with a single routing rule in vercel.json, with no changes to your application code.

This guide adds one routing rule to your Vercel project. The rule matches requests by `User-Agent`, excludes Profound's own loop-protected fetches, and rewrites matching requests to Profound with your API key and public host attached as request headers.

Nothing in your application changes. Vercel's edge network evaluates the rule before your functions run, and requests from anything other than a supported assistant never match it.

Requests are rewritten to `https://concierge.tryprofound.com/v1/concierge/vercel/{path}`. Profound reconstructs the original path and query from the rewritten URL, so there is no `x-concierge-url` header to maintain. Which paths are rendered is decided by your [policy](/dynamic-bot-rendering/overview#how-rendering-and-serving-are-controlled) in Profound, not by the rule: an assistant request for a path you have not opted in is passed straight through to your deployment.

> [!WARNING]
>
> **Define the rule in `vercel.json` or `vercel.config.ts`, not in the dashboard.** Confirm that the deployed transform expands the environment variable rather than sending `$PROFOUND_CONCIERGE_KEY` as literal text. A literal value is rejected with `401` and `x-concierge-auth: unknown-key`.
>
> Before production, validate environment expansion and route precedence in an isolated deployment using a nonsecret test variable and a diagnostic endpoint you control. Do not send your API key to a diagnostic endpoint or log it. Confirm that existing dashboard and framework routes do not override the tested rule.

> [!WARNING]
>
> **Vercel's routing integration does not provide CDN-level failover.** Profound falls back to your origin only after a request is authenticated and valid. A Profound outage or a hard rejection can reach the assistant as an error, because this external rewrite has no secondary origin to retry.

## Prerequisites

* A Vercel project you can deploy
* Your Dynamic Bot Rendering API key, a registered domain, a policy that opts your public pages in, and a known published test page. See [Before you start](/dynamic-bot-rendering/overview#before-you-start).
* Confirm that deployment protection, firewall rules, and bot challenges permit assistant traffic and Profound's origin fetches without weakening access controls.

## Setup

**1. Add the API key to your project environment**

In the Vercel dashboard, open **Project → Settings → Environment Variables** and add:

| Name | Value | Environments |
| - | - | - |
| `PROFOUND_CONCIERGE_KEY` | Your Dynamic Bot Rendering API key | Production only |

Keep the key in the environment, never in the rule file. The rule file is committed to your repository.

**2. Add the routing rule**

Add the following route to your project configuration, replacing `www.example.com` with your registered public host. Use whichever file your project already has.

```json vercel.json theme={null}
{
  "routes": [
    {
      "src": "^/(.*)$",
      "has": [
        {
          "type": "header",
          "key": "User-Agent",
          "value": "(?i).*(?:duckassistbot|chatgpt-user|gemini-deep-research|perplexity-user|amzn-user|mistralai-user|claude-user|claude-code|codex).*"
        }
      ],
      "missing": [
        {
          "type": "header",
          "key": "x-concierge-request",
          "value": "^1$"
        }
      ],
      "dest": "https://concierge.tryprofound.com/v1/concierge/vercel/$1",
      "transforms": [
        {
          "type": "request.headers",
          "op": "set",
          "target": {
            "key": "x-concierge-api-key"
          },
          "args": "$PROFOUND_CONCIERGE_KEY",
          "env": [
            "PROFOUND_CONCIERGE_KEY"
          ]
        },
        {
          "type": "request.headers",
          "op": "set",
          "target": {
            "key": "x-concierge-host"
          },
          "args": "www.example.com"
        }
      ]
    }
  ]
}
```

```ts vercel.config.ts theme={null}
import type { VercelConfig } from '@vercel/config/v1';

export const config: VercelConfig = {
  routes: [
    // Route recognized assistants to Profound unless x-concierge-request is set to 1
    {
      src: '^/(.*)$',
      has: [
        {
          type: 'header',
          key: 'User-Agent',
          value:
            '(?i).*(?:duckassistbot|chatgpt-user|gemini-deep-research|perplexity-user|amzn-user|mistralai-user|claude-user|claude-code|codex).*',
        },
      ],
      missing: [
        {
          type: 'header',
          key: 'x-concierge-request',
          value: '^1$',
        },
      ],
      dest: 'https://concierge.tryprofound.com/v1/concierge/vercel/$1',
      transforms: [
        {
          type: 'request.headers',
          op: 'set',
          target: {
            key: 'x-concierge-api-key',
          },
          args: '$PROFOUND_CONCIERGE_KEY',
          env: ['PROFOUND_CONCIERGE_KEY'],
        },
        {
          type: 'request.headers',
          op: 'set',
          target: {
            key: 'x-concierge-host',
          },
          args: 'www.example.com',
        },
      ],
    },
  ],
};
```

> [!WARNING]
>
> Preserve existing routing, redirects, authentication, and framework behavior. Check configuration validation and generated build output for route order and compatibility with existing properties. Current [Vercel routing documentation](https://vercel.com/docs/project-configuration/vercel-json#routes-vs-higher-level-properties) describes coexistence with higher-level properties; do not automatically convert your routing configuration. If integration requires a routing redesign, stop and review that separately.

**3. Deploy and verify**

Deploy the project, then:

```bash theme={null}
curl -sD - -o /dev/null https://www.example.com/pricing -A 'chatgpt-user'
```

Check the HTTP status, a valid `x-concierge-request-id`, and the expected `x-concierge-bot-kind`. On your known published page, also require `x-concierge-cache: HIT` and the expected rendered body. Verify that browser requests and search crawlers keep their existing behavior, and that a request carrying `x-concierge-request: 1` reaches your deployment directly.

Send a `GET` request. Profound's contract covers `GET` only, so use `curl -sD - -o /dev/null` rather than `curl -I`. Full checks are in [Verify and troubleshoot](/dynamic-bot-rendering/verify-and-troubleshoot).

## What each part of the rule does

| Part | Purpose |
| - | - |
| `src` + `dest` | Captures the whole path as `$1` and rewrites it onto Profound's Vercel endpoint, preserving the query string. Profound decides per path whether to serve a render or pass the request through to your deployment. |
| `has` on `User-Agent` | Matches the [supported assistants](/dynamic-bot-rendering/overview#supported-ai-assistants), case-insensitively. Browsers and search crawlers never match, so they never leave your existing route. |
| `missing` on `x-concierge-request` | Loop protection. Profound stamps `x-concierge-request: 1` on its own origin fetches. Excluding that value lets those fetches bypass the rule and reach your deployment directly. |
| `x-concierge-api-key` transform | Injects the key from the project environment at request time. |
| `x-concierge-host` transform | The public host, hard-coded per route. It must match a domain registered in Profound. |

> [!WARNING]
>
> Keep the `missing` condition's `value` exactly `^1$`, or omit `value` to test for the header's absence. Any other regular expression that doesn't match the stamped value silently makes the exclusion a no-op. Profound's own fail-open fetches are then rewritten back to Profound and rejected with `400`.

Non-`GET` requests from a matching `User-Agent` are also rewritten and receive `405` from Profound. Supported assistants fetch pages with `GET`, so this does not affect normal traffic. Add `"methods": ["GET"]` to the route if you want other methods to stay on your deployment.

## Multiple domains

The `x-concierge-host` transform is a fixed string, so one rule serves one public host. If a single Vercel project serves several registered domains, add one route per host and narrow each with a `has` condition on the `host` header:

```json theme={null}
{
  "type": "header",
  "key": "host",
  "value": "^www\\.example\\.com$"
}
```

Place these routes before any catch-all route.

## Preview deployments

The rule above fires on every hostname the project serves, including preview deployments. Because `x-concierge-host` is fixed, a preview request would be identified to Profound as production. Scoping `PROFOUND_CONCIERGE_KEY` to Production only, as in step 1, means preview requests carry an unexpanded literal key and Profound rejects them with `401`, so they can never serve or render production content.

If you want assistant requests on preview hosts to reach your preview deployment normally instead, add the `host` condition from [Multiple domains](#multiple-domains) so the rule only matches your production hostname. See [environment variables in routes](https://vercel.com/docs/project-configuration/vercel-json#using-environment-variables-in-routes) for how unset variables are handled.

## Cache and content checks

Confirm deployed rewrite cache behavior before enabling traffic. [Vercel honors upstream cache headers](https://vercel.com/docs/routing/rewrites#caching-rewrites-to-external-origins) by default for projects created on or after April 6, 2026. Older projects can opt in through project settings or `x-vercel-enable-rewrite-caching`; that header set to `0` opts out. Check project age, the current setting, and route overrides rather than assuming a default. Do not assume a rendered response has `Cache-Control: no-store`: the Vercel adapter does not guarantee it. If your cache rules make these responses cacheable, verify separation between assistant and human responses and between host, path, and query variants. If caching is enabled, verify expiration and invalidation after publishing a replacement, unpublishing, or pausing serving: a CDN entry can outlive the change in Profound.

Profound does not forward viewer cookies or authorization to your origin, and stored pages are not personalized by session, language, or geography. Review same-URL locale and header variants before opting a page in.

## Security notes

* **The loop marker is client-spoofable by design.** Anyone sending `x-concierge-request: 1` skips the rewrite and gets your live origin page, the same content a non-assistant `User-Agent` receives. It skips rendering and telemetry for that request. Never use this marker to bypass authentication or firewall checks.
* **The key stays out of the repository.** It's referenced from the project environment, never written into the rule file.
* **Profound re-derives the assistant itself.** It never trusts a caller-supplied identity header, only the forwarded `User-Agent`.
* **Credentials never reach your origin through Profound.** The rule forwards whatever headers the assistant sent, but Profound drops cookies and authorization headers when it fetches your origin, and only opted-in public pages are rendered.


# Verify and troubleshoot
Source: https://docs.tryprofound.com/dynamic-bot-rendering/verify-and-troubleshoot

Confirm Dynamic Bot Rendering is routing correctly, read the x-concierge-* diagnostic headers, and resolve the common setup failures.

## Test your setup

Every check below sends a `GET` request. Profound's contract covers `GET` only, so use `curl -sD - -o /dev/null` rather than `curl -I`, which sends `HEAD`. A `HEAD` response may come from your origin or from a handler that omits the body, and tells you nothing about routing.

**1. Assistant traffic should reach Profound**

```bash theme={null}
curl -sD - -o /dev/null https://www.example.com/pricing -A 'chatgpt-user'
```

Look for a UUID in `x-concierge-request-id` and `x-concierge-bot-kind: chatgpt-user`. Together they prove your CDN routed the request to Profound and Profound recognized the assistant. They do not prove a rendered page was served: a `MISS` with `x-concierge-failopen: 1` means Profound returned your origin's live content.

**2. Your published test page should be served from Profound**

On the page you published during [setup](/dynamic-bot-rendering/overview#before-you-start), also require `x-concierge-cache: HIT` and the rendered HTML. Save the body for comparison, and let curl decompress it in case Profound served it gzip-encoded:

```bash theme={null}
curl -sS --compressed -D /tmp/dbr-headers.txt -o /tmp/dbr-page.html \
  https://www.example.com/pricing -A 'chatgpt-user'
```

If this request is a `MISS`, check the `x-concierge-policy` headers as described in [Reading the result](#reading-the-result). Repeating the request does not turn a `MISS` into a `HIT`; only publishing a render does.

**3. Human traffic should be untouched**

```bash theme={null}
curl -sD - -o /dev/null https://www.example.com/pricing \
  -A 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36'
```

Expect no `x-concierge-*` headers and the same page content as before. Compare latency and cache hit rates with your baseline too: a Worker or a shared cache-policy change can affect them even when routing is unchanged.

**4. Search crawlers should be untouched**

```bash theme={null}
curl -sD - -o /dev/null https://www.example.com/pricing -A 'Googlebot/2.1'
```

Also expect no `x-concierge-*` headers. If this request is routed, see [Common problems](#common-problems).

**5. Excluded traffic should keep its existing behavior**

In staging, send non-`GET` methods, requests to `/api/users`, `/graphql`, authentication, admin, framework resources, and assets, including extensionless endpoints. Use harmless fixture endpoints for methods that could modify data, and nonsecret test values for cookies and authorization headers. All of these must stay on your existing route. On Vercel, also test preview and alternate hosts.

**6. Test failure behavior safely**

Use staging or an isolated configuration, record the original settings, and restore them after each test.

* **Cloudflare:** test an unreachable upstream, a response-header timeout, and one of the configured fallback statuses. Expect origin content with `x-concierge-cdn-failover: 1`. Redirects, `404`, and `429` pass through, and a response that has already started streaming cannot be replaced.
* **CloudFront:** test primary connection and response timeouts within the configured budget and the selected status criteria. Expect the fallback origin for configured failures, but not for `401`. Send uncached requests, and expect cached fallback responses to persist until their TTL expires even after you restore the primary.
* **Vercel:** there is no CDN-level failover. During a Profound outage or after a hard rejection, an assistant can receive an error. Do not expect transparent origin fallback.

After restoring the configuration, wait for propagation and repeat the routing and published-page checks.

## Run setup verification from the dashboard

The Profound dashboard can test your routing for you. It sends a request with each of the nine supported assistant `User-Agent` values to a URL on your registered host, from outside your network, and reports which ones your CDN routed to Profound.

These probes carry no credentials and no loop marker, so they look like real assistant traffic to your CDN. An assistant counts as routed only when its response carries a valid request ID and the matching bot kind. A `MISS`, a paused domain, a redirect, or a non-`200` status can still confirm routing; page health is reported separately. Make sure the URL you test is on your public-page allowlist, or every assistant will correctly show as not routed.

> [!NOTE]
>
> Setup verification is a live observation of one URL at one moment. It confirms that your CDN rule fires for each assistant. It does not verify domain ownership, and it is not a substitute for checking your published test page by hand.

## Response header reference

| Header | Meaning |
| - | - |
| `x-concierge-request-id` | A UUID Profound assigns to every request it handles. Its absence means Profound never saw the request: the rule did not match, the CDN failed over, or a proxy stripped the header. |
| `x-concierge-cache` | `HIT` when a published render was served, `MISS` when Profound fell back to your origin. |
| `x-concierge-bot-kind` | The assistant Profound recognized, or `unknown`. |
| `x-concierge-failopen` | `1` when live origin content was returned instead of a stored render. |
| `x-concierge-origin-status` | Your origin's HTTP status, when your origin was fetched. |
| `x-concierge-policy` | The effective mode for this request: `serve`, `warm`, or `off`. See [policy modes](/dynamic-bot-rendering/overview#policy-modes). |
| `x-concierge-policy-source` | Which level decided that mode: `page` for a page rule, `domain` for the domain default or a paused domain, or `default` when no rule applied. |
| `x-concierge-policy-rule` | The ID of the winning rule, when there was one. |
| `x-concierge-auth` | On `401` only: `missing-key`, `unknown-key`, or `inactive-key`. |
| `x-concierge-cdn-failover` | Set by the Cloudflare Worker in these guides, not by Profound, when it served your origin after a failed upstream call. |

## Reading the result

| What you see | What it means |
| - | - |
| `cache: HIT` | A published render was served. Also check the status and the body. |
| `MISS` + `failopen: 1` + `policy: serve` | Policy allows serving, but the URL has no usable publication. Render, review, and publish a candidate for this exact path and query. |
| `MISS` + `failopen: 1` + `policy: warm` | `warm` serves origin content even when a publication exists. Change the winning rule to `serve` once you are ready. |
| `MISS` + `policy: off` + `policy-source: default` | No rule covers this path, so it is not opted in. Add a domain default or a page rule. |
| `MISS` + `policy: off` + `policy-source: domain` + no rule ID | The domain is paused. Resume it in the dashboard. |
| `MISS` + `policy: off` + a rule ID | The winning rule is `off`, or it excludes this assistant from its allowed list. With `policy-source: domain` that rule is the domain default. Check its mode and allowed assistants. |
| `bot-kind: unknown` | Your CDN routed a `User-Agent` Profound doesn't recognize. The request added a round trip and got origin content. Align your rule with the [supported list](/dynamic-bot-rendering/overview#supported-ai-assistants). |
| No `x-concierge-*` headers on an assistant request | Profound never saw the request. Check the rule's match conditions and, on Cloudflare and CloudFront, whether CDN failover fired. |

## Error responses

Profound only falls back to your origin after a request is authenticated and valid. Configuration errors are rejected rather than proxied. Cloudflare's Worker masks its configured rejection statuses by serving your origin, and CloudFront masks `400` and `403` but not `401`. Vercel has no retry. When an origin page appears where you expected a rejection, inspect your CDN's diagnostics.

| Status | Cause | Fix |
| - | - | - |
| `401` + `x-concierge-auth: missing-key` | No `x-concierge-api-key` header arrived. | Confirm that the rule attaches the key. Never log or echo a real key while diagnosing. |
| `401` + `unknown-key` | A value arrived but matches no key. | Check for truncation, whitespace, or a literal `$PROFOUND_CONCIERGE_KEY` that was never expanded. On Vercel, validate variable expansion with a nonsecret test value before using the real key. |
| `401` + `inactive-key` | The key was revoked. | Create a new key and roll it out. |
| `403` | The host isn't registered, or belongs to another organization. | Register the exact public host in the dashboard. It must match after normalization: lowercase, no trailing dot, no port. |
| `400` | A malformed host or path, an absolute URL where a path was expected, a `#` fragment, or an inbound loop marker. | Usually a rule sending a full URL instead of a path, or a broken loop-protection condition. |
| `405` | A non-`GET` method reached Profound. | Route only `GET` to Profound, and test with `curl -sD - -o /dev/null` rather than `curl -I`. |
| `502` | Profound could not reach your origin and had no publication to serve, or a backend lookup failed. | Check that your origin is reachable from the public internet and not blocking Profound's fetches. A CDN can also generate its own `502`, so check its diagnostics too. |

## Common problems

> **Googlebot or other search crawlers are being routed**
>
> Your `User-Agent` pattern matches something outside the supported token list. Compare it with the pattern in your setup guide and the [supported assistants](/dynamic-bot-rendering/overview#supported-ai-assistants). Search crawlers such as `Googlebot` must keep reaching your origin.

> **Every assistant request shows bot-kind: unknown (CloudFront)**
>
> CloudFront replaces the viewer `User-Agent` with `Amazon CloudFront` on origin requests unless an origin request policy forwards it. Profound then has nothing to classify. Use the narrow custom origin request policy in the [CloudFront guide](/dynamic-bot-rendering/cloudfront). Avoid **AllViewer**, which also forwards the viewer `Host` and can break the connection to Profound. Add `User-Agent` to the origin request policy, not the cache key.

> **One assistant gets origin content while others get the render**
>
> The winning rule's allowed-assistant list excludes it. The response shows `x-concierge-policy: off` together with the rule ID. Add the assistant to the list, or clear the list so the rule inherits the domain default.

> **Assistants get a cached human page, or humans get a rendered page**
>
> Some cache layer stores HTML without separating assistant and human responses, or without separating hosts, paths, and query strings.
>
> On CloudFront, include `x-concierge-route` in the cache key, keep Minimum TTL at `0`, and check host and query variants. On Cloudflare, verify that the Worker's upstream call bypasses cache; `User-Agent` keying alone cannot separate page identities because every page uses the same upstream URL. On Vercel, inspect the project's external-rewrite cache settings. On every provider, also check what happens to cached entries after you publish a replacement, unpublish, or pause: a CDN entry can outlive the change in Profound.

> **Requests loop, or return 400 with a valid-looking rule**
>
> Loop protection is failing. Your rule must let requests carrying `x-concierge-request` pass straight through to your origin. Profound stamps that header on its own origin fetches, and rejects any inbound request that carries it with `400`, so a broken exclusion turns into one failed request rather than a loop.
>
> On Vercel, this is a `missing` condition on `x-concierge-request` with the value `^1$`, the marker Profound stamps. Any other regex silently disables the exclusion, because a `missing` condition with a value only excludes requests whose header matches it.

> **Query-string variants serve the wrong content**
>
> Query strings are part of the page identity. If your CDN strips them before the origin request, `/product?id=a` and `/product?id=b` collapse to `/product`, and the wrong variant is rendered and served.
>
> On CloudFront, include the query strings your pages vary on, or "All", in the cache policy. An origin request policy can forward extra query strings without keying them, so verify forwarding and cache identity separately.

> **Nothing is ever served from cache, even after many requests**
>
> Read `x-concierge-policy` and `x-concierge-policy-source` on the response. No matching rule means `off`, and `warm` serves live origin content even when a publication exists. Set the winning rule to `serve`.
>
> Then confirm that a candidate for this exact path and query was reviewed and published. Approval alone does not publish it, and repeated requests do not publish it either. A candidate held as pending stays pending until someone approves, rejects, or publishes it. If the page is published and policy is `serve` but you still see a `MISS`, check your CDN caches and your origin's reachability.

> **Requests are challenged, blocked, or return login content**
>
> Check WAF rules, bot challenges, deployment protection, and origin allowlists for both incoming assistant traffic and Profound's origin fetches. Profound does not forward viewer cookies or authorization headers, so protected or session-sensitive pages belong outside this setup. The loop marker is spoofable and must not be used as a firewall or authentication bypass. If your origin needs a way to identify Profound's fetches, contact Profound support.

## Turning it off

To stop serving renders without touching your CDN, pause the domain in the Profound dashboard. Every assistant request goes to your origin while your rules, renders, and publications are preserved, and resuming restores them. A pause takes effect in Profound immediately, but responses your CDN has already cached persist until they expire. A pause is not a substitute for CDN-level failover during an outage, because the request still has to reach Profound.

To remove the integration, restore the CDN configuration you recorded before setup:

* **Cloudflare:** remove the added Worker route or restore its previous binding without removing other Worker logic.
* **CloudFront:** restore the previous viewer-request association and policies. Use **No association** only when there was none before.
* **Vercel:** remove only the added route from `vercel.json`, preserve other routing, and redeploy.

Routing changes take effect after configuration or deployment propagation. Previously cached fallback responses may persist until expiry or your normal invalidation process. Verify original routing and content after rollback. Registered domains, keys, and publications remain in Profound until you remove them.


# Categories in Profound
Source: https://docs.tryprofound.com/getting-started/categories

How categories group the prompts, topics, assets, and regions in Answer Engine Insights

> [!TIP]
>
> Find definitions of unfamiliar terms and concepts in [Profound glossary](/getting-started/glossary).

Categories organize your [Answer Engine Insights (AEI)](/answer-engine-insights/about) data. Each category groups the prompts, topics, assets (your own brands and competitors), and regions you track for a market, business unit, brand, or geography.

Categories keep markets separate. For example, a bank that competes in credit cards, mortgages, and wealth management faces different competitors and questions in each market, so each gets its own category. Profound runs the prompts and reports the results inside that boundary, so every AI search data point compares you against the right competitors on the right questions.

## Category structure and how it relates to other Profound building blocks

![Diagram of the hierarchy of Profound categories, including their relationship to organizations and a Profound account.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/categories-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=60be3ca5b18a94ccf55bcc33de5d41d5)

**Your Profound account** holds your total prompt allowance and the regions, languages, and platforms it can monitor. Categories draw from that pool: each one takes a share of the prompts and a subset of the regions, languages, and platforms.

**An organization** is a bucket for all of your Profound data, including [AEI](/answer-engine-insights/about), [Agents](/agents/about), [Sheets](/sheets/getting-started), [brand context](/context-manager/about), [Aim](/aim/about) conversations, and [Documents](/documents/about). An organization uses categories to organize its AEI data.

**A category** is the market unit you analyze, for example "financial services" or "home decor products." Each category has a name, a share of the account's prompts, and the regions, languages, and platforms it runs in, chosen from those available to your account. An organization can own and use multiple categories.

**Assets** are the brands or products you track in a category. An asset is either owned (yours) or a competitor. Every category needs at least one enabled owned asset. Competitors default to a limit of five per category.

**Topics** are the subject areas your prompts cover, such as "Purchase decisions" or "Travel rewards." Every prompt belongs to one topic, so topics are how analytics reports group prompts. The Prompt generator uses topics to generate relevant prompts.

**Prompts** are the questions Profound sends to answer engines. A prompt lives in one category and one topic, and runs in one language. Prompts have [types](/getting-started/glossary#analysis-type) that define which analysis Profound runs on the responses: Visibility, [FactCheck](/answer-engine-insights/factcheck/about), or [Sentiment](/answer-engine-insights/sentiment/about).

**Tags** are optional labels you add to prompts inside a category to organize them in a way the topics don't cover, for example by campaign or buying stage. A prompt can carry multiple tags.

**Personas** are the audience segments you define for a category. Learn more in [Personas in Answer Engine Insights](/answer-engine-insights/personas).

**Regions**, **languages**, and **platforms** define where prompts run, in which language, and on which answer engines. View your organization's lists in [Analytics Setup](/platform-config/analytics-setup).

> [!TIP]
>
> Learn more about prompts, topics and tags in [How to create topics and prompts to track in Profound](/answer-engine-insights/how-tos/create-topics-and-prompts).

### When you have multiple organizations and categories

If you work with more than one organization or category, here's how they relate:

![Example diagram showing categories owned, used, and shared across multiple organizations.](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/answer-engine-insights/categories-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=a893343a7fca86c2b811053f45d291ee)

* An organization can create, own, and share categories with other organizations.
* Your organization can use a shared category, but only the owning organization can edit it.
* Your Profound contract sets the number of categories your organization can use.

## Where you work with categories

* **Answer Engine Insights (AEI):** Every category-scoped page has a selector at the very top of the page that shows the owning organization, the categories you can view, and the owned assets in the selected category. Filters for topics, tags, regions, platforms, and personas apply within that category.

  Learn more in [Interpret Answer Engine Insights v2](/answer-engine-insights/interpret).

> [!TIP]
>
> [AEI](/answer-engine-insights/about) reports on one category at a time. When you need the bigger picture, use [Aim](/aim/about) and [Dashboards](/custom-dashboards/about) to combine analytics data across your categories.

* **Analytics Setup:** Admins of enterprise organizations use [Analytics Setup](/platform-config/analytics-setup) to create and manage categories, allocate prompts, choose regions and languages, and add assets.

* **People.** When you invite a member on the **People** page, choose the categories they can open in the **Category** field. Learn more in [Add and manage organization members](/platform-config/people/add-and-manage-org-members).

* **Onboarding.** New users set up their first category in the onboarding flow: confirm your brand, add owned and competitor assets, review generated topics, and generate prompts.

* **Profound API and MCP.** Reports provided by the API or MCP tools are scoped by category. Learn more in the [REST API reference](/rest-api/introduction) and [MCP documentation.](/mcp/capabilities/analytics-capabilities)


# Profound usage pricing: Credits
Source: https://docs.tryprofound.com/getting-started/credits

How AI Marketer (Aim) conversations and Agent runs consume credits, what your plan includes, and what happens when credits run out

## What credits are

Credits are the unit Profound uses to measure the AI work your organization runs: conversations with [AI Marketer (Aim)](/aim/about), [Agent](/agents/about) runs, and [Sheets](/sheets/getting-started). Each Profound plan includes a credit allotment, and every piece of AI work draws from that shared balance.

The cost of a piece of work depends on how much it does. A short question to Aim about last week's visibility uses fewer credits than a message that researches a topic, generates images, or drafts an article. A Workflow with a few steps uses fewer credits than an Agent that calls several AI models and reasons through a complex task.

## What uses credits

* **AI Marketer conversations**: every message you send to Aim consumes credits, whether you chat in the platform, inside a [Project](/projects/about), a [Profound Doc](/documents/about), a [Sheet](/sheets/getting-started), or in [Slack](/integrations/slack/connect-slack-to-profound). The cost of each message reflects the work Aim does to answer it, including any tools it calls and images it generates. When Aim runs one of your Agents, that run is billed as an Agent run.
* **Agent and Workflow runs**: every run consumes credits, including test runs in the Agent and Workflow Builders, scheduled runs, runs started from Aim, and runs started through [Profound MCP](/mcp/overview).
* **Sheets runs**: a [Sheet](/sheets/run-workflows-from-sheets) runs a Workflow once per row, and each row is billed as a run.
* **Prompt Research Reports**: generating a [Prompt Research Report](/prompt-volumes/about#prompt-research-reports) consumes credits.

Browsing your Answer Engine Insights, Prompt Volumes, and Agent Analytics data doesn't consume credits.

## Plans and credit allotments

| Plan | Credits |
| - | - |
| Trial | A limited, one-time allotment of AI Marketer credits. Trial credits don't renew. |
| Enterprise | A credit package sized for your team by the Profound accounts team. Packages can be adjusted over time as your needs change. |
| Agency Growth | 400 credits per month per client workspace. |

To size or adjust your package, [contact the Profound accounts team](https://www.tryprofound.com/contact-sales).

> [!NOTE]
>
> Pitch workspace credits in [Agency Mode](/agencies/agency-mode-overview) are different from the usage credits described on this page. They are a separate allowance for creating and extending pitch workspaces.

## Estimate and track credit usage

Admins can check the organization's credit balance and usage in the organization settings. The estimated cost of a Workflow run also shows before you start it:

* **In the Workflow Builder**: hover over **Run test** to see the **Estimated usage** for the whole Workflow, or over a node's **Test node** button to see the estimate for that node. After a test, **Run details** shows the **Credits used**.
* **On a Workflow's run page**: **Estimated usage** appears before you select **Run**, and **Credits used** replaces it once the run completes.
* **In Sheets**: the Workflow preview shows the estimated cost to run one row before you run the Sheet. Final credit usage may vary depending on the input and generated output.

Enterprise admins can also set a credit limit per member, so one heavy user can't consume the organization's whole balance. A member who reaches their limit can't run Workflows, Sheets, or Prompt Research Reports until their allocation is increased or the organization balance renews. Their usage still counts toward the organization total.

## When you run low or run out

Profound warns you before work stops:

* **Before a costly Workflow run**: if the estimated cost is higher than the credits you have left, a dialog warns that **This run may exceed your credit limit** and shows the projected overage. Confirm to continue, or select **Cancel**. Sheets runs estimated at 500 credits or more ask you to confirm the high-cost run even when the balance covers it.
* **Overage**: depending on how your account is configured, your organization either keeps running on overage credits after the allotment is used, or pauses until the balance renews. While you're on overage credits, Workflow pages show a **Plan limit reached** banner. Ask your account team which setting applies to you.
* **Out of credits**: Workflow pages show **Agent runs disabled** and **Run test** is unavailable. Work resumes when the balance renews or your account team adds credits.
* **Per-user limit reached**: only that member is blocked. The rest of the organization keeps running.
* **Trial**: when trial credits run out, Profound asks you to update your plan to Enterprise.

## Do API calls consume credits?

No. Pulling data from the [Profound API](/rest-api/introduction) doesn't consume credits. Credits are only charged where AI work runs: an Aim conversation, an Agent or Workflow run, a Sheet run, or a Prompt Research Report, including runs you start through [Profound MCP](/mcp/overview).


# Glossary
Source: https://docs.tryprofound.com/getting-started/glossary

Definitions of the metrics, objects, and features you'll find in Profound

This glossary defines the metrics, objects, and features you'll find in Profound. For definitions of more general AEO/AI search terms, see our [AI Search Glossary](https://www.tryprofound.com/blog/ai-search-glossary).

> [!TIP]
>
> Search the terms you're interested in with Ctrl+F or Cmd+F.

## A

### Accuracy prompt

A prompt that checks whether what AI says about your product is factually correct: pricing, features, integrations, specs, and anything else with a right answer. In Profound, accuracy prompts run with [FactCheck](/answer-engine-insights/factcheck/about).

Learn more: [Six types of prompts in Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts), [About FactCheck](/answer-engine-insights/factcheck/about)

### Accuracy Score

A [FactCheck](/answer-engine-insights/factcheck/about) metric that shows how much of what answer engines say about your brand is correct. Profound extracts the claims made in AI responses about your brand, checks each one against your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits), and reports the share that is accurate. The score is calculated as accurate claims / (accurate claims + inaccurate claims).

For example, FactCheck extracts 50 claims. 30 of those can be verified against the Knowledge Base, of which 24 are accurate and 6 are inaccurate. The accuracy score is 24 / (24 + 6) = 80%. The 20 unverifiable claims don't affect the score.

Learn more: [About FactCheck](/answer-engine-insights/factcheck/about)

### Activity Logs

A single audit trail of events in your Profound organization, such as sign-ins, permission changes, and configuration updates. Admins use Activity Logs to review who did what and when. Available under Platform Management.

Learn more: [About Activity Logs](/platform-config/activity-logs/about), [Activity logs platform documentation](/platform-config/activity-logs/about)

### Admin

A Profound [role](/platform-config/people/roles-and-permissions) with full access to organization settings, including team management, [Prompt Designer](/answer-engine-insights/prompt-designer) access, and usage reporting. Some actions, such as modifying prompts or creating [Agency Mode](/agencies/agency-mode-overview) workspaces, require the Admin role.

Learn more: [Roles and permissions](/platform-config/people/roles-and-permissions)

### AEO Content Scorecard

A content optimization node for [Workflows](/agents/workflows/about-workflows) that scores a piece of content against AEO best practices and returns recommendations.

Learn more: [AEO Content Scorecard](/agents/workflows/nodes/content-optimization/aeo-content-scorecard)

### Agency Mode

A mode for agencies and consultancies that manage several brands under one account. Each brand gets its own workspace with independent prompts, competitors, tracked platforms, and configuration. Agency Mode supports short-term pitch workspaces and long-term client workspaces, both managed from the Agency Mode dashboard.

Learn more: [Agency Mode overview](/agencies/agency-mode-overview)

### Agent

An automation that [AI Marketer (Aim)](/aim/about) builds and runs for you from a plain-English description. You tell it the job, connect the tools it needs, and choose when it runs: on demand, on a schedule, or when something happens in an app, on the web, or in your Profound data. On each run, Aim works out the steps itself, so the wording and the steps can differ between runs within the guardrails you set. For a step that must run the same way every time, an Agent can run a [Workflow](/getting-started/glossary#workflow).

Learn more: [About Agents](/agents/about)

### Agent Analytics

Infrastructure-level monitoring of how AI crawlers and assistants access your website. Profound collects request data at the CDN or server layer, classifies the AI traffic it finds, and reports on bot visits and AI referrals by page. Agent Analytics feeds the bot and referral data in [Pages](/pages/about) and powers [network benchmarking](/pages/about#network-benchmarking).

Learn more: [About Agent Analytics](/agent-analytics/overview)

### Agent Builder

The page where you create and refine an [Agent](/getting-started/glossary#agent). On the left is a chat with Aim: describe what the Agent should do, answer its questions, run the Agent, and ask for changes. On the right is the settings panel with the Agent's run triggers, inputs, instructions, outputs, tools, context, and advanced settings. Edits apply to the Agent right away, with no separate publish step.

Learn more: [Get started with Agents](/agents/getting-started), [Building blocks of an Agent](/agents/building-blocks-of-agent)

### AI Marketer

Your AI marketing teammate in Profound, also known as [Aim](/aim/about). It powers every chat surface in the platform, suggests [Projects](/projects/about) and tasks, and can use your [Agents](/agents/about) to do the work. It's grounded in the context you keep in [Context Manager](/context-manager/about).

Learn more: [About Profound's AI Marketer (Aim) ](/aim/about)

### AI referral

A visit to your website by a human who arrived from an AI answer, as opposed to a visit by an AI [bot](/agent-analytics/crawler-verification). Tracked by [Agent Analytics](/agent-analytics/overview).

Learn more: [About Agent Analytics](/agent-analytics/overview)

### Aim

The short name for [AI Marketer](/aim/about).

Learn more: [About Profound's AI Marketer (Aim) ](/aim/about)

### Analysis type

A setting on a prompt that controls which analyses Profound runs on its response: Visibility, [Sentiment](/answer-engine-insights/sentiment/about), or [FactCheck](/answer-engine-insights/factcheck/about). Every prompt produces visibility data, while other prompt types need to be assigned in the [Prompt Designer](/answer-engine-insights/prompt-designer). Also known as prompt type.

Learn more: [About Sentiment](/answer-engine-insights/sentiment/about)

### Answer engine

An AI service that answers a user's prompt in real time, such as ChatGPT, Gemini, Claude, Perplexity, or Google AI Overviews. Profound sends your prompts to answer engines every day and analyzes the responses. In Profound and in the API, a tracked answer engine is called a platform or model.

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Answer Engine Insights (AEI)

The Profound product that shows how your brand performs across answer engines. Profound runs your prompts daily, collects the responses, and turns them into metrics: Visibility Score, Share of Voice, Average Position, Citation Share, Citation Coverage, Sentiment Score, and Accuracy Score.

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Article Research Report

A research node for [Workflows](/agents/workflows/about-workflows) that compiles a research report for a planned article, which downstream nodes can use to write grounded content.

Learn more: [Article Research Report](/agents/workflows/nodes/research/article-research-report)

### Article Title Suggestion

A content creation node for [Workflows](/agents/workflows/about-workflows) that proposes article titles for a topic or brief.

Learn more: [Article Title Suggestion](/agents/workflows/nodes/content-creation/article-title-suggestion)

### Artifact

An output that an [AI Marketer (Aim)](/aim/about) chat or an [Agent](/agents/about) produces, such as a [Profound Document](/documents/about), a [Sheet](/sheets/getting-started), or an image. Artifacts from chats in a [Project](/projects/about) appear in the Project's **Artifacts** section, where every member can open them.

Learn more: [About Projects](/projects/about)

### Asset

A brand, company, or product that Profound tracks in a category. Your own brand is an owned asset; the others are competitors.

### Attribute

In [Sentiment](/answer-engine-insights/sentiment/about), a specific claim or pattern within a [theme](/answer-engine-insights/sentiment/about#themes-and-theme-attributes), with a positive or negative polarity. For example, "opaque pricing" is a negative attribute under a Pricing theme.

Learn more: [About Sentiment](/answer-engine-insights/sentiment/about)

### Average Position

The [average position](/answer-engine-insights/about#average-position) of your brand relative to competitors across responses. For example, McDonald's is listed first in half of the responses that mention it and second in the other half. Its Average Position is 1.5.

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

## B

### Blacklist

A mode in [Configure Competitors](/answer-engine-insights/settings#configure-competitors) that removes specific brands from all Answer Engine Insights analysis. Use it for brands that appear in your data but aren't relevant competitors, or that you aren't permitted to mention. Compare [Whitelist](/getting-started/glossary#whitelist).

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Bot

An automated agent that requests pages from your website. [Agent Analytics](/agent-analytics/overview) identifies and classifies AI bots by purpose: AI citation, AI indexing, AI training, and AI agent visits. Profound also operates [its own bots](/bots/introduction).

Learn more: [Crawler verification](/agent-analytics/crawler-verification), [Profound bots](/bots/introduction)

### Bot Visits

A view in [Pages](/pages/about) and [Agent Analytics](/agent-analytics/overview) that shows what AI bots see and can reach on your site: which crawlers visit a page, what content is hidden from them, and a `robots.txt` audit that lists the AI crawlers your site disallows. See also [Bot vs. Human Readable Content](/pages/about#bot-vs-human-readable-content).

Learn more: [About Pages](/pages/about)

### Bot vs. Human Readable Content

A [Pages](/pages/about) feature that compares what a person sees on a page with what an AI [bot](/agent-analytics/crawler-verification) can access. Content rendered with client-side JavaScript is often invisible to bots, so Pages flags hidden text blocks and headings and recommends how to expose them.

Learn more: [About Pages](/pages/about)

### Brand

The entity whose presence in AI answers you track. In Profound data, a brand is an asset. Your own brand is an owned asset, and competitors are the other assets in your category.

### Brand Context

The tab in [Context Manager](/context-manager/about) that holds your [knowledge records](/context-manager/about#knowledge-records-that-build-themselves) and identifies [knowledge gaps](/context-manager/about#identified-gaps).

Learn more: [About Context Manager](/context-manager/about)

### Brand Kit

A brand identity profile that defines how Profound should write on your behalf: brand information, writing style, and a writing sample. Profound generates a Brand Kit from your website URL, and you can edit it at any time. Where a [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) defines what to say, a Brand Kit defines how to say it. Part of [Context Manager](/context-manager/about).

Learn more: [Knowledge Bases and Brand Kits](/context-manager/knowledge-bases-and-brand-kits)

### Bulk Analysis

A [Prompt Volumes](/prompt-volumes/about) feature that compares several keywords at once and ranks them by prompt volume, so you can prioritize what to track.

Learn more: [About Prompt Volumes](/prompt-volumes/about)

### Bulk upload

Adding many prompts at once by uploading a CSV in the [Prompt Designer](/answer-engine-insights/prompt-designer).

Learn more: [Create, manage, and tag prompts in Prompt Designer](/answer-engine-insights/prompt-designer)

## C

### Call API

A node for [Workflows](/agents/workflows/about-workflows) that sends an HTTP request to any external API and returns the response for later nodes to use.

Learn more: [Call API](/agents/workflows/nodes/call-api)

### Category

The workspace that holds one brand and its competitive set. A category contains assets, topics, prompts, tags, regions, Personas, and platforms.

Learn more: [Categories in Profound](/getting-started/categories)

### Chart annotation

A note pinned to a date on a time-series line chart in [Answer Engine Insights](/answer-engine-insights/about) that records an event, such as a product launch, a content refresh, or a change to your prompts.

Learn more: [Chart annotations in AEI](/answer-engine-insights/chart-annotations)

### Citation

A reference or link to a specific web page, article, or resource inside an answer engine response. Profound records which domains and pages are cited, how often, and for which prompts, and classifies each source into a citation category.

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Citation Categories

Classification of the sources answer engines cite.

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Citation Coverage

A metric in [Answer Engine Insights](/answer-engine-insights/interpret). Where Citation Share measures the proportion of all citations that point to a domain, Citation Coverage measures how widely that domain is cited across responses. For example, `mcdonalds.com` is cited in 30 of 100 responses, so its Citation Coverage is 30%, even if it accounts for a smaller share of all citations.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Citation Decay

A table in the Citations tab of [Answer Engine Insights](/answer-engine-insights/interpret) that shows how long sources keep getting cited after they first appear.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Citation Domains

A data node for [Workflows](/agents/workflows/about-workflows) that returns the top cited domains for a set of prompts, topics, or the whole category.

Learn more: [Citation Domains Node](/agents/workflows/nodes/data/citation-domains)

### Citation Pages

A data node for [Workflows](/agents/workflows/about-workflows) that returns the most cited pages for a set of prompts, topics, or the whole category.

Learn more: [Citation Pages](/agents/workflows/nodes/data/citation-pages)

### Citation Rank

The rank of a domain or page when cited sources are sorted by [Citation Share](/answer-engine-insights/interpret#citation-share-and-citation-coverage). A lower rank means the source is cited more often than its peers. For example, mcdonalds.com has the third-highest Citation Share among all cited domains. Its Citation Rank is 3.

Learn more: [Interpret Answer Engine Insights (legacy)](/answer-engine-insights/interpret)

### Citation Share

The percentage of all citations that point to a given domain, page, or citation category. For example, if your owned domains receive 100 citations out of 1,000 total, your Citation Share is 10%.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Citation Share (Workflow node)

A data node for [Workflows](/agents/workflows/about-workflows) that returns [Citation Share](/answer-engine-insights/interpret#citation-share-and-citation-coverage) values for use inside a Workflow.

Learn more: [Citation Share Node](/agents/workflows/nodes/data/citation-share)

### Claim

A statement an answer engine makes about your brand. [FactCheck](/answer-engine-insights/factcheck/about) extracts claims, checks them against your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits), and classifies each as accurate, inaccurate, or not relevant. Subjective statements are routed to [Sentiment](/answer-engine-insights/sentiment/about) instead.

Learn more: [About FactCheck](/answer-engine-insights/factcheck/about)

### Client workspace

A long-running brand [workspace](/agencies/agency-mode-overview) in [Agency Mode](/agencies/agency-mode-overview) for an agency's active clients. Client workspaces last for the duration of the agency's contract and are typically larger than [pitch workspaces](/agencies/agency-mode-overview), with more prompts, platforms, regions, and competitors tracked.

Learn more: [Agency Mode overview](/agencies/agency-mode-overview)

### Command Center

A Platform Management view that shows your team's Profound adoption, hours saved with Agents, and peer benchmarks.

Learn more: [About Command Center](/platform-config/command-center/about)

### Competitor

A brand you monitor alongside your own. Profound tracks competitors' presence in AI responses so you can compare Visibility Score, Share of Voice, Citation Share, and Sentiment.

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Competitor prompt

A prompt that asks which brand AI picks when the question narrows to a head-to-head comparison, for example:

> "I'm looking for an email marketing platform with the best AI features. Would you recommend ActiveCampaign or HubSpot?"

Learn more: [Six types of prompts in Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts)

### Conditional

Branching logic in a [Workflow](/agents/workflows/about-workflows). A conditional routes the workflow down different paths based on the value of a variable, for example publishing content only when a quality check passes.

Learn more: [Add conditionals to Workflows](/agents/how-tos/add-conditionals-to-workflows)

### Connector

A connection between Profound and a tool in your marketing stack, such as a CMS, analytics platform, or messaging app. Connectors, also called tools, are what [Agents](/agents/about) and AI Marketer use, while [integrations](/integrations/overview) are what [Workflows](/agents/workflows/about-workflows) use. Connectors also power the sources that [Context Manager](/context-manager/about) learns from.

Learn more: [Integrations vs. connectors](/integrations/integrations-vs-connectors)

### Context Manager

The one place for everything Profound knows about your brand: positioning, audience, strategy, and how you like to work. It stores that context as knowledge records, keeps Personal Context, and houses [Knowledge Bases](/context-manager/knowledge-bases-and-brand-kits) and [Brand Kits](/context-manager/knowledge-bases-and-brand-kits). [AI Marketer](/aim/about), [Agents](/agents/about), and [Projects](/projects/about) all draw on it.

Learn more: [About Context Manager](/context-manager/about)

### Coverage topic

A broad [topic](/getting-started/glossary#topic) that tracks whether your brand is part of the conversation for a category, such as "historical data". Compare [Depth topic](/getting-started/glossary#depth-topic), which targets a specific opportunity.

### Create Content Brief

A content creation node for [Workflows](/agents/workflows/about-workflows) that drafts a content brief, typically from AEO data or research produced earlier in the workflow.

Learn more: [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief)

### Credits

The unit Profound uses to measure AI work in your organization. Credits are consumed by [AI Marketer](/aim/about) conversations and by [Workflow](/agents/workflows/about-workflows) runs, whether stand-alone, on a schedule, from [Sheets](/sheets/getting-started), as a step in an [Agent](/agents/about), or through [Profound MCP](/mcp/overview). While Agents are in public beta, Agent runs don't use credits.

Learn more: [Credits](/getting-started/credits)

### Custom Dashboards

See [Dashboards](/getting-started/glossary#dashboards).

Learn more: [About Dashboards](/custom-dashboards/about)

### Custom role

A [role](/platform-config/people/roles-and-permissions) with a permission set you define, assigned to organization members in place of the built-in roles.

Learn more: [Custom roles](/platform-config/people/custom-roles)

## D

### Dashboards

Configurable views of your Profound data. Pick charts and tables for metrics such as Visibility Score, Share of Voice, and Citation Share, apply filters, arrange the layout, and share the dashboard with your organization or with a link.

Learn more: [About Dashboards](/custom-dashboards/about)

### Depth topic

A focused [topic](/getting-started/glossary#topic) built to win a specific opportunity, such as "historical data for marketers". It answers the question "are we the default recommendation?" Compare [Coverage topic](/getting-started/glossary#coverage-topic), which tracks a whole category.

### Discovery prompt

A prompt that asks for a recommendation in your category without naming any brand, for example *"What are the best AEO tools for enterprise brands?"* Because nothing in the question points at your brand, discovery prompts give the cleanest read of which brands shape your category.

Learn more: [Six types of prompts in Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts)

### Domain

The website behind a citation, such as example.com, as opposed to the specific page cited. Profound ranks domains in Top Citation Domains and assigns each to a citation category.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Domain Segments

Groupings of website data in [Agent Analytics](/agent-analytics/overview) by sub-brand, product line, or region. A segment can split one domain into slices or combine several domains into one view, so each team sees only its part of the site while leadership sees the brand-wide roll-up.

Learn more: [About Agent Analytics](/agent-analytics/overview)

## E

### Earned Media

The citation category for sites whose primary purpose is publishing content: news outlets, magazines, trade publications, editorial review sites, affiliate and comparison sites, and personal blogs. It excludes corporate blogs, newsrooms, and lead-generation sites.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Entity Parser

The component that extracts entities (companies, products, people, product features, or publication sources) from answer engine responses. Your Profound team configures it during onboarding.

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Execution

One run of a prompt against one platform.

Learn more: [Create, manage, and tag prompts in Prompt Designer](/answer-engine-insights/prompt-designer)

## F

### FactCheck

Profound's accuracy layer, part of [Answer Engine Insights](/answer-engine-insights/interpret). FactCheck extracts the claims answer engines make about your brand, checks each against your [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits), clusters them by theme, and traces inaccurate claims back to the citation URLs driving them.

Learn more: [About FactCheck](/answer-engine-insights/factcheck/about)

## G

### Generate Article

A content creation node for [Workflows](/agents/workflows/about-workflows) that writes a full article from a brief, research, or instructions, optionally grounded in a [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) and styled with a [Brand Kit](/context-manager/knowledge-bases-and-brand-kits).

Learn more: [Generate Article](/agents/workflows/nodes/content-creation/generate-article)

### Get Sitemap

A node for [Workflows](/agents/workflows/about-workflows) that fetches a website's sitemap so you can iterate over its URLs. See also [Profound Sheets Sitemap Importer](/bots/sitemap-importer).

Learn more: [Get Sitemap](/agents/workflows/nodes/get-sitemap)

### Google Ads integration

A set of integration [nodes](/agents/workflows/about-workflows) that pull Google Ads data into [Workflows](/agents/workflows/about-workflows).

Learn more: [Google Ads integration and nodes](/integrations/google-ads/overview)

## H

### Half-life

In [Citation Decay](/answer-engine-insights/interpret#citation-decay), the number of days for a source's citations to fall to half of their peak and stay there. A short half-life means the page or domain lost ground quickly and is a candidate for a content refresh. Compare [Rise](/getting-started/glossary#rise).

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Human Review

A node for [Workflows](/agents/workflows/about-workflows) that pauses the workflow until a person reviews and approves the content before later nodes run.

Learn more: [Human Review](/agents/workflows/nodes/human-review)

## I

### Indexing

In [Pages](/pages/about), submitting a page to search engines so they pick up new or updated content sooner. Profound uses the IndexNow protocol, which requests priority indexing but doesn't guarantee it. In [Agent Analytics](/agent-analytics/overview), AI indexing is one of the [bot](/agent-analytics/crawler-verification) traffic types.

Learn more: [About Pages](/pages/about)

### Input

Information you provide each time you start an Agent or a Workflow: text, Profound data, files, or dropdown selections. In [Sheets](/sheets/getting-started), each row supplies the inputs for one [Workflow run](/agents/workflows/workflow-runs).

Learn more: [About Agents](/agents/about)

### Intent

In [Prompt Volumes](/prompt-volumes/about), the reason behind a user's prompt: informational, commercial, generative, or conversational.

Learn more: [About Prompt Volumes](/prompt-volumes/about)

### Iteration

A loop in a [Workflow](/agents/workflows/about-workflows). An iteration applies the same sequence of nodes to every item in a list, for example extracting pricing from each competitor site in turn.

Learn more: [Add iterations to Workflows](/agents/how-tos/add-iterations-to-workflows)

## K

### Keyword list

A named collection of related keywords in [Prompt Volumes](/prompt-volumes/about) for tracking a topic cluster over time.

Learn more: [Keyword searches and lists](/prompt-volumes/how-tos/keyword-searches-and-lists)

### Knowledge Base

A repository of documents, web pages, and other source material that Profound treats as the source of truth about your brand. [Context Manager](/context-manager/about) distills Knowledge Bases into [knowledge records](/context-manager/about), [FactCheck](/answer-engine-insights/factcheck/about) verifies [claims](/answer-engine-insights/factcheck/about) against them, and a Knowledge Base node in [Workflows](/agents/workflows/about-workflows) runs a semantic search over them so generated content rests on verified material.

Learn more: [Knowledge Bases and Brand Kits](/context-manager/knowledge-bases-and-brand-kits)

### Knowledge record

A document in [Context Manager](/context-manager/about) that Profound generates and maintains about one part of your business: company, audience, competitors, offerings, strategy, or execution.

Learn more: [About Context Manager](/context-manager/about)

## M

### Market perception prompt

A prompt that asks how AI frames your category rather than your brand, for example *"What is a digital asset management platform and how should I evaluate one?"* It's useful for checking whether the way AI explains the space favors capabilities you have or capabilities your competitors lead on.

Learn more: [Six types of prompts in Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts)

### Mention

A reference to a brand within an answer engine response. One response can mention the same brand several times.

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Model

See [platform](/getting-started/glossary#platform).

## N

### Name Matching

The [Answer Engine Insights](/answer-engine-insights/interpret) setting that groups the different names answer engines use for a brand, such as "Profound" and "Try Profound," under one asset.

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Network benchmarking

A [Pages](/pages/about) feature that scores a page's citations as a percentile against roughly 3 million pages tracked across the Profound network, from Poor to Great. Benchmark against all sites or a peer group. Requires [Agent Analytics](/agent-analytics/overview).

Learn more: [About Pages](/pages/about)

### Node

One step in a [Workflow](/agents/workflows/about-workflows). Each node performs a specific task: prompting an AI model, fetching a page, querying Profound data, creating content, and so on.

Learn more: [About Workflows](/agents/workflows/about-workflows), [Workflow nodes](/agents/workflows/about-workflows#workflow-components)

## O

### Organization

Your company's space, where everything you do in Profound lives. An organization is a way to organize your Profound data and access: one Profound login can have several organizations, each with its own members and their [roles](/platform-config/people/roles-and-permissions), [teams](/getting-started/glossary#team), [credits](/getting-started/credits), [categories](/getting-started/glossary#category) and resources, such as Agents, Dashboards, Documents, and Knowledge Bases.

Learn more: [How your Profound data and access are organized](/getting-started/organizations-and-teams)

## P

### Page

A specific URL cited by an answer engine, as distinct from its domain.

Learn more: [About Pages](/pages/about)

### Page Health

A [Pages](/pages/about) feature that reports whether a page loads, renders, and performs well enough for users, crawlers, and answer engines. It draws on Google Lighthouse and PageSpeed Insights and covers core web performance metrics such as Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift.

Learn more: [About Pages](/pages/about)

### Pages

The Profound product that shows how each of your web pages performs in AI search. Pages combines Citation Share and Citation Rank from [Answer Engine Insights](/answer-engine-insights/interpret) with bot visits and AI referrals from [Agent Analytics](/agent-analytics/overview), plus content scores, Page Health, Bot vs. Human Readable Content, Indexing, and network benchmarking, in one table.

Learn more: [About Pages](/pages/about)

### Perplexity Search

A node for [Workflows](/agents/workflows/about-workflows) that runs a live web search through Perplexity and returns the results for later nodes.

Learn more: [Perplexity Search](/agents/workflows/nodes/perplexity-search)

### Persona

An audience profile that Profound adopts when it runs a prompt, such as "small business owner" or "enterprise IT buyer."

Learn more: [Personas in Answer Engine Insights](/answer-engine-insights/personas)

### Personal Context

The part of [Context Manager](/context-manager/about) that keeps memories about how you work. Profound uses it to personalize your chats without mixing your preferences into the shared [Brand Context](/context-manager/about).

Learn more: [About Context Manager](/context-manager/about)

### Pitch workspace

A short-term brand workspace in [Agency Mode](/agencies/agency-mode-overview), used to assess a prospect's AI search visibility or prepare a pitch. Pitch workspaces expire after a defined period, typically 30 days.

Learn more: [Agency Mode overview](/agencies/agency-mode-overview)

### Platform

An answer engine that Profound tracks, such as ChatGPT, Gemini, Claude, Perplexity, or Google AI Overviews. Also called a model.

### Position

Where your brand ranks relative to competitors. Position aggregates performance indicators such as Share of Voice into a ranked score; a lower value means a higher rank. See also [Average Position](/getting-started/glossary#average-position).

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Product tile

The card-style component that displays one product when an answer engine enters Shopping mode. A tile carries an image, price, merchant links, and buy buttons, and can expand to show sections such as "What to know."

Learn more: [About Shopping](/shopping/about)

### Profound Documents

A built-in collaborative editor for storing and refining content inside Profound. Documents support rich text, real-time co-editing, and comments. Also known as Profound Docs.

Learn more: [Profound Documents](/documents/about)

### Profound Knowledge Base Fetcher

A Profound bot that reads a website's sitemap when you import the site into a [Knowledge Base](/context-manager/manage-knowledge-bases), and re-checks imported PDF files for changes during auto-sync.

Learn more: [Profound Knowledge Base Fetcher](/bots/knowledge-base-fetcher)

### Profound MCP

Profound's hosted Model Context Protocol server, which lets AI tools such as ChatGPT, Claude, or Cursor access Profound data.

Learn more: [Profound MCP overview](/mcp/overview)

### Profound Sheets Sitemap Importer

A Profound bot that fetches a website's sitemap.xml and robots.txt so you can import the sitemap URLs into a [Sheet](/sheets/getting-started).

Learn more: [Profound Sheets Sitemap Importer](/bots/sitemap-importer)

### ProfoundBot

A Profound bot that fetches a single public web page on demand, so an [Agent](/agents/about) or [Sheet](/sheets/getting-started) can read its content as Markdown or HTML.

Learn more: [ProfoundBot](/bots/profoundbot)

### Project

One workspace in [Projects](/projects/about). A Project holds the chats, instructions, context, members, and Artifacts for one initiative, such as a product launch or a sentiment fix.

Learn more: [About Projects](/projects/about)

### Projects

The Profound feature for team work that spans more than one task. Each Project is a shared workspace with its own instructions, context (files, [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) documents, and brand information from [Context Manager](/context-manager/about)), chats with [AI Marketer](/aim/about), and Artifacts. AI Marketer suggests new Projects and tasks based on your data. See also [Suggestion](/getting-started/glossary#suggestion).

Learn more: [About Projects](/projects/about)

### Prompt

A query that Profound sends to answer engines on your behalf, such as "What is the best tool for answer engine analysis?" Prompts run once a day on each configured platform, and their responses form the dataset behind [Answer Engine Insights](/answer-engine-insights/interpret). Prompts belong to a topic, can carry tags, regions, and Personas, and are managed in the [Prompt Designer](/answer-engine-insights/prompt-designer).

Learn more: [Create, manage, and tag prompts in Prompt Designer](/answer-engine-insights/prompt-designer)

### Prompt Designer

The interface for creating, editing, tagging, and organizing prompts and topics. Open it from [Answer Engine Insights](/answer-engine-insights/interpret) > Prompts > **Modify Prompts**. Add prompts by bulk upload, with the Prompt Generator, or manually. Requires the [Admin](/platform-config/people/roles-and-permissions) role.

Learn more: [Create, manage, and tag prompts in Prompt Designer](/answer-engine-insights/prompt-designer)

### Prompt Generator

A tool in the [Prompt Designer](/answer-engine-insights/prompt-designer) that generates collections of prompts for your existing topics. You choose the topics, regions, languages, and platforms, and can add extra instructions.

Learn more: [Create, manage, and tag prompts in Prompt Designer](/answer-engine-insights/prompt-designer)

### Prompt LLM

The core node for [Workflows](/agents/workflows/about-workflows): it sends a prompt you write to an AI model and returns the response. Most content workflows chain several Prompt LLM nodes.

Learn more: [Prompt LLM](/agents/workflows/nodes/prompt-llm)

### Prompt Research Reports

Reports in [Prompt Volumes](/prompt-volumes/about) that compare what real users ask answer engines with what you track in [Answer Engine Insights](/answer-engine-insights/interpret), to surface gaps in your prompt coverage.

Learn more: [About Prompt Volumes](/prompt-volumes/about)

### Prompt type

See [Analysis type](/getting-started/glossary#analysis-type).

### Prompt Volume

The projected demand for a prompt or topic, drawn from the [Prompt Volumes](/prompt-volumes/about) dataset.

Learn more: [About Prompt Volumes](/prompt-volumes/about)

### Prompt Volumes

Profound's dataset of real prompts that users submit to answer engines. Data comes from ChatGPT, Gemini, Claude, and Perplexity and is anonymized and aggregated.

Learn more: [About Prompt Volumes](/prompt-volumes/about)

## Q

### Query fanout

A search query an answer engine generates behind the scenes while answering a prompt.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Query Fanout Estimator

A content optimization node and template for [Workflows](/agents/workflows/about-workflows) that estimates how a topic expands into related query fanouts, so content can cover them.

Learn more: [Query Fanout Estimator](/agents/workflows/nodes/content-optimization/query-fanout-estimator)

### Query fanouts

A view in the Prompts tab of [Answer Engine Insights](/answer-engine-insights/interpret) that lists the queries answer engines generate from each prompt you track.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

## R

### Region

The country a prompt is scoped to when Profound runs it.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Research Snippets Generator

A research node for [Workflows](/agents/workflows/about-workflows) that produces short research snippets on a topic for use in briefs or articles.

Learn more: [Research Snippets Generator](/agents/workflows/nodes/research/research-snippets-generator)

### Response

The answer an answer engine returns for one execution of a prompt. Profound analyzes each response for brand mentions, citations, sentiment, and claims.

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Rise

In Citation Decay, the number of days from a source's first citation to its peak citation count. The summary shows the Average Rise across all sources. Compare [Half-life](/getting-started/glossary#half-life).

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Role

The permission level of a member in your [organization](/getting-started/organizations-and-teams). Roles control who can modify prompts, manage settings, and see credit usage. You can use pre-defined roles, such as Admin or Member, or define a [custom role](/platform-config/people/custom-roles) with a custom set of permissions.

Learn more: [Roles and permissions](/platform-config/people/roles-and-permissions)

## S

### Saved Views

A stored combination of date range, comparison period, and filters in [Answer Engine Insights](/answer-engine-insights/interpret) or [Pages](/pages/about).

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Sentiment

The part of [Answer Engine Insights](/answer-engine-insights/interpret) that tracks the tone and content of how answer engines describe your brand. Sentiment extracts claims from responses, clusters them into themes and attributes with positive or negative polarity, shows the exact text behind each signal, and attributes signals to the citations driving them.

Learn more: [About Sentiment](/answer-engine-insights/sentiment/about)

### Sentiment (node)

A data node for [Workflows](/agents/workflows/about-workflows) that returns positive or negative sentiment scores from current Sentiment data for use inside a Workflow.

Learn more: [Sentiment node](/agents/workflows/nodes/data/sentiment)

### Sentiment Score

The share of positive sentiment for a brand, averaged across tracked [platforms](/answer-engine-insights/interpret). Claims are weighted by [mention](/answer-engine-insights/settings) frequency, so a theme that appears more frequently carries more weight. For example, 75 positive and 25 negative mentions give a Sentiment Score of 75%.

Learn more: [About Sentiment](/answer-engine-insights/sentiment/about)

### Sentiment Score (node)

A data node for [Workflows](/agents/workflows/about-workflows) that returns Sentiment Score values for use inside a workflow.

Learn more: [Sentiment Score Node](/agents/workflows/nodes/data/sentiment-score)

### Share of Voice

How often your brand is mentioned compared with all brands. Share of Voice divides the number of responses that mention your brand by the total number of brand mentions across all responses. For example, 20 mentions out of 100 total brand mentions gives a Share of Voice of 20%. Compare [Visibility Score](/getting-started/glossary#visibility-score).

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Share of Voice Rank

Your brand's rank by Share of Voice relative to competitors for a prompt or topic. Shown in the rank list beside the Share of Voice chart and as a column in the Tracked prompts table.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Sheets

The Profound product for running [Workflows](/agents/workflows/about-workflows) at scale. A Sheet is a spreadsheet where each row is an input (a page, topic, keyword, or prompt) and each column is a data field, a Workflow step, or an output. Rows run in parallel, and outputs write back into the table.

Learn more: [Get started with Sheets](/sheets/getting-started)

### Sheets Templates

Ready-made [Sheet](/sheets/getting-started) starting points for common workflows, such as Article Translator, Content Refresher, FAQ Generator, and Internal Link Suggestions. Each comes pre-loaded with the right [Workflow](/agents/workflows/about-workflows), columns, and sample data.

Learn more: [Get started with Sheets](/sheets/getting-started)

### Shopping

The Profound product that shows how your products appear inside AI shopping experiences.

Learn more: [About Shopping](/shopping/about)

### Shopping mode

The state an AI chat enters when a prompt signals purchase intent. Instead of a text answer, the answer engine returns a carousel of product tiles.

Learn more: [About Shopping](/shopping/about)

### Shopping Mode Rate

The percentage of your tracked prompts that trigger Shopping mode rather than a standard text answer.

Learn more: [About Shopping](/shopping/about)

### Shopping Visibility Score

How consistently your brand and products appear across Shopping mode responses. A score of 100% means your brand appears every time the answer engine enters shopping mode for that prompt set. Review it by topic, because scores can swing between topics for the same brand.

Learn more: [About Shopping](/shopping/about)

### Strongholds

The Topic Prioritization group for topics with high Visibility Score and high Citation Share.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Suggestion

A new Project or task that [AI Marketer](/aim/about) suggests based on your data and context. Project suggestions appear at the top of the **Projects** page, and task suggestions appear in the **Suggestions for you** panel of a Project. Select **Start task** on a Suggestion to open a chat where AI Marketer starts on the plan with you. If a Suggestion fits one of your existing [Agents](/agents/about), AI Marketer can use it.

Learn more: [About Projects](/projects/about)

## T

### Tag

A free-form label you apply to a prompt as a second layer of organization beyond topics.

Learn more: [Create, manage, and tag prompts in Prompt Designer](/answer-engine-insights/prompt-designer)

### Team

A named group of members inside your [organization](/getting-started/glossary#organization). Assign categories to a team to give every member access to them at once, share Dashboards, Documents, Sheets, Agents, Workflows, and other resources with the whole team, and give the team its own [credit](/getting-started/glossary#credits) wallet.

Learn more: [How teams work in Profound](/platform-config/people/teams/overview), [How your Profound data and access are organized](/getting-started/organizations-and-teams)

### Template

A pre-built [Agent](/agents/about), [Workflow](/agents/workflows/about-workflows), or [Sheet](/sheets/getting-started) for a common use case, such as AEO Content Refresh or FAQ Generator.

Learn more: [About Agents](/agents/about)

### Theme

A neutral, high-level topic area that answer engines associate with a brand, such as Customer Support or Pricing. In [Sentiment](/answer-engine-insights/sentiment/about), themes group attributes with positive or negative polarity; in [FactCheck](/answer-engine-insights/factcheck/about), themes cluster similar claims.

Learn more: [About Sentiment](/answer-engine-insights/sentiment/about), [About FactCheck](/answer-engine-insights/factcheck/about)

### Top Citation Domains

A ranked list of the [domains](/answer-engine-insights/interpret) most frequently cited in AI responses.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Top Citation Pages

A ranked list of the pages most frequently cited in AI responses. Several pages from the same domain can appear.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Topic

A group of related prompts organized around a common theme, product, or business objective, such as "Pricing" or "Integrations." Topics structure your prompt set and are the main way to compare performance across areas of your business. Create topics in the [Prompt Designer](/answer-engine-insights/prompt-designer).

Learn more: [How to create topics and prompts to track in Profound](/answer-engine-insights/how-tos/create-topics-and-prompts)

## V

### Validation prompt

A yes-or-no prompt that checks whether AI knows your brand has a specific capability, for example *"Does Profound offer prompt volume tracking?"*

Learn more: [Six types of prompts in Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts)

### Variable

A piece of data that flows in or out of a [Workflow](/agents/workflows/about-workflows) node. Variables pass between steps and can be reused anywhere later in the Workflow.

Learn more: [About Workflows](/agents/workflows/about-workflows)

### Visibility

Whether your brand appears in an answer engine response. Measured by [Answer Engine Insights](/answer-engine-insights/interpret). Related metrics: Visibility Score, Share of Voice, Average Position.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Visibility Rank

Your brand's rank by Visibility Score relative to competitors for a prompt or topic. Compare [Share of Voice Rank](/getting-started/glossary#share-of-voice-rank).

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Visibility Score

The percentage of responses that mention your brand, out of all responses that mention at least one brand. For example, appearing in 5 out of 10 responses gives a Visibility Score of 50%. It's the headline metric in [Answer Engine Insights](/answer-engine-insights/interpret). Compare [Share of Voice](/getting-started/glossary#share-of-voice), which counts mentions rather than responses.

Learn more: [About Answer Engine Insights](/answer-engine-insights/about)

### Visibility Score (node)

A data node for [Workflows](/agents/workflows/about-workflows) that returns Visibility Score values for a brand, topic, or prompt set for use inside a Workflow.

Learn more: [Visibility Score](/agents/workflows/nodes/data/visibility-score)

## W

### Watchlist

The view of the All Citations table in [Answer Engine Insights](/answer-engine-insights/interpret) that monitors a custom set of pages you define, rather than the most cited pages overall.

Learn more: [Interpret Answer Engine Insights](/answer-engine-insights/interpret)

### Web Page Scrape

A node for [Workflows](/agents/workflows/about-workflows) that fetches the content of a web page so later Workflow steps can analyze or rewrite it. Uses [ProfoundBot](/bots/profoundbot).

Learn more: [Web Page Scrape](/agents/workflows/nodes/web-page-scrape)

### Whitelist

A mode in [Configure Competitors](/answer-engine-insights/settings#configure-competitors) that shows only the brands on your list and excludes any response that mentions none of them. Compare [Blacklist](/getting-started/glossary#blacklist).

Learn more: [Answer Engine Insights settings](/answer-engine-insights/settings)

### Workflow

A node-based automation you build yourself in the [Workflow Builder](/getting-started/glossary#workflow-builder). Each node does one job, such as querying Profound data, prompting an AI model, fetching a web page, or publishing to a CMS, and data moves between nodes as variables. Every run follows the same path, so the output has the same structure each time. Run a Workflow on demand, on a schedule, across many inputs from [Sheets](/sheets/getting-started), as a step inside an [Agent](/getting-started/glossary#agent), or through the Profound API and [Profound MCP](/mcp/overview). Workflow runs consume [credits](/getting-started/credits).

Learn more: [About Workflows](/agents/workflows/about-workflows)

### Workflow Builder

The canvas where you assemble a [Workflow](/getting-started/glossary#workflow): add nodes, connect them, define the inputs, and run tests to check what each node returns and what the run is expected to cost in [credits](/getting-started/credits) before you publish.

Learn more: [Create a Workflow](/agents/workflows/create-workflow)

### Workflow output

What a [Workflow](/getting-started/glossary#workflow) returns when its last node finishes: generated content, a published article, a Slack message, a [Profound Document](/documents/about), or a confirmation that the run succeeded. In [Sheets](/sheets/getting-started), the output lands in the row that supplied the input.

Learn more: [About Workflows](/agents/workflows/about-workflows)

### Workflow run

One execution of a [Workflow](/getting-started/glossary#workflow), from the moment its inputs are submitted to its final output. Runs are independent of each other: each has its own status, result, and [credit](/getting-started/credits) cost, and the same Workflow can return different results for different inputs. Test runs in the Workflow Builder also show the variables each node produced. Review earlier runs from the Workflow's **Past runs** view.

Learn more: [Workflow runs](/agents/workflows/workflow-runs)

### Workspace

An isolated brand environment in Agency Mode with its own prompts, competitors, tracked platforms, and configuration.

Learn more: [Agency Mode overview](/agencies/agency-mode-overview)

## Y

### YouTube Title & Description

A node for [Workflows](/agents/workflows/about-workflows) that fetches the title and description of a YouTube video for use in a content workflow.

Learn more: [YouTube Title & Description](/agents/workflows/nodes/youtube-title-description)

### YouTube Video Transcript

A node for [Workflows](/agents/workflows/about-workflows) that fetches a YouTube video's transcript so later nodes can turn it into written content.

Learn more: [YouTube Video Transcript](/agents/workflows/nodes/youtube-video-transcript)


# How your Profound data and access are organized
Source: https://docs.tryprofound.com/getting-started/organizations-and-teams

Learn how organizations, teams, and categories fit together in Profound, and when to use teams or separate organizations

> [!TIP]
>
> Find definitions of unfamiliar terms and concepts in the [Profound glossary](/getting-started/glossary).

Everything you do in Profound lives in an **organization**: your categories, members, credits, and resources, such as Agents, Dashboards, or Knowledge Bases. Inside an organization, **teams** group people who work together, so you can give the whole group access to the same categories and resources at once.

## Organizations

An organization is your company's space in Profound. It holds your data, your members and their [roles](/platform-config/people/roles-and-permissions), and your [credits](/getting-started/credits).

### What data belongs to an organization

Some data belongs to the organization as a whole, and some belongs to a single [category](/getting-started/categories) inside it.

**Organization-wide** data is available across all your categories:

* [Agents](/agents/about) and [Workflows](/agents/workflows/about-workflows)
* [Sheets](/sheets/getting-started)
* [Documents](/documents/about)
* [Dashboards](/custom-dashboards/about)
* [Projects](/projects/about)
* [AI Marketer (Aim)](/aim/about) chats
* [Integrations](/integrations/overview)
* [Agent Analytics](/agent-analytics/overview) data
* [Context](/context-manager/about): Brand context, [Knowledge Bases](/context-manager/knowledge-bases-and-brand-kits), and Skills. You can [limit a Knowledge Base to specific categories](/context-manager/manage-knowledge-bases#choose-which-categories-a-knowledge-base-applies-to).

This data is scoped to your organization, but that doesn't mean that all members can see it by default. Resources such as Dashboards, Documents, Projects, Sheets, Aim chats, Agents, and Workflows stay private to the person who created them until they're shared with people, teams, or the whole organization.

**Category-specific** data is available only inside the selected category:

* Everything inside [Answer Engine Insights](/answer-engine-insights/about): prompts, topics, tags, assets, Personas, competitors, and so on
* [Prompt Volumes](/prompt-volumes/about)
* [Sentiment](/answer-engine-insights/sentiment/about)
* [Shopping](/shopping/about)
* [Pages](/pages/about)

Learn more in [Categories in Profound](/getting-started/categories).

### Creating an organization

Profound creates an organization in one of the following ways:

* **With a Profound contract**: when your company buys Profound through the sales team, Profound sets up the organization for you.
* **With a sign-up on the Profound website**: when you create an account yourself, Profound creates a new organization on a trial plan during onboarding and makes you its admin.

To join an organization that already exists, ask an [admin](/platform-config/people/roles-and-permissions) to invite you. If your company uses [Directory Sync](/platform-config/people/directory-sync), your identity provider adds you automatically.

## Teams

A team is a named group of members inside your organization. Teams solve a common problem for larger companies: different groups work on different brands, markets, or clients, and each group needs access to its own categories and resources. Without teams, an admin grants that access to one person at a time. With teams, you grant it once to the team, and everyone in it gets the same access. Learn more in [How teams work in Profound](/platform-config/people/teams/overview).

### What data belongs to a team

A team doesn't own data: the data belongs to the organization, and the team controls who can reach it. Each team has:

* **Members**: the people in the team.
* **Categories** (optional): the categories every member of the team can access.
* **Shared resources** (optional): Dashboards, Documents, Sheets, Agents, Workflows, and other resources shared with the team.
* **Team wallet** (optional): [credits](/getting-started/credits) reserved for the team's use. Learn more in [Team wallets](/platform-config/people/teams/overview#team-wallets).

## How organizations, teams, and categories relate

* A team belongs to one organization, and only members of that organization can join it.
* A member can belong to several teams.
* A team can access only categories your organization uses, including categories [shared with it by another organization](/getting-started/categories#when-you-have-multiple-organizations-and-categories).
* Team access adds to the access a member already has. Members keep the categories granted to them individually on the [People page](/platform-config/people/add-and-manage-org-members#add-a-member), plus everything their teams grant.
* When you remove a member from a team, they lose the categories and shared resources they had through that team. Anything granted to them directly or through another team stays. If they were spending from the team's wallet, they go back to spending from the organization's wallet.

## Multiple organizations and teams

Teams and organizations both split your company into groups. You can keep all of your Profound data in one organization split into teams, or have multiple organizations, each with its own data. The difference between the two approaches is how complete you want the split to be.

### When to use teams

Use teams when groups need to work separately but stay connected under one set of admins. With teams:

* **Admins oversee every group.** Organization admins manage all teams in one place and can see every member.
* **Groups can still work together.** A member can belong to several teams, and anyone can share organization-wide resources (such as Projects, Documents, or Agents) with another team, without switching organizations.
* **Reporting is unified.** Dashboards can combine data from every team's categories, so leadership sees the whole business in one place.

**Example use case**: a company with "EMEA Marketing" and "North America Marketing" teams, each with its own categories and budget, and a central marketing operations admin who oversees both.

### When to use multiple organizations

Use separate organizations only when groups need full separation, including different admins. With separate organizations:

* **Each organization has its own admins and members.** No one can see an organization's data unless they're a member of it, so an admin of one organization can't see the members of another.
* **Members switch between organizations manually** with the organization selector to work in each one.
* **Reporting stays separate.** Dashboards and other reports can't combine data across organizations.

**Example use case**: a holding company whose subsidiaries run as independent businesses. Each subsidiary's marketing team manages its own Profound setup, and no one at the parent company or at another subsidiary should have admin access to it.

### How multiple organizations and teams relate

* A team belongs to one organization. It can't include members or categories from another organization.
* One Profound login can belong to several organizations. Your role and teams in one organization don't carry over to another.
* An organization can share categories with other organizations. The receiving organization can use a shared category and give its own teams access to it, but only the owning organization can edit it.


# Profound platform walkthrough: see how it works
Source: https://docs.tryprofound.com/getting-started/platform-walkthrough

A video walkthrough of how to understand and control how your brand shows up in AI search

> [!NOTE]
>
> The video below shows the legacy Answer Engine Insights design. Learn more about the current features in [Interpret Answer Engine Insights](/answer-engine-insights/interpret).

[https://player.mux.com/WKUPfYhw1Fl02PkcqGCMYASxkDnIP8bysjl6FI5CrbwI?\&accent-color=%23000000\&primary-color=%23ffffff\&default-hidden-captions=true\&thumbnail-time=146](https://player.mux.com/WKUPfYhw1Fl02PkcqGCMYASxkDnIP8bysjl6FI5CrbwI?\&accent-color=%23000000\&primary-color=%23ffffff\&default-hidden-captions=true\&thumbnail-time=146)

This walkthrough covers everything you need to start understanding and controlling how your brand shows up in AI Search.

You'll see:

* **Prompt monitoring:** what questions people are asking that surface your brand (or your competitors')
* **Answer engine visibility:** how Profound tracks your brand's presence, share of voice, and sentiment across AI-generated answers
* **Citation Tracking:** where your content is being pulled from (or missed) when AI models generate responses
* **Competitive Benchmarking:** how you stack up against competitors in AI Search visibility
* **Actionable Recommendations using Agents:** where to focus content and optimization efforts for maximum AI engine optimization (AEO) impact using agentic workflows

Whether you're new to AI Search optimization or looking to get more out of your Profound account, this walkthrough will help you navigate the platform with confidence and turn insights into action.


# Profound Knowledge Base
Source: https://docs.tryprofound.com/index

Guides, references, and recipes for using Profound and building on its data

# Profound Knowledge Base

Guides, references, and recipes for using Profound and building on its data

> **[Analyze your AI search presence](/getting-started/glossary)**
>
> Answer Engine Insights, Prompt Volumes, dashboards, and the basics to get started

> **[Get the work done with AI Marketer (Aim) and Agents](/aim/about)**
>
> Your AI marketing teammate, Projects, Agents, Sheets, and the brand context behind them

> **[Set up your organization](/platform-config/authentication/configure-sso)**
>
> Authentication, Activity Logs, Analytics Setup, team access, and agency mode

> **[Connect your tools](/integrations/overview)**
>
> Integrations and connectors that make your CMS, analytics, and messaging tools available to Agents and Aim

> **[Build with the API and SDKs](/rest-api/introduction)**
>
> Pull Profound data into your own applications, reports, and pipelines

> **[Work with Profound from AI tools](/mcp/overview)**
>
> Give ChatGPT, Copilot, and coding assistants access to your Profound data through MCP

> **[Follow reporting recipes](/cookbook/introduction)**
>
> Ready-made API queries for visibility, citations, and metrics

## More ways to learn

> **[Blog](https://www.tryprofound.com/blog)**
>
> Product news and perspectives on AI search

> **[Community](https://community.tryprofound.com)**
>
> Meet other Profound users, swap tips, and ask questions

> **[Profound University](https://university.tryprofound.com)**
>
> Courses and resources to build your expertise

> **[Research hub](https://www.tryprofound.com/research)**
>
> Studies and insights on AI search from the Profound team

> **[The Marketing Engineer Podcast](https://www.tryprofound.com/podcasts)**
>
> Conversations with the people shaping marketing in the AI era

> **[Webinars](https://www.tryprofound.com/webinars)**
>
> Live and recorded sessions on the platform and AI search strategy

> **[White papers](https://www.tryprofound.com/reports-guides)**
>
> In-depth reports and guides on AI search

## Need help?

> **[Status](https://status.tryprofound.com)**
>
> Check service health and current incidents

> **[Contact support](mailto:support@tryprofound.com)**
>
> Reach the Profound team for help with your account or setup


# About Pages
Source: https://docs.tryprofound.com/pages/about

How Pages shows the performance of your content across AI search

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Analytics > Pages

Pages gives you a clear view of how your content performs in AI search. It brings everything you need from [Answer Engine Insights](/answer-engine-insights/about) and [Agent Analytics](/agent-analytics/overview) into one place to track the performance of your web pages. See how each page performs right now and how it's trending over time, all in a single view.

![Pages table for tryprofound.com with Category, Citation Share, Citation Rank, AI Citations, and AI Indexing columns](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/pages/about-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=d9f9b488b5e4bf5408de748b0bd8b220)

## One view for how your content performs

Your content is the primary way you influence how your brand appears in answer engines. Until now, getting a clear picture of how it performs involved too many steps: the story for any one page was split between [Answer Engine Insights](/answer-engine-insights/about) and [Agent Analytics](/agent-analytics/overview). Even without Agent Analytics, digging into citation data in Answer Engine Insights takes several clicks and a clear idea of what you're looking for. That's a lot of effort to answer a basic question: is this content working?

Pages brings it all together. It centralizes the data Profound has about each piece of owned content: citation share from Answer Engine Insights, bot visits and benchmarks from Agent Analytics, Page Health, Bot vs. Human Readable Content, and content scores. From one place, see what content drives impact, spot opportunities to improve, and follow trends over time.

## Working with Pages

Pages works for every customer, but what you see depends on whether Agent Analytics is configured for your account.

### Without Agent Analytics

You get Pages with your [Answer Engine Insights](/answer-engine-insights/about) citation data, content scores, and AEO readiness scoring. Page Health, Bot vs. Human Readable Content, and Indexing are available too, so you can crawl your content, score how ready it is for answer engines, submit updates for indexing, and export rows. This removes a major activation blocker: act on your content right away, before any content delivery network (CDN) setup is complete.

### With Agent Analytics

Adding [Agent Analytics](/agent-analytics/overview) layers in live bot and referral activity from your server logs: AI citation bot traffic, AI indexing, AI training, AI agent visits, and human referrals, along with per-page benchmarks. This gives you the complete view of how AI systems both cite and crawl each page.

## Key features

### Unified page-level analytics dashboard

A single table lists every page with its key metrics in one row, combining citation data from Answer Engine Insights with live bot and referral data from Agent Analytics. The overview ranks pages by **citation share** and enriches them with bot visits (**AI citation** traffic, **AI indexing**, **AI training**, and **AI agent visits**), **human referrals**, tags, and category context. It's the fastest way to find the pages driving your visibility.

### Citations breakdown

See exactly where and why a page is cited. Pages breaks a page's citations down by prompt, topic, tag, platform, region, persona, and even text chunks, each with citation share and citation rank, so you can tell which queries and audiences the page is winning.

### Content analysis and recommendations

The **Content** tab analyzes a page's content and scores how ready it is for answer engines, with an overall score and a target zone. The score breaks down into factors like readability, content freshness, content structure, information density, and machine readability, and Pages turns each into specific optimization recommendations with the projected score impact of the change.

### Network benchmarking

Your Page Ranking shows how a page's citations compare against roughly 3 million pages tracked across the Profound Network, scored from Poor to Great as a percentile. Benchmark against all sites or narrow to a peer group, so a raw number becomes a clear sense of how the page stacks up. This feature requires Agent Analytics.

### Page Health

Page Health tells you whether your best content can load, render, and perform well enough for users, crawlers, and answer experiences to use it. It's an SEO-focused performance read on each page, sourced from Google Lighthouse and PageSpeed Insights. It covers core web performance metrics such as Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift.

### Bot vs. Human Readable Content

See the difference between what a person sees on a page and what an AI bot can access. Content rendered with client-side JavaScript often isn't visible to bots, so a page that looks complete to you can appear nearly empty to an answer engine. Pages flags the specific text blocks and headings hidden this way and recommends how to make them visible, so AI systems can find and cite the content that matters.

### Indexing

Submit a page for indexing directly from Pages to prompt search engines to pick up new or updated content sooner. Indexing uses [IndexNow,](https://www.indexnow.org/index) a protocol to request priority indexing across subscribing search engines. The protocol doesn't guarantee priority indexing. If the request is accepted, indexing can take a few days for large enterprises and up to a few weeks for smaller sites.


# Pages FAQ
Source: https://docs.tryprofound.com/pages/faq

Frequently asked questions about Pages

> **Do I need Agent Analytics to use Pages?**
>
> No. Agent Analytics is only needed for live bot and referral activity. Without it, Pages still shows your Answer Engine Insights citation data, runs Page Health, Bot vs. Human Readable Content view, supports submission for indexing, crawls your content, scores AEO readiness, and lets you export.

> **Why does a page show bot activity, but no citation data?**
>
> The two measure different things, so a gap is expected. Citation share in [Answer Engine Insights](/answer-engine-insights/about) comes from the prompts Profound runs through answer engines, while [Agent Analytics](/agent-analytics/overview) records real visits from AI bots crawling your site. Some reasons that explain the mismatch:
>
> * Answer Engine Insights tracks some engines that Agent Analytics doesn't
> * Your tracked prompts may not match how people actually search, which skews cited pages toward your prompts rather than real-world usage
> * Answer engines often reuse a cached copy of a page instead of fetching it fresh, so a citation can show up with no matching bot visit

> **Why are the Agent Analytics data columns hidden or locked?**
>
> Agent Analytics must be set up for the columns (like Bot Visits, AI indexing, and so on) to appear in the Pages dashboard. Once Agent Analytics is set up for the domain, the columns appear.

> **What is the AEO readiness score for?**
>
> It shows how prepared a specific page is for answer engine performance and gives your content team concrete recommendations to act on. Use it to decide what to improve on a page, and in what order.

> **Why doesn't the Agent Analytics data in Pages match the Agent Analytics tab?**
>
> The two use slightly different date ranges. Answer Engine Insights data is always one day behind, so Pages shows Agent Analytics data on that same one-day-delayed range to keep everything in one view aligned. The Agent Analytics tab updates in real time (the last 7 days up to now), so for the same window its numbers can differ from what you see in Pages.


# Get started with Pages
Source: https://docs.tryprofound.com/pages/getting-started

Open Pages, review your content, and act on page-level insights

## Step 1: Open Pages

Log in to the Profound platform and open **Pages** from the left sidebar.

## Step 2: Choose your domain and date range

Select the domain you want to review, then set the date range and comparison period at the top of the view, for example, Last 7 Days versus the previous period.

![A screenshot of the Pages view with the date range filter highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/pages/getting-started-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=2824c02ce6feefec4f823128dcaf00ea)

## Step 3: Find your top pages

Pages lists every page with its metrics in one row. Sort by **Citation Share** to see which pages answer engines reference most, use the **Customize** option to choose which columns appear in the dashboard, or use Group by to roll pages up by category.

## Step 4: Open a page to see its analytics

Click any page to open its detailed view. The **Analytics** tab shows the citation share trend, AI bot activity by platform, what information on the page is accessible to bots, how the page ranks against the Profound Network, and a breakdown of where the page is cited by prompt, topic, platform, and more.

![A Page view screenshot with Analytics tab highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/pages/getting-started-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=e77d62f5d44e4195526f9fdef413f10f)

## Step 5: Review and improve content

Switch to the **Content** tab to see the page's content score and optimization recommendations. Work through the suggested content fixes. Select **Index page** in the top right of the view to submit updated content for faster indexing.

![Page detail with the Content tab and Index page button highlighted, Overall Score gauge shown](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/pages/getting-started-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=a0c7e508cb23c61d078755ae0dc41f04)

## Step 6: Connect Agent Analytics for the full picture

To add live bot and referral activity, AI agent visits, and per-page benchmarks, set up Agent Analytics through your CDN or server-level integration.

> [!TIP]
>
> Learn more about setting up Agent Analytics in the [Agent Analytics](/agent-analytics/self-serve-onboarding) section.


# Consent and privacy
Source: https://docs.tryprofound.com/pixel/consent-and-privacy

How Profound Pixel handles consent, which cookies it writes, and what data it collects

> [!NOTE]
>
> This page is a factual description of the pixel's behavior. It is not legal advice, and it makes no recommendation about what your policies, notices, or agreements should say. Those are decisions for your own counsel.

## Consent handling

The way Profound Pixel handles consent depends on whether your website uses a consent management platform (CMP). On sites with a CMP, the pixel reads the consent signal the CMP emits. On sites without one, it follows the consent posture set in your snippet.

Either way, the pixel doesn't ask your visitors for consent and adds no cookie banner of its own to your site.

### The consent signal

Whatever its source, the consent signal the pixel reads consists of the following three permissions:

| Permission | Controls | When denied |
| - | - | - |
| Storage | Reading and writing Profound's cookies, the only browser storage the pixel uses | No cookies are read or written; the pixel works from memory only |
| Send | Whether events may leave the browser | Events wait in memory (up to 100; the oldest are dropped past that); nothing is sent |
| Ad-platform sharing | Any future forwarding of conversions to an ad platform | Nothing is shared. No forwarding exists today |

These permissions are independent, and none of them implies another. For example, consent to cookie storage alone never allows ad-platform sharing or sending out events.

### Sites with CMP

If your site uses a CMP, it handles consent for you: it emits the consent signal that Profound Pixel detects and follows. The recognized standards and vendors include:

| Standard or vendor | Detected via | Notes |
| - | - | - |
| IAB TCF v2.2 (OneTrust, Cookiebot, Sourcepoint, Didomi, and others) | `__tcfapi` | - |
| Google Consent Mode v2 | `ad_storage`, `analytics_storage`, `ad_user_data`, `ad_personalization` | - |
| OneTrust, Cookiebot, Osano, TrustArc, vanilla-cookieconsent | Built-in support for each vendor's consent categories | - |
| US Privacy / GPP | `__uspapi`, `__gpp` | - |
| Global Privacy Control, Do Not Track | Browser signals | Both always turn off ad-platform sharing and, under the opt-out posture, also stop events from being sent. Neither affects storage: cookies are still written |

#### If your CMP isn't detected

If your CMP isn't recognized, or if you built your own consent banner, you'll need to map your consent categories to the [permissions](#the-consent-signal) Profound Pixel can read.

Call `pfq('consent', …)` from your CMP's change handler, and pass it the permissions map:

```javascript theme={null}
pfq('consent', { storage: true, send: true, share_ads: false });
```

Permissions you leave out keep their default values: denied in the EU/UK, or defined in your [consent posture](#sites-without-cmp) in other regions.

### Sites without CMP

#### Consent posture

If your site doesn't use a CMP, you have to explicitly set your consent posture when you initialize your Profound Pixel. The options are:

* **Opt-in (default)**: all three consent permissions start denied until a consent signal grants them. This is the best setting for EU/UK traffic.
* **Opt-out**: the permissions start granted until a signal denies them, with Global Privacy Control honored. This is the usual setting for US-based traffic. Set it with `pfq("init", "<YOUR_PIXEL_ID>", { posture: "opt-out" })`. Learn more in the [pixel installation guide](/pixel/install).

#### Timezone check

When reading the consent signal, the pixel also checks the browser's timezone. A visitor who appears to be in a European timezone is treated as opt-in even under the opt-out posture. Profound applies this check to comply with European data protection regulations.

The check covers more than the EU region: any `Europe/*` timezone counts, including cities outside the EU, such as Istanbul, Belgrade, and Kyiv, plus several Atlantic islands.

If your pixel is scoped to a single non-EU region, or your local data protection regulations otherwise allow the opt-out posture, you can disable the timezone check by adding `detectRegion: false` to the `pfq('init', ...)` call in your install snippet:

```javascript theme={null}
pfq("init", "<YOUR_PIXEL_ID>", { detectRegion: false });
```

> [!WARNING]
>
> Always review your local data protection laws before disabling the timezone check.

### Consent withdrawal

When a visitor withdraws consent, the changes take effect immediately: the pixel deletes its cookies and clears its memory. The visitor also gets a new random visitor ID, so granting consent again later can't restore the old ID.

## Cookies

Profound Pixel writes the following cookies into the site visitor's browser storage:

| Name | Purpose | Lifetime (set by Profound) |
| - | - | - |
| `_pf_id` | Random visitor ID, first and last acquisition source, and ad attribution data | 365 days |
| `_pf_clk` | Ad and click identifiers | 90 days |

> [!NOTE]
>
> Some browsers cap the lifetime of script-written cookies below the set lifetime values. For example, Safari's Intelligent Tracking Prevention shortens both to roughly 7 days.

Both cookies are first-party and marked `SameSite=Lax` and `Secure`. Only the pixel script in the browser writes them, and only after your consent signal allows storage. Profound's servers set no cookies at all.

## Data collection

Profound Pixel collects the following data:

| Category | Examples | Notes |
| - | - | - |
| Visitor ID | A random browser-generated ID | Generated by Profound Pixel, not derived from any personal attribute |
| Ad and click IDs | Ad and campaign identifiers from the landing URL | - |
| Acquisition source | The channel or hostname the visitor came from | This does not include the full referrer URL |
| Page context | Page URL, referrer, title | Any personal data is redacted on Profound's servers |
| Query string | The page URL's query parameters, such as `?ref=newsletter` | Derived on Profound's servers, with sensitive keys dropped |
| Your user IDs | `external_user_id` and `external_account_id` | Collected if you send them in [`identify` call](/pixel/events#tie-events-to-your-own-user-ids) |
| Network | IP address | Never stored for visitors in the EU/EEA, the UK, or Switzerland |
| Country | A two-letter country code (such as US, GB, FR) | No city or precise location is collected |
| Device | The browser's raw User-Agent string, taken from the request header | - |
| Event properties | Whatever you send in event properties | - |
| Consent record | The permission flags and the signal they came from | - |

**Notes**:

* Whether the IP address is discarded or stored depends on the country Profound's servers detect for the request. The detection is reliable in practice, but it isn't guaranteed for any individual request. If the detection fails, the country is recorded as empty, and the IP address is stored.
* Collected data is processed in the United States.
* Conversion data is not forwarded to OpenAI, Google, or any other ad platform today. If that ever changes, Profound will inform you before it is switched on.

## Optional: forwarding conversions to OpenAI

> [!NOTE]
>
> This capability is still in development and subject to change.

When you run ads with OpenAI, OpenAI normally learns about your conversions through [its own pixel](https://developers.openai.com/ads/measurement-pixel) or its [Conversions API](https://developers.openai.com/ads/conversions-api). Profound can handle this for you: because Profound Pixel uses the same event names as OpenAI, Profound can pass your conversions on to OpenAI's Conversions API from its own servers, matched to the ad through OpenAI's click ID.

For you, this would mean:

* OpenAI receives the conversions it needs to credit your ads and adjust bidding.
* Profound Pixel stays the only tag on your site: you don't install OpenAI's pixel, and you keep the full first-party record with Profound.
* Visitors stay in control: the forwarding is governed by the [Ad-platform sharing permission](#the-consent-signal), which stays denied until consent allows it.

If this is an option you would like to explore, speak to your Profound team.


# Events reference
Source: https://docs.tryprofound.com/pixel/events

Profound Pixel events, their fields, and usage

## Standard events

Profound Pixel recognizes the following standard events. Each event accepts a set of optional fields. See [Field sets](#field-sets) for what each set accepts.

| Event | When to send it | Field set |
| - | - | - |
| `page_viewed` | A visitor views a page | `contents` |
| `contents_viewed` | A visitor views a product, listing, article, or other content unit | `contents` |
| `items_added` | A visitor adds items to a cart or selection | `contents` |
| `checkout_started` | A visitor begins checkout | `contents` |
| `order_created` | A purchase completes | `contents` |
| `subscription_created` | A paid subscription starts | `plan_enrollment` |
| `trial_started` | A free trial starts | `plan_enrollment` |
| `lead_created` | A visitor submits a lead form or requests contact | `customer_action` |
| `registration_completed` | A visitor finishes account or event registration | `customer_action` |
| `appointment_scheduled` | A visitor books a meeting, demo, or consultation | `customer_action` |

### Field sets

| Field set | Recognized fields |
| - | - |
| `contents` | `amount`, `currency`, `contents[]` |
| `plan_enrollment` | `plan_id`, `amount`, `currency`, `contents[]` |
| `customer_action` | `amount`, `currency` |

> **contents set properties**
>
> | Property | Type | Required | Description |
> | - | - | - | - |
> | `amount` | number | no | A whole number in minor units for the given currency (cents for US dollars). For example, \$129.99 = `12999`. Always pair it with `currency` |
> | `currency` | string | yes, if `amount` is present | An [ISO 4217](https://en.wikipedia.org/wiki/ISO_4217) currency code, such as `USD` |
> | `contents[]` | array | no | A list of items involved in the event. Each entry takes `id`, `name`, `content_type`, `quantity` (a whole number), `amount`, and `currency` |
>
> ```javascript Example event theme={null}
>   pfq('track', 'order_created', {
>     amount: 12999, currency: 'USD',
>     contents: [
>       {
>         id: 'SKU123',
>         name: 'Widget',
>         content_type: 'product',
>         quantity: 1,
>         amount: 12999,
>         currency: 'USD'
>       }
>     ]
>   });
> ```

> **plan\_enrollment set properties**
>
> | Property | Type | Required | Description |
> | - | - | - | - |
> | `plan_id` | string | no | Your internal plan identifier |
> | `amount` | number | no | A whole number in minor units for the given currency (cents for US dollars). For example, \$129.99 = `12999`. Always pair it with `currency`. |
> | `currency` | string | yes, if `amount` is present | An [ISO 4217](https://en.wikipedia.org/wiki/ISO_4217) currency code, such as `USD` |
> | `contents[]` | array | no | A list of items involved in the event. Each entry takes `id`, `name`, `content_type`, `quantity` (a whole number), `amount`, and `currency` |
>
> ```javascript Example event theme={null}
>   pfq('track', 'subscription_created', {
>     plan_id: 'pro-monthly',
>     amount: 4900,
>     currency: 'USD',
>     contents: [
>       {
>         id: 'pro-monthly',
>         name: 'Pro plan, monthly billing',
>         content_type: 'plan',
>         quantity: 1,
>         amount: 4900,
>         currency: 'USD'
>       }
>     ]
>   });
> ```

> **customer\_action set properties**
>
> | Property | Type | Required | Description |
> | - | - | - | - |
> | `amount` | number | no | A whole number in minor units for the given currency (cents for US dollars). For example, \$129.99 = `12999`. Always pair it with `currency`. |
> | `currency` | string | yes, if `amount` is present | An [ISO 4217](https://en.wikipedia.org/wiki/ISO_4217) currency code, such as `USD`. |
>
> ```javascript Example event theme={null}
>   pfq('track', 'lead_created', {
>     amount: 50000,
>     currency: 'USD',
>     form: 'contact-sales'
>   });
> ```

### Custom properties

Add your own custom properties (such as `order_id` or `product_id`) alongside the recognized fields. Two limits apply: up to 128 properties per event, and up to 64 KiB per request body. An event that exceeds either limit is rejected with a `400` error.

## Custom events

For anything outside the standard set, send a custom event: call `pfq('trackCustom', …)` with any event name and any properties you choose.

```javascript theme={null}
  pfq('trackCustom', 'requested_demo', { plan: 'enterprise' });
```

Maximum length for event names is 128 characters. The suggested naming convention is lowercase letters, numbers, underscores or dashes, and a name that doesn't match a [standard event](#standard-events) name, so your custom events are easy to tell apart in reporting.

> [!WARNING]
>
> **Never send customers' personal data in events.** This includes emails, phone numbers, names, government IDs, payment data, street addresses, and special-category data, such as health, financial account details, or biometrics. As a safety measure, Profound drops personal data keys and redacts values that look like personal data.

## Tie events to your own user IDs

If you need to match conversions to records in your own systems, make an `identify` call with your internal identifiers after the customer logs in and before any events are sent:

```javascript theme={null}
  pfq('identify', {
    external_user_id: 'u_123',
    external_account_id: 'acct_456'
  });
```

The following fields are accepted:

| Field | Type | Required |
| - | - | - |
| `external_user_id` | string | no |
| `external_account_id` | string | no |

Any other fields are silently ignored.

The pixel holds the IDs in memory until your consent signal allows sending, then attaches them to the events that follow.

> [!WARNING]
>
> Send internal IDs only, never customers' names, emails, phone numbers, or other identifiers. Personal data in the `identify` fields causes the whole event to be rejected with a `400` error.


# FAQ
Source: https://docs.tryprofound.com/pixel/faq

Frequently asked questions about Profound Pixel

> **Does Profound Pixel work only with OpenAI ads?**
>
> No. Profound Pixel measures AI-channel traffic broadly: it attributes conversions to paid OpenAI ads and to organic AI answer traffic, and it also records other acquisition channels, such as Google or Bing.

> **How large is the Profound Pixel tag?**
>
> It's a lightweight JavaScript tag, about 8KB compressed. It loads asynchronously and never blocks your page from rendering. Learn more in [How the pixel script works](/pixel/overview#how-the-pixel-script-works).

> **Is the pixel ID safe to expose in my web page source code?**
>
> Yes. The pixel ID identifies you, and it isn't a secret or token. Profound protects it on its side: each pixel ID has a list of allowed domains, and events from other domains are not processed. Learn more in [Your pixel ID](/pixel/overview#your-pixel-id).

> **What happens if I send personal data in the event payload?**
>
> Profound has protections in place if that happens:
>
> * The pixel identifies visitors by random IDs only.
> * In the browser, the pixel drops personal data keys, such as name or email, before sending the events.
> * On its servers, Profound drops a broader set of personal data keys and redacts values that look like personal data.
>
> However, even with those protections in place, don't send personal data in events.
>
> Learn more in the [Events reference](/pixel/events).

> **I have OpenAI's pixel on my site. Can I run Profound Pixel alongside it?**
>
> Yes, running both pixels works today. Profound doesn't forward conversions to OpenAI, so conversions are not counted twice.
>
> If you'd rather keep one pixel on your site, Profound Pixel can be set up to send your conversions to OpenAI for you. If you'd like that, speak to your Profound team. Learn more in [Forwarding conversions to OpenAI](/pixel/consent-and-privacy#optional-forwarding-conversions-to-openai).

> **Can I send conversions to Profound directly from my own servers instead of through the browser pixel?**
>
> Yes, a server-to-server integration is supported alongside the browser pixel. Speak to your Profound team about setting it up.


# Install Profound Pixel
Source: https://docs.tryprofound.com/pixel/install

A step-by-step guide to install and use your Profound Pixel

This guide walks you through everything you need to get started with Profound Pixel: getting your pixel ID, adding the code snippet to your website, setting up and verifying conversion tracking.

## Before you start

Have these things in place before any code goes on your site:

1. **Sign the data processing addendum.** The pixel sends Profound end-user data your current agreement doesn't cover. Reach out to your ads representative, and don't install until the addendum is signed.
2. **Tell your Profound team your domains.** Each pixel ID carries its own list of allowed domains, and events from a domain that isn't on the list are not processed. Share every hostname you'll install the pixel on.

## Installation steps

**1. Get your pixel ID**

In the Profound platform, go to **Ad Manager** (in the left sidebar, under **Ads**), select **Conversions** in the top-right corner, then under **Your pixels**, select **Generate pixel**.

> [!NOTE]
>
> Each pixel ID covers one site or environment.

**2. Add the snippet to your website**

Install the snippet manually or through Google Tag Manager.

**Manual installation**

Paste this code snippet into the `<head>` of every page you want to measure, and replace `<YOUR_PIXEL_ID>` with the pixel ID you retrieved in [step 1](#step-1):

```html expandable theme={null}
  <script>
    (function (w, d, s, u) {
      try {
        if (w.pfq) return;
        var q = function () { q.q.push([].slice.call(arguments)); };
        q.q = [];
        w.pfq = q;
        var js = d.createElement(s);
        js.async = true;
        js.src = u;
        js.onerror = function () { w.pfq = function () {}; };
        var f = d.getElementsByTagName(s)[0];
        if (f && f.parentNode) f.parentNode.insertBefore(js, f);
        else d.head.appendChild(js);
      } catch (e) {}
    })(window, document, "script", "https://pixel.profound.click/sdk/v1/p.js");

    pfq("init", "<YOUR_PIXEL_ID>");
    pfq("track", "page_viewed");
  </script>
```

Keep both `pfq` function calls. They do different jobs:

* `pfq("init", …)` sets up identity and consent handling and captures the ad click ID from the landing URL. It records nothing by itself.
* `pfq("track", "page_viewed")` records the page view. Without it, the page reports nothing.

**Placement rules**:

* Place the snippet on every page you want measured. A good practice is to add it to every page through your global layout or base template.
* Place it in `<head>`, as early as practical. The snippet must run before anything can rewrite the landing URL, to prevent the loss of ad click IDs.
* Place it once per page. The snippet ignores copies of itself (`if (w.pfq) return`), but the `page_viewed` line doesn't: copies count every page view multiple times.

> [!NOTE]
>
> **Single-page apps must record their own page views.** The pixel doesn't watch for route changes, so call `pfq('track', 'page_viewed')` once on each route change you want recorded.

**Google Tag Manager**

If you manage your site's tags in Google Tag Manager, install the pixel as a [Custom HTML tag](https://support.google.com/tagmanager/answer/6107167):

1. In Google Tag Manager, create a new tag and choose the **Custom HTML** tag type.
2. Paste in the code snippet from the **Manual installation** tab, and replace `<YOUR_PIXEL_ID>` with the pixel ID you retrieved in [step 1](#step-1).
3. Set the trigger to **Initialization - All Pages**.
4. Under **Advanced Settings** > **Tag firing options**, choose **Once per page**.
5. Under [Consent Settings](https://support.google.com/tagmanager/answer/10718549), choose **No additional consent required**.

> [!NOTE]
>
> The last two settings may look overly permissive, but they're intentional: the pixel manages consent itself, and firing it too late loses the ad click ID, so conversions can't be tied back to ads.

If your site is a single-page app, add a second tag that calls `pfq('track', 'page_viewed')` with a [History Change trigger](https://support.google.com/tagmanager/answer/7679322), so route changes are recorded as page views.

**3. Set your consent posture**

> [!TIP]
>
> Learn about how the pixel reads consent signals and what each setting does in [Consent and privacy](/pixel/consent-and-privacy#consent-handling).

The pixel starts in opt-in mode: it stores and sends nothing until a consent signal grants it.

If your site is in the UK, the EU, or the wider European Economic Area, no action is required. Your consent management platform (CMP) provides the consent signal for the pixel.

If you have a US-only site, you likely don't have a CMP, so you need to explicitly set the consent posture to opt-out for the pixel to be able to record events. Edit the `pfq("init", …)` line inside the snippet you added in [step 2](#step-2) so it passes the posture option:

```javascript theme={null}
  pfq("init", "<YOUR_PIXEL_ID>", { posture: "opt-out" });
```

**4. (optional) Allow the pixel in your Content Security Policy**

If your site enforces a strict [Content Security Policy](https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/CSP), it blocks the pixel until you allow Profound's domains.

Add these directives to your policy:

* [`script-src`](https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/script-src) `https://pixel.profound.click` allows the browser to load the pixel script.
* [`connect-src`](https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/connect-src) `https://events.profound.click` allows the pixel to send events.

The inline snippet from [step 2](#step-2) also needs a [nonce or hash](https://developer.mozilla.org/en-US/docs/Web/HTTP/Reference/Headers/Content-Security-Policy/script-src) on its `<script>` block, the same way you authorize your site's other inline scripts.

**5. Track your conversions**

The snippet from [step 2](#step-2) records page views on its own, but it can't know when a visitor becomes a customer. Your site reports that by calling the `pfq` function, which the snippet makes available on every page that carries it.

**Manual installation**

Add a `pfq('track', …)` call to your site's own code at the moment a conversion completes: on the order confirmation page, in the form's submit handler, wherever the conversion becomes definite. Use the standard [event name](/pixel/events#standard-events) that matches the conversion, and pass the details you have, for example:

```javascript theme={null}
  pfq('track', 'items_added',   { amount: 4900,  currency: 'USD', product_id: 'SKU123' });
  pfq('track', 'order_created', { amount: 12999, currency: 'USD', order_id: 'A-1001' });
  pfq('track', 'lead_created',  { form: 'contact-sales' });
```

**Google Tag Manager**

Create one more [Custom HTML tag](https://support.google.com/tagmanager/answer/6107167#custom-html-tag) per conversion type, and fire it on the trigger that marks that conversion on your site. Prefix each call with `window.pfq &&` so the tag does nothing if the install snippet's tag from [step 2](#step-2) never ran. For example:

```html theme={null}
  <script>
    window.pfq && pfq('track', 'order_created', {
      order_id: {{DLV - transaction_id}},
      amount: Math.round(Number({{DLV - value}}) * 100),
      currency: {{DLV - currency}},
    });
  </script>
```

**Notes**:

* The `{{DLV - …}}` variable names given above are examples only. Use the variable names defined in your Google Tag Manager container. Learn more in the [Data Layer Variables documentation](https://support.google.com/tagmanager/answer/7683362).
* The `* 100` multiplier in the code example only suits currencies with two decimal places, such as USD or EUR. For other currencies, match the multiplier to the currency's number of decimals.

> [!TIP]
>
> For the detailed breakdown of events and their fields, see the [Events reference](/pixel/events).

**6. Verify it works**

Complete a test conversion on your own site (for example, a test purchase). In your browser's network tab, filter for `events.profound.click`.

Locate the event request produced by your test conversion and check the following:

1. It's a `POST` request with a `202` status response.
2. The `amount` field in the payload body is a whole number in minor units (the smallest unit for the given currency): for example, an amount of 129.99 US dollars should show as `12999`. If the amount you see is in different units, go back to the tracking code you added in [step 5](#step-5) and correct the units it sends.
3. If your site receives European/UK traffic, test the conversion as a European visitor. Until you accept the consent banner, the pixel should send no requests and write no cookies. After you accept, requests to `events.profound.click` should appear in the network tab, and the pixel's cookies should appear in your browser's cookie storage.

   > [!TIP]
   >
   > Simulate a visit to your site from Europe by using a VPN, or by setting an EU-based location override in your browser's DevTools.

After completing these steps, your pixel should be set up and ready to go. If you have issues during setup, see [Troubleshooting](/pixel/troubleshooting) for the most common issues, or [contact support](mailto:support@tryprofound.com).


# About Profound Pixel
Source: https://docs.tryprofound.com/pixel/overview

A script that ties conversions on your site back to the AI traffic that drove them

> [!NOTE]
>
> Profound Pixel and Ad Manager are beta features. We're gradually rolling them out to customers. Follow [the Profound blog](https://www.tryprofound.com/blog) for updates.

Profound Pixel is a lightweight JavaScript tag that records conversions on your site and attributes them to the AI-channel traffic that drove them. This covers both paid ads and organic AI answer traffic.

## Key facts

* **The tag follows the standard pixel pattern.** The install shape matches other major pixel providers', and the events use [the same names as OpenAI's](https://developers.openai.com/ads/measurement-pixel).
* **The async loader never blocks rendering.** If the Profound content delivery network is unreachable or the browser version is too old, the pixel goes inert and doesn't throw errors on your page.
* **The pixel doesn't ship a cookie banner.** It reads the consent signal your existing consent management platform emits and only fires events when consent is granted. See [Consent and privacy](/pixel/consent-and-privacy).
* **No package to install, no code to bundle.** Profound pre-builds the script in its continuous integration pipeline and hosts it on its own domain, `pixel.profound.click`.

## How the pixel script works

To use the pixel, you paste a code snippet into your site and add function calls that record [events](/pixel/events). Learn more in the [installation guide](/pixel/install).

Profound hosts the pixel script on its own domain. When a page loads, the snippet fetches the pixel script (`p.js`) from `https://pixel.profound.click/sdk/v1/`.

> [!NOTE]
>
> Profound's build process enforces the script's size limit: under 8 KB compressed.

Once the script runs, it reads your site's [consent signal](/pixel/consent-and-privacy) and captures any ad and campaign IDs from the URL the visitor landed on. It then processes the events the page reports: the page view that the code snippet records on every page load, plus any conversions your code tracks. The pixel sends every event that consent allows to `https://events.profound.click/api/v1/events`. Profound records the event there and attributes it to the AI ad or AI answer that brought the visitor.

> [!TIP]
>
> Serving the script from your own domain (for example, `pix.yourbrand.com`) is available as an optional enterprise add-on. Ask your Profound team if you need it.

## Your pixel ID

Get your pixel ID from the **Ad Manager** page in the Profound platform. Each pixel ID covers one site or environment, and your account can hold more than one. The ID is safe to expose in page source: it identifies you, it isn't a secret or an auth token, and it never needs rotating.

Profound enforces integrity on its side. Each pixel ID has an allowlist of origins that should include every hostname you'll install the pixel on: the root domain (`example.com`), the `www` variant (`www.example.com`), and any checkout, landing-page, or regional subdomain.

> [!TIP]
>
> Follow the [installation guide](/pixel/install) to install Profound Pixel.


# Troubleshooting
Source: https://docs.tryprofound.com/pixel/troubleshooting

Common Profound Pixel issues, their causes, and fixes

## The events report the wrong amounts

Your code likely sends `amount` in major units (dollars), but Profound reads the value as minor units (cents). GA4-style dataLayers carry major units, so this is common with Google Tag Manager.

Send `amount` as a whole number in the currency's smallest unit: `12999` for \$129.99, not `129.99`. In Google Tag Manager, convert the dataLayer value:

```javascript theme={null}
Math.round(Number({{DLV - value}}) * 100)
```

Replace `{{DLV - value}}` with a [Data Layer Variable](https://support.google.com/tagmanager/answer/7683362) name defined in your Google Tag Manager container. If your currency doesn't use two decimal places, adjust the `* 100` multiplier to match its number of decimals.

## Conversions aren't attributed to ads

This issue can be caused by the pixel snippet running too late.

If you installed the snippet manually, check that it's placed directly inside the `<head>` tag of your HTML page and loads unconditionally.

If you used Google Tag Manager, make sure that:

* the **Consent Settings** in your tag are set to **No additional consent required**, as per [step 2 of the install guide](/pixel/install#step-2)
* the trigger is set to **Initialization - All Pages**

Learn more in [step 2 of the pixel installation guide](/pixel/install#step-2).

## Page views are missing on a single-page app

The pixel doesn't watch for route changes, so it records one page view per full page load.

Call `pfq('track', 'page_viewed')` on each route change you want recorded. In Google Tag Manager, use a [History Change trigger](https://support.google.com/tagmanager/answer/7679322).

## Page views are counted twice

If you installed the snippet manually, check that the snippet is placed only once per page.

If you used Google Tag Manager, check that **Advanced Settings** > **Tag firing options** is set to **Once per page**, as per [step 2 of the install guide](/pixel/install#step-2).

## Events return a `202` but don't appear in reporting

The events are likely rejected silently: with origin enforcement on, an unknown pixel ID or an origin missing from the allowlist returns the same `202` as an accepted event.

Check that the snippet has the correct pixel ID, and give your Profound team every origin you send events from: the root domain, the `www` variant, and any checkout or landing subdomains.

## The pixel records no events at all

This can happen when the pixel is waiting for a consent signal that never comes.

If your website doesn't use a consent management platform (common for US-only sites), set the consent posture to opt-out as instructed in [step 3 of the installation guide](/pixel/install#step-3) and redeploy the snippet.

## An event is rejected with a `400` error

Some possible reasons for this are:

* Profound detected personal data in the [`identify`](/pixel/events#tie-events-to-your-own-user-ids) fields. Check that they carry only internal identifiers.
* The event has more than 128 properties.
* The request body of the event exceeds 64 KiB.

Learn more about the event data limitations in the [Events reference](/pixel/events).

## Properties or IDs are missing from recorded events

This can be one of the following:

* The property key or value looks like personal data, so it was dropped in the browser or on Profound's servers.
* If an [`identify`](/pixel/events#tie-events-to-your-own-user-ids) field is missing, check if the property key is correctly typed.

Learn more in the [Events reference](/pixel/events).

## Attribution stops after about 7 days in Safari

Safari's Intelligent Tracking Prevention limits script-written cookies to roughly 7 days. This is browser behavior that can't be changed. Using a different browser will fix the issue.

## The opt-out consent posture behaves as opt-in for some visitors

The pixel checks each browser's timezone and treats every visitor with a European timezone as opt-in: it records nothing until they give consent.

If your pixel is scoped to a single non-EU region, or your local data protection regulations otherwise allow the opt-out posture, you can disable the check by adding `detectRegion: false` to the `pfq('init', ...)` call in your install snippet:

```javascript theme={null}
pfq("init", "<YOUR_PIXEL_ID>", { detectRegion: false });
```

Learn more in [Consent and privacy](/pixel/consent-and-privacy#sites-without-cmp).


# About Prompt Volumes
Source: https://docs.tryprofound.com/prompt-volumes/about

Explore real prompts that users submit to AI answer engines

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Analytics > Prompt Volumes

Prompt Volumes give you direct access to Profound's dataset of real prompts submitted to AI answer engines by actual users. Use it to discover what questions people are actually asking ChatGPT, Gemini, and Perplexity about topics relevant to your business, so you can validate your tracking strategy, understand user language, and prioritize content based on real signal.

## The problem with traditional keyword research

When someone wants to find the best corporate credit card option, they used to Google "best corporate credit card." Now they ask their preferred AI chat: "What's the best corporate credit card for small teams?"

Traditional keyword research tools capture what people type into search bars, not how they hold conversations with AI. The language is different, the intent is different, and the questions are longer and more specific.

Without visibility into real AI prompting behavior, teams face three compounding problems:

* **Misdirected tracking:** monitoring prompts that reflect SEO assumptions, not actual user questions.
* **Misaligned content:** writing for search engines while users are talking to AI assistants.
* **Missed opportunities:** finding out about rising topics only after competitors have already moved on them.

## Profound's solution

Prompt Volumes is a proprietary dataset built by aggregating, cleaning, and modeling real conversations between users and AI answer engines. It transforms raw AI conversation data into structured, searchable insights about what your market is actually asking, so you can:

* Validate whether the topics you're tracking reflect **real user demand**
* **Discover the exact phrasing** people use when asking about your category
* **Understand user intent** across informational, commercial, generative, and conversational queries
* **Identify** which **regions** are driving specific queries
* **Prioritize** content and tracking **strategy** on signal, not assumption

## How Prompt Volumes work

Prompt Volumes is a dataset built by licensing AI answer engine conversations, cleaning the data, applying probabilistic modeling, and visualizing the results as intuitive charts and tables your team can act on.

> [!NOTE]
>
> All data is anonymized, aggregated, scrubbed of PII, and fully compliant with GDPR and CCPA.

**Data sources:** ChatGPT, Gemini, and Perplexity.

## Key features

### Keyword search and Bulk analysis

Enter a short, broad keyword to query the dataset and see how often that topic appears across real AI conversations. You can also compare multiple keywords at once using Bulk analysis (available on the Enterprise plan), which ranks topics by volume and helps you prioritize what to track before you invest in it.

Keywords can be searched using phrase match (any word order) or exact match (specific word order). Start broad and use the results to narrow down into more specific language.

### Volume and trend visualization

The **Overview** tab shows how prompt volume for your keyword changes over time, alongside a breakdown of conversations by intent and a list of related terms worth exploring. Use it to spot rising topics and understand the specific language your audience is using.

### Intent analysis

The Conversation Share by Intent chart on the Overview tab breaks down why people are asking about a topic, classifying prompts into informational, commercial, generative, and conversational categories. This helps you match content to where users are in their journey.

### Regions

The Regions tab shows where prompts about your keywords come from, as a heat map and a per-region volume table. Note that coverage is strongest in the US, UK, and select European markets.

### Keyword lists

Organize related keywords into named lists to track topic clusters over time and keep your research organized by project or audience segment.

### Relevant prompts sample

Discover a sample of real user prompts that caused an AI answer engine to cite a specific page on your (or a competitor's) domain in its response. Compare your citations with competitors and explore which exact URLs on your site are being cited and for what kinds of questions.

### Prompt Research Reports

Prompt Research Reports provide data-driven gap analysis, comparing what real users are asking AI engines against what you're currently tracking in Answer Engine Insights. Use them to get a data-backed starting point, surface what you may be missing, or understand where your customers are in their purchasing journey.

## Historical data

Prompt Volumes historical data availability depends on your region and platform. For the US, ChatGPT, Gemini, and Perplexity data goes back to January 2025. For all other regions Profound currently tracks, data is available from January 2026.

This means you can query any keyword (even one you just added) and view historical volume trends from **January 2025 onward** for the US, or **January 2026 onward** for other regions.

> [!NOTE]
>
> Data coverage varies by region. For example, in the US, data is currently available for ChatGPT, Gemini, and Perplexity, while Norway only has ChatGPT.


# Prompt Volumes FAQ
Source: https://docs.tryprofound.com/prompt-volumes/faq

Frequently asked questions about Prompt Volumes

> **Where does the Prompt Volumes data come from?**
>
> Profound licenses data from well-known, established double opt-in consumer panel providers. These panels consist of millions of active AI assistant users who have opted in to have their behavior tracked and anonymized. Profound receives tens of millions of real prompts each month from these panels, cleans the data, applies probabilistic modeling to correct for demographic and geographic biases, and produces the insights visible in the platform.
>
> All data is fully compliant with CCPA and GDPR, and Profound also applies additional anonymization and aggregation on top of vendor-level protections.

> **Do you use synthetic or simulated data in the dataset?**
>
> No. Every prompt in the dataset comes from real users interacting with AI answer engines. Synthetic or simulated data is not used at any point.

> **How fresh is the data?**
>
> Prompt volume data is updated on a rolling weekly basis, with a data latency of less than two weeks. This means the trends you see in Prompt Volumes reflect near-current conversations.

> **How far back does the data go?**
>
> This depends on your region and the platform in question. Generally, for the US it's January 2025, and for all other regions Profound currently tracks, it's January 2026.
>
> To learn more about data coverage, see the **Historical data** section in [About Prompt Volumes](/prompt-volumes/about) article.

> **Which AI platforms are included?**
>
> ChatGPT, Gemini, and Perplexity data is available. Coverage per country reflects where each platform is available, so it may vary by region. Get in touch with your Profound representative to learn more about what's available to you.
>
> Additional platforms are in scope for future expansion.

> **Why does the data primarily show ChatGPT usage?**
>
> ChatGPT typically accounts for the largest share of prompt volume. For date ranges before January 2026, only US data is available, so earlier periods reflect US usage only. For more recent date ranges, data is available across all platforms and covered regions.

> **Are all regions covered?**
>
> No. Coverage is strongest in the US, UK, and select European markets.

> **Can I see rising topics and trends over time?**
>
> Yes. The Prompt Volumes chart shows how keyword volume changes over time so you can spot rising topics and seasonal shifts. The Keyword Hierarchy tab shows how a topic expands into related terms, and the Similar Keywords table surfaces adjacent terms that commonly appear alongside your keyword.

> **Where do tracked prompts come from?**
>
> You create them yourself, either manually or with AI-assisted suggestions based on your brand and topics. Prompt Volumes helps you write better prompts by showing you the actual language real users are submitting to AI engines in your category, so your tracked prompts reflect genuine market language.

> **Why aren't my keyword phrase searches returning results?**
>
> Prompt Volumes match keywords across AI conversations, so volume scales inversely with specificity: single words return the broadest results, while full sentences return very few. The sweet spot is typically 2–4 words that capture the core intent. Use Keyword Hierarchy and Similar Keywords to find related terms and drill into subcategories.

> **What's the difference between Prompt Volumes and the Prompts table in Answer Engine Insights?**
>
> The Prompts table in Answer Engine Insights shows the specific prompts you're actively tracking and how your brand performs against them. Prompt Volumes shows the full landscape of what users are asking AI about any topic, including questions you haven't started tracking yet. Use Prompt Volumes to decide what goes into your Prompts table.


# Get started with Prompt Volumes
Source: https://docs.tryprofound.com/prompt-volumes/getting-started

The quickest way to start searching and analyzing Prompt Volumes

This is a short guide on the quickest way to start using and analyzing Prompt Volumes. More detailed how-to guides are available in [Analyze Prompt Volumes](/prompt-volumes/how-tos/analyze-prompt-volumes), [Keyword searches and lists](/prompt-volumes/how-tos/keyword-searches-and-lists), and other articles in this section.

## Step 1: Analyze a prompt

The simplest way to get started is with a keyword or phrase analysis:

1. Open the **Prompt Volumes** section in the left sidebar navigation.
2. Enter a starting keyword or phrase (for example, "AI search optimization" or "business credit cards") in the text input at the top of the page and click **Analyze**.
3. Toggle **Bulk analysis** to compare multiple topics and rank by volume (available on the Enterprise plan).
4. Prioritize keywords with meaningful volume. If you see no volume, go broader until you get signal.

Usually, the keywords or phrases for analysis come from prompt topics you defined in **Answer Engine Insights**.

## Step 2: Review your results

After clicking **Analyze**, you'll land on a results page that includes:

* **Prompt Volume chart** — how often your keyword appears across AI platforms
* **Conversation Share by Intent** — why people are prompting about your keyword
* **Similar Keywords** — related terms worth exploring

Switch to the **Regions** and **Keyword Hierarchy** tabs to see where prompts are coming from and how terms relate to your keyword.

## Step 3: Create a keyword list

Once you find useful keywords or phrases, group them into lists for ongoing tracking:

* From the Prompt Volumes landing page, click **+ Create List**
* Add or remove keywords directly from the list view
* Click any keyword in the list to view its performance

## Step 4: Revisit regularly

User phrasing evolves as AI adoption grows. Check Prompt Volumes regularly to catch rising topics, shifting language, and emerging intent patterns before they become table stakes.


# Analyze prompt volumes
Source: https://docs.tryprofound.com/prompt-volumes/how-tos/analyze-prompt-volumes

Read the Overview, Regions, and Keyword Hierarchy tabs and apply filters in Prompt Volumes

Submitting a keyword query, or selecting a keyword from a list, redirects you to the corresponding results within the Prompt Volumes dataset. From there, you can review several charts and tables that show how frequently your keywords appear across different answer engine platforms. These visualizations are organized into the following tabs: Overview, Regions, Keyword Hierarchy.

## Overview tab

Three main charts and tables provide an overview of your keyword query results. Use these to understand the volume of prompts that contain your queried keywords.

### Prompt Volume

The Prompt Volume chart shows how many keyword matches were found in the prompt dataset.

Adjust the visualization using the Chart Config controls above the chart. Display results as either a line or bar chart, and group results by platform or show them without grouping.

![pv\_7.png](https://mintcdn.com/profound-37face47/wPfOqxyNzo8IXjy3/images/prompt-volumes/how-tos/analyze-prompt-volumes-1.png?fit=max\&auto=format\&n=wPfOqxyNzo8IXjy3\&q=85\&s=f9a32c54d8c65e55633c134bdbff4063)

### Conversation Share by Intent

The Conversation Share by Intent chart shows the share of conversations by intent, based on a representative sample of user conversation data over the selected time range. Profound classifies intent as informational, commercial, conversational, or generative (requests to create content).

![pv\_8.png](https://mintcdn.com/profound-37face47/wPfOqxyNzo8IXjy3/images/prompt-volumes/how-tos/analyze-prompt-volumes-2.png?fit=max\&auto=format\&n=wPfOqxyNzo8IXjy3\&q=85\&s=9b7a96b1e9d9c0ef8bc71fea57fb5d3e)

Switch between the sunburst chart and a table view using the toggle above the chart.

### Similar Keywords

The Similar Keywords table displays keywords that commonly appear in similar prompting patterns. Select any keyword in this list to run a new query for that term.

## Regions

The Regions tab displays keyword query results broken down by region.

The Prompt Volume by Region chart is a heat map showing the volume of prompts containing the queried keyword across geographic regions. A table to the right lists the exact prompt counts for each region.

## Keyword Hierarchy

The Keyword Hierarchy tab contains an interactive hierarchy chart. The queried keyword heads the chart. Subsequent levels show expanded keywords and their corresponding prevalence in the results.

![pv\_9.png](https://mintcdn.com/profound-37face47/wPfOqxyNzo8IXjy3/images/prompt-volumes/how-tos/analyze-prompt-volumes-3.png?fit=max\&auto=format\&n=wPfOqxyNzo8IXjy3\&q=85\&s=2632a0e895588d91936ad5db2f57c111)

Adjust the number of keywords per level and depth of the hierarchy using the number stepper on the left side of the chart.

## Filters

Filters are available across Prompt Volumes. They let you adjust the analytical time frame and narrow the data view. Filters appear as dropdown menus beneath the tabs.

![pv\_6.png](https://mintcdn.com/profound-37face47/wPfOqxyNzo8IXjy3/images/prompt-volumes/how-tos/analyze-prompt-volumes-4.png?fit=max\&auto=format\&n=wPfOqxyNzo8IXjy3\&q=85\&s=1024dc67032d8afcda926c3d8f7051ee)

### Date Range filter

Use the date range filter to select the time window for the prompt data. Choose **All Time**, switch between **Weeks** and **Months**, or turn on **Range** to define a custom range using the calendar.

Only prompts submitted during the selected period are counted in the Prompt Volumes charts.

Charts display differences by calculating the change between the selected time frame and a previous period. Adjust this comparison period using the **Prev. Period** dropdown.

### Attribute filters

The **Platforms** and **Regions** filters let you narrow chart data. While each chart includes filters relevant to its context, filter functionality is consistent across charts.

To adjust filters, open the dropdown for the desired attribute and select any platform or region to include in the chart view.


# Keyword searches and lists
Source: https://docs.tryprofound.com/prompt-volumes/how-tos/keyword-searches-and-lists

Query the Prompt Volumes dataset with keywords and save results as lists

There are several ways to query the Prompt Volumes dataset. Profound scans prompts for keywords and generates data points when matches are found. Effective keyword creation and management are essential for producing meaningful query results.

## How to use keywords to query Prompt Volumes data

The Prompt Volumes search bar is the fastest way to query the Prompt Volumes dataset. Enter a keyword into the search bar and click **Analyze** to go directly to the query results.

![pv\_1.png](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/prompt-volumes/how-tos/keyword-searches-and-lists-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=75a4882f0b27f4c9bbc0354ee5c224d3)

### Bulk analysis

The **Bulk analysis** option lets you input multiple keywords into the search and run queries against the dataset using all of them at once. Bulk analysis is available on the Enterprise plan.

### Phrase matching

Profound supports two matching modes. **Exact matching** returns only prompts containing the exact keyword string. **Phrase matching** scans for prompts containing the keywords in any order.

For example, if the keyword is "ai tools," exact matching excludes a prompt like "what tools are good in the ai space?" With phrase matching enabled, that prompt appears in the results since both "ai" and "tools" are present.

## How to create keyword lists

Group related keywords into lists to track how specific initiatives or strategies perform across answer engines. This lets you analyze the collective performance of a set of terms.

From the Prompt Volumes landing page, select **+ Create List** to create a new list.

Add or remove keywords directly from the list view. Click any keyword in the list to view its performance within the Prompt Volumes dataset.

![pv\_5.png](https://mintcdn.com/profound-37face47/wPfOqxyNzo8IXjy3/images/prompt-volumes/how-tos/keyword-searches-and-lists-2.png?fit=max\&auto=format\&n=wPfOqxyNzo8IXjy3\&q=85\&s=627ceb23d445b76ef8672966f5807f59)


# Using Prompt Volumes with Workflows
Source: https://docs.tryprofound.com/prompt-volumes/using-with-workflows

Pull Prompt Volumes data into Workflows with the Prompt Volumes node

Prompt Volumes data can be pulled directly into Profound's [Workflows](/agents/workflows/about-workflows) using the **Prompt Volumes node** in the Workflow Builder.

## What the Prompt Volumes node does

The **Prompt Volumes node** retrieves keyword volume data from your Prompt Volumes dataset and surfaces it as structured output within your Workflow. Use this node when you want to:

* Pull volume trends for specific keywords as part of an automated content strategy workflow
* Feed keyword frequency data into a [Prompt LLM](/agents/workflows/nodes/prompt-llm) node for analysis, prioritization, or content brief generation
* Monitor keyword volume changes over time as part of a recurring reporting Workflow
* Incorporate real-world prompt demand data alongside your Answer Engine Insights metrics in a single automated workflow

## Passing Prompt Volumes output to other Workflow nodes

Like all Profound Data nodes, the ​ **Prompt Volumes node** produces a named output variable you can pass into any other node that accepts text or structured data inputs.

For example, a [Prompt LLM](/agents/workflows/nodes/prompt-llm) node can summarize trending topics, surface content gaps, or draft briefs. A [Slack](/integrations/slack/connect-slack-to-profound) node can deliver keyword insights directly to your team.

Don't be afraid to experiment with different node combinations to find the workflows that work best for your team.


# About Shopping
Source: https://docs.tryprofound.com/shopping/about

See how your products appear inside AI-powered shopping experiences

> [!NOTE]
>
> **Available on:** the Enterprise plan
>
> **Find it in:** Analytics > Shopping

Shopping Analysis gives retail and e-commerce brands a clear view of how their products appear inside AI-powered shopping experiences. Profound captures the full AI response: product tiles, images, prices, merchant links, and buy buttons. This lets you see exactly which products are featured, which retailers own the checkout, and how your brand compares to competitors.

Shopping Analysis is currently focused on ChatGPT Shopping. Coverage of additional AI commerce surfaces, including Perplexity and Gemini, is evolving. Shopping Analysis is most relevant for retail, e-commerce, and CPG brands with physical or digital products.

> [!NOTE]
>
> **Shopping mode** is a feature that activates in an AI chat when a prompt signals purchase intent. Instead of a text answer, AI returns a visual carousel of product tiles.
>
> **Product tiles** are the individual card-style UI components that display a product in AI Shopping Mode. Click a tile to see pricing and purchase options, and additional information sections such as "What to know" and "What people are saying," depending on the product.

## AI shopping is a blind spot for brands

More customers are starting their purchase journey by asking an AI rather than visiting a website. When they do, the AI recommends specific products, describes them, and links to merchants, all without any input from brands.

Brands have no visibility into which products are recommended, how they're described, or which competitors appear. When a product does appear, the checkout link may route to a third-party retailer rather than the brand's own store. Because AI responses surface individual SKUs, a brand can appear to have strong category visibility while quietly losing product slots to competitors.

## How Profound fills the visibility gap

Profound Shopping Analysis treats AI shopping surfaces the same way behavioral analytics tools treat a website: as a measurable, improvable channel. The platform runs your tracked prompts through an LLM, monitors which ones trigger a shopping mode response, and captures the full result when they do. You get product-level data on placement, merchant attribution, attribute accuracy, and competitive positioning. This is the same depth of insight you previously got from web analytics, now applied to conversational commerce.

## Key features

### Shopping Mode Rate

Prompts that reliably trigger AI shopping mode typically combine a baseline commercial intent with one or two buyer profile factors, such as price range, gender, or preference. For example:

* Best running shoes under \$150
* What refrigerator should I buy with an in-door ice machine
* Affordable standing desk for small apartments

Shopping Mode Rate shows what percentage of your tracked prompts trigger the shopping mode answer, rather than a standard text answer. Check it in **Shopping > Brands > Shopping Mode Rate** to understand how much signal you’re working with before reading any other data in the module.

### Shopping Visibility Score

The Visibility Score tracks how consistently your brand and products appear across shopping mode responses. A score of 100% means your brand appears every time AI triggers shopping mode for that prompt set. The score breaks down by topic, and visibility can swing significantly between topics for the same brand, so topic-level review is more revealing than the overall figure.

### Merchant layer

When your product appears in an AI Shopping tile, it typically includes multiple checkout links: some from third-party retailers, some from your own store. The merchant layer shows who those retailers are, how often they appear, and what share of their recommended products belong to your brand. Use this to understand whether your direct-to-consumer store is competitive at checkout or whether third-party retailers are capturing most of the traffic.

### Product layer (SKU-level analysis)

The product layer tracks individual SKUs, not just brand mentions. You can see which specific products appear most in shopping results, which prompts surface them, and how AI describes their attributes. Competitors or unrelated products may be winning product placements in your category, and this view surfaces exactly where those gaps exist.


# Shopping FAQ
Source: https://docs.tryprofound.com/shopping/faq

Frequently asked questions about Shopping analysis

> **How does Profound track product appearances in ChatGPT Shopping?**
>
> Profound runs your tracked prompts through ChatGPT and checks whether each one triggers shopping mode. When it does, the full response is captured and analysed: product tiles, images, prices, merchant links, and product attributes. The data is structured and surfaced in Shopping Analysis, broken down by brand, merchant, product, and topic.

> **Why is my Shopping Mode Rate low?**
>
> A low Shopping Mode Rate usually means your tracked prompts don't signal strong commercial intent to AI. Terms like "buy," "shop," "where to find," or references to price and budget tend to trigger shopping results. Reviewing your prompt set for commercial intent signals is the most effective way to improve the rate.

> **Why don't some retailers appear in shopping feeds?**
>
> Shopping Analysis reflects what AI actually returns in response to your tracked prompts. Regional or niche retailers that don't appear in shopping responses for your category won't appear in the data. Coverage is not a complete inventory of all retailers in your market.

> **Why does a competitor appear in my shopping results when they have no organic visibility?**
>
> AI shopping results are not determined by organic search rankings. AI may surface a competitor's product based on their product feed quality, merchant relationships, structured data, or attribute signals that are independent of their search engine optimization (SEO) performance. Shopping Analysis helps you identify exactly this pattern: products that win AI shopping placements without traditional search visibility.

> **What content strategy helps with AI shopping visibility?**
>
> Profound's research found that AI shopping results lean on community discussion, consumer editorial, and review-heavy sources rather than B2B analyst content. Content optimised for AI shopping should be consumer-facing, review-oriented, and community-validated. This is a different content mix from what drives standard Answer Engine Optimization visibility.

> **Can Shopping Analysis handle large product catalogues?**
>
> Shopping Analysis works at the SKU level and can surface which specific products appear in AI shopping results across a large catalogue. The platform is designed to handle retail-scale product sets. If you have questions about catalogue size or configuration for a very large product feed, speak with your customer success manager.


# Get started with Shopping
Source: https://docs.tryprofound.com/shopping/getting-started

Navigate the Brands, Merchants, Products, and Prompts tabs in Shopping

> [!TIP]
>
> To learn more about making the best use of Shopping Analysis, check out the Profound University module [Tracking Your Products in AI Shopping](https://university.tryprofound.com/courses/profound-101/modules/analyze/lessons/tracking-your-products-in-ai-shopping).

The Shopping Analysis section is organized across four tabs: Brands, Merchants, Products, and Prompts.

## Step 1: Start with the Brands tab

The **Brands** tab shows which brands appear in shopping carousels across your tracked prompts.

Check your **Visibility Score** to see how often your brand appears and how it ranks against competitors.

Review your **Shopping Mode Rate**. If it's very low (for example, below 5%), your tracked prompts may lack commercial intent and data will be sparse. Filter by topic to find where your brand dominates and where gaps exist.

## Step 2: Follow the checkout in the Merchants tab

The **Merchants** tab shows who owns the purchase transaction when your products appear in shopping mode.

Review **Merchant Share** rankings to see which retailers capture the most checkout options. Your direct-to-consumer store may rank below third-party retailers.

Check **Brand Share by Merchant**. A dominant merchant with low brand share means your product is competing at checkout.

## Step 3: Review SKU performance in the Products tab

The **Products** tab tracks which individual products appear in shopping mode responses.

Review which SKUs appear most frequently in **Product Visibility** view. The top products in your category may not be yours.

**Click into a product** to see how often it appears, which prompts surface it, and how AI describes its attributes.

## Step 4: Diagnose queries in the Prompts tab

The **Prompts** tab shows which specific tracked prompts return your brand's products in shopping mode. Use it to identify the commercially oriented queries where your products appear in the carousel, and to find prompts where shopping mode triggers but your brand does not appear.


# Agency Mode overview
Source: https://docs.tryprofound.com/agencies/agency-mode-overview

Manage multiple brands or client workspaces under a single Profound account

[https://player.mux.com/jdABwUsFg3lhFGxBxH7QbTAxZrxcYOw200PXBg3RHWVU?default-hidden-captions=true\&thumbnail-time=2](https://player.mux.com/jdABwUsFg3lhFGxBxH7QbTAxZrxcYOw200PXBg3RHWVU?default-hidden-captions=true\&thumbnail-time=2)

Agency Mode is designed for organizations that manage multiple brands or client workspaces under a single account. It allows agencies to create, configure, and manage separate brand environments, enabling focused brand analysis for multiple brand assets.

Each brand workspace operates independently, with its own configuration, data, and lifecycle. This makes it possible to run short-term audits, or pitches, for prospective clients. It also addresses upsell opportunities that agencies have while also maintaining long-running workspaces for an agency’s active clients.

> [!NOTE]
>
> Agency Mode is for marketing agencies and consultancies monitoring multiple brands and clients. This is distinct from Profound's [Agent Analytics](/agent-analytics/overview), which measures how answer engines assess your web presence.

## Agency Mode workspaces

Agency Mode supports two primary types of workspaces: short-term pitch workspaces, and long-term client workspaces.

### Pitch workspaces

Pitch workspaces are designed for short-term evaluations. Agencies use them to assess a prospective client’s AI search visibility, identify AEO growth and upsell opportunities, or prepare materials for other types of sales and business development conversations.

A pitch workspace can be set up by agency Profound users [by entering details about the client and any relevant context for the prompts](/agencies/manage-brands#create-a-new-pitch). Profound uses this information to automatically generate industry-specific prompts and launch the workspace. Data generation typically takes 24-48 hours after workspace creation or when new prompts are added. Agency users can also upload a CSV with their own prompts if they prefer that to Profound prompt generation. After uploading prompts via CSV, it takes 48 hours to start seeing data for those new prompts in the workspace.

A pitch workspace can be [promoted](/agencies/manage-brands#promote-a-pitch-workspace-into-a-client-workspace) to a full client workspace if the agency wins ongoing AEO business or wants to monitor a brand workspace on a long-term basis. To keep workspaces organized, pitch workspaces expire after a defined period (typically 30 days). Before expiration, agencies can extend the workspace for another 30 days using a credit, or promote it to a full client workspace by purchasing one from Profound. If no action is taken, the workspace expires automatically at the end of the designated time period.

Expired pitches move to Inactive Workspaces, where historical data remains available in read-only mode for agency users to access later if they need it. Inactive workspaces can be reactivated later for an additional credit use. This can be done if business resumes with a prior prospect.

### Active client workspaces

Active client workspaces support ongoing client work. These workspaces are configured for longer durations and are built for continuous monitoring, analysis, and content optimization. This gives agencies the ability to collect data and measure impact of AEO work on clients' AI search success metrics.

Each client workspace maintains its own prompt set, data stream, competitors, LLMs tracked, and configuration. This allows agencies to manage multiple clients in parallel without overlap. Client workspaces last for the duration of the agency's contract with Profound or an alternative, agreed-upon time period. Client workspaces are typically larger in size and have more prompts, LLMs, regions, and competitors tracked when compared to pitch workspaces.

### Inactive workspaces

Inactive workspaces are workspaces no longer operating. These are primarily pitch workspaces that have expired after their set duration. Pitch workspaces are automatically moved to the inactive section of the agency’s landing page, below the pitch workspaces and client workspace tables. If an agency lets a Pitch workspace expire and needs to reactivate it at any point, they can click the “Start a pitch” button on that inactive workspace to use a credit to reactivate it and begin collecting data for another time increment (typically 30 days).

## Agency Mode dashboard

All pitch and client workspaces are managed from the Agency Mode dashboard. From this dashboard, all agency users can view active workspaces, see usage of their Profound contract credits, and track pitch workspace expiration timelines. Admin users from the agency also have the ability to create new pitch workspaces or client workspaces if they have workspace credits available on their account.

![Agency Mode Client Management dashboard with Pitch Workspaces, Client Workspaces, and Inactive sections and usage counters](https://mintcdn.com/profound-37face47/J_F2xMhMTt8_rix9/images/agencies/agency-mode-overview-1.png?fit=max\&auto=format\&n=J_F2xMhMTt8_rix9\&q=85\&s=9045a0f2476b233dee65ec9e5a428772)

Pitch workspaces can be [extended to move](/agencies/manage-brands#extend-or-end-a-pitch) past expiration. Alternatively, pitch workspaces can [be promoted](/agencies/manage-brands#promote-a-pitch-workspace-into-a-client-workspace)into a full brand workspace when a prospect becomes a client.

Changes made to any brand workspace do not affect other client workspaces. This separation reduces operational risk and facilitates scaling as the agency’s client base grows.

> [!NOTE]
>
> Agents and Agent Analytics are only available on Enterprise plans and are not isolated to one workspace. They are configured at the organization level in Profound and the same information is accessible across client and pitch workspaces.

## Track workspace usage and allocations

Agencies can track their Profound resource consumption across workspaces from the Agency Mode Dashboard and the Usage dashboard.

For agencies on the Agency Growth plan, billing and usage is consolidated under a single account. Enterprise agency partners of Profound handle their billing through their dedicated Profound account team.

## Manage and increase workspace volume

After an agency purchases a package of Pitch Workspaces from Profound, agency users who are “Admin” roles in their Profound organization can create new workspaces directly within the Profound agency environment.

If you run low on Pitch workspace credits, you can purchase additional workspace packages from Profound.

### Enterprise

Agencies that work with a Profound account team can request additional workspace volume directly through their Profound partner.

### Self-Serve

Agencies on a Growth plan can contact Profound support to request an increase in workspace volume.

## Agency Mode access

Agency Mode is available to agencies on supported plans. Contact your Profound account team to confirm eligibility or enable Agency Mode.

Roles are assigned at the agency level. Admins can create new workspaces, extend workspaces that are about to expire, and invite new agency users from their organization. They can also change roles and permissions for existing users.


# Manage Your Agency's Brands
Source: https://docs.tryprofound.com/agencies/manage-brands

Create, configure, and manage pitch and client workspaces in Agency Mode

Manage your [Agency Mode](/agencies/agency-mode-overview) pitch and client workspaces from the Agency Mode dashboard. From this dashboard, you can create, configure, and manage separate brand workspaces.

> [!NOTE]
>
> Agency Mode is for marketing agencies and consultancies monitoring multiple brands and clients. This is distinct from Profound's [Agent Analytics](/agent-analytics/overview), which measures how answer engines assess your web presence.

## Access the Agency Mode Dashboard

To access the Agency Mode dashboard, click the brand asset dropdown in the upper-left corner of the Profound platform. Select “My Agency” to open the Agency Mode dashboard.

![Brand asset dropdown in the upper-left corner open with the My Agency option highlighted](https://mintcdn.com/profound-37face47/J_F2xMhMTt8_rix9/images/agencies/manage-brands-1.png?fit=max\&auto=format\&n=J_F2xMhMTt8_rix9\&q=85\&s=f9b2567dd0b8d39b82e6e4332de4de4f)

## Create and Manage a Pitch

Every prospective client or business development opportunity for an agency should have its own unique [pitch workspace.](/agencies/agency-mode-overview#pitch-workspaces)You can create and manage your pitch workspaces from the pitch workspace table in the Agency Mode dashboard.

### Create a New Pitch

Click “Start a pitch” on the right side of the pitch workspace table to open the pitch creation menu.

![Client Management dashboard with the Start a pitch button highlighted in the Pitch Workspaces section](https://mintcdn.com/profound-37face47/J_F2xMhMTt8_rix9/images/agencies/manage-brands-2.png?fit=max\&auto=format\&n=J_F2xMhMTt8_rix9\&q=85\&s=0b7f48df86b4e477f46cd4eca4534967)

You can choose to generate workspace prompts using AI or upload them via CSV.

Provide the following information, then click “Create Pitch Workspace”:

* **Brand Name and Website** - The client’s brand name and primary domain URL.
* **Industry** -  The prospective client’s unbranded industry that they operate in.
* **Regions** -  The countries the prospective client serves and the language spoken.
  * This creates prompts in the natural language of the country being tracked.
* **Topics** - Key themes that you want to explore with your prospective client.
* **Competitors -** The brand names and main website URLs of the prospective client’s key industry competitors.
* **Contextual Information -** Additional details used to generate client prompts.
  * This can be content such as key audiences that you want targeted in the prompt generation, or journeys to reflect. Examples include:
    * “Prompts should be from the perspective of retail shoppers, and should include a mix of research/awareness, consideration and purchase related prompts across the full funnel.”
    * “Please include a mix of prompts with some from the perspective of a patient and some from the perspective of a physician, and ensure they focus more on upper- and mid-funnel queries these audiences would ask about the diabetes industry.”

**• Prompt Creation Method -** Choose between uploading your own CSV or by using Profound’s prompt generation.

> [!NOTE]
>
> You can edit auto-generated prompts in the [Prompt Designer](/answer-engine-insights/prompt-designer) once the client workspace is set-up.

### Manage Pitch Workspace Prompts

To access prompts for a pitch workspace, click the brand name in the pitch workspace table on your Agency landing page. This opens the Prompt Designer for that specific pitch.

You can also access the [Prompt Designer](/answer-engine-insights/prompt-designer) from the “Prompts” tab in Answer Engine Insights while inside the brand workspace.

### Promote a Pitch Workspace into a Client Workspace

You can expand pitch workspace capabilities by promoting it into a client workspace. Profound recommends promoting any pitch workspace when a prospective customer converts into a paid customer.

> [!NOTE]
>
> Reach out to your Profound account management team or support for help designing the client workspace configuration for your client’s ongoing business needs.

Click “Promote” on the right side of the corresponding row in the workspace table to promote a pitch workspace to a client workspace.

![Client Management dashboard with the Promote button highlighted on a pitch workspace row](https://mintcdn.com/profound-37face47/J_F2xMhMTt8_rix9/images/agencies/manage-brands-3.png?fit=max\&auto=format\&n=J_F2xMhMTt8_rix9\&q=85\&s=90766fbd245ddefd7d9ca712ea485296)

### Extend or End a Pitch

Click “Extend” on the right side of the corresponding row in the workspace table to extend a pitch workspace. You can extend a pitch for 30 days or for a pre-set time period for pitch workspaces.

![Client Management dashboard with the Extend button highlighted on a pitch workspace row](https://mintcdn.com/profound-37face47/J_F2xMhMTt8_rix9/images/agencies/manage-brands-4.png?fit=max\&auto=format\&n=J_F2xMhMTt8_rix9\&q=85\&s=5cd2b6623eec80a63f13706f8c5dd8fd)

Extending a pitch prevents it from expiring and moving into an inactive state, and uses another one of your available credits for pitch workspaces. You [can purchase more workspaces](/agencies/agency-mode-overview#manage-and-increase-workspace-volume)if you do not have any more available.

## Create and Manage a Client Workspace

Every client should have its own [client workspace](/agencies/agency-mode-overview#active-client-workspaces), and some clients may require multiple workspaces for different lines of business, product categories, or countries. You can create and manage a client workspace from the client workspace table in the Agency Mode dashboard, or partner with Profound to create the client workspace on your behalf with your inputs.

### Create a New Client Workspace

Click “Add a new client” on the right side of the pitch workspace table to open the client workspace creation menu.

![Client Management dashboard with the Add a new client button highlighted in the Client Workspaces section](https://mintcdn.com/profound-37face47/J_F2xMhMTt8_rix9/images/agencies/manage-brands-5.png?fit=max\&auto=format\&n=J_F2xMhMTt8_rix9\&q=85\&s=13ac341d20f4ff2832848cfdfcca53a0)

You can determine if you want to generate workspace prompts using AI or alternatively via CSV upload.

Input the following parameters and click “Create Client Workspace” to create the workspace:

* **Brand Name and Website** - The client’s brand name and primary domain URL.
* **Industry** -  The prospective client’s unbranded industry that they operate in.
* **Regions** -  The countries the prospective client serves and the language spoken.
  * This creates prompts in the natural language of the country being tracked.
* **Topics** - Key themes that you want to explore with your prospective client.
* **Competitors -** The brand names and main website URLs of the prospective client’s key industry competitors.
* **Contextual Information -** Additional details used to generate client prompts.
  * This can be content such as key audiences that you want targeted in the prompt generation, or journeys to reflect. Examples include:
    * “Prompts should be from the perspective of retail shoppers, and should include a mix of research/awareness, consideration and purchase related prompts across the full funnel.”
    * “Please include a mix of prompts with some from the perspective of a patient and some from the perspective of a physician, and ensure they focus more on upper- and mid-funnel queries these audiences would ask about the diabetes industry.”
* **Prompt Creation Method -** Choose between uploading your own CSV or by using Profound’s prompt generation.

### Manage Client Workspace Prompts

As with any Profound workspace, you can manage prompts for a client workspace using the [Prompt Designer](/answer-engine-insights/prompt-designer).

### Remove a Client Workspace

If a client ends its longer term engagement with your agency and you need to repurpose the client workspace for a different client, you can reach out to Profound. Profound will consult on how to leverage this client workspace allocation for a different ongoing business need if needed.

## Assess Workspace Usage

The Agency Mode dashboard shows how many pitch and client workspaces you are using relative to your allotment or package of each workspace type purchased from Profound.

Each row also displays prompt usage for the workspace beneath the brand name, as well as the number of days remaining before a pitch workspace becomes inactive.\
You can increase your workspace allotment by [purchasing more workspaces](/agencies/agency-mode-overview#manage-and-increase-workspace-volume).


# Add conditionals to Workflows
Source: https://docs.tryprofound.com/agents/how-tos/add-conditionals-to-workflows

Direct a Workflow down different paths based on variable values

Conditionals are decision points within your Workflow that direct the Workflow down different paths based on variable values.

Use a Conditional if a Workflow must operate differently depending on an input or a value generated in a previous step. For example, route workflows to trigger different prompts based on specific user input.

## Add a Conditional

Click the **+** icon in the Workflow Builder and select **Conditional**. Alternatively, drag **Conditional** from the node selection menu into the workflow.

![Workflow Builder node picker with the Conditional option highlighted under Logic](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-conditionals-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=02d6096eeed2d778b4ff356400c7842c)

Conditionals require you to define a condition using a comparison expression (for example: equals, greater than, less than, or contains). If the condition evaluates as true, the Workflow follows the **If true** branch. If not, it follows the **If false** branch.

![Workflow Builder with a Conditional node showing If true and If false branches and the condition settings panel](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-conditionals-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=b2f8d6fd68068dc707dfecb656d16d94)

Each branch can contain its own sequence of steps.

## Configure conditions

When configuring a Conditional, select:

* The variable to evaluate
* The evaluation operator (equals, greater than, less than, and so on)
* The value to compare it against

Select **+ Add Condition** to add multiple conditions to the same branch. Multiple conditions can be combined using **AND** (all conditions must be true) or **OR** (at least one condition must be true). When multiple conditions are configured, all must be satisfied for the Workflow to follow the **If True** branch.

![Conditional node settings with two number conditions joined by AND and the Add condition button highlighted](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-conditionals-3.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=6c7e2d049f05dbec70914ec87ec6eb3f)

If the conditions are not satisfied, the Workflow automatically follows the **If false** branch.

## Conditional branch requirements

There are several requirements to ensure a Workflow executes as expected:

* The variable used in a condition must exist and must be created in a previous step
* Both branches must ultimately lead to an **End** step to complete the Workflow run
* The evaluation operator must align with the input variable type:
  * Numerical operators only work with numeric values. For example, **greater than** will not work on a string such as "Best AEO Tools"
  * String operators only work on text strings. For example, **contains** will not work on a numeric value of 126


# Add iterations to Workflows
Source: https://docs.tryprofound.com/agents/how-tos/add-iterations-to-workflows

Loop a Workflow through a list and run the same steps for each item

## Overview

The Iteration node loops through a list within a Workflow, executing the same sequence of steps for each item automatically.

By default, a Workflow processes a single input and produces a single output. The Iteration node extends this by applying the same process across an entire list, taking multiple inputs and producing multiple outputs within a single Workflow run.

Use an Iteration node when a Workflow must perform a repeated action across multiple values, such as running an LLM prompt for each item in a list of keywords or URLs. Examples include:

* Analyzing a list of URLs to identify common themes in their content and structure, then optimizing an owned piece of content accordingly.
* Collecting prompt responses across multiple answer engines, regions, or topics and conducting bulk analysis on brand and product mentions to surface trends.
* Iterating over a list of Reddit threads to generate a clean report on top narratives and recurring themes.

## Add an Iteration node

Select the **+** icon in the Workflow Builder and select **Iteration**. Alternatively, drag **Iteration** from the node selection menu into the workflow.

![iterations-1.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-iterations-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=e27ce5af180c5721a5fe148b19f6e95e)

Iterations require a list of items as the input. The node loops through each item sequentially, passing the current item as a variable to the steps inside the loop.

## Configure an Iteration node

Configuring an Iteration node requires two things: passing in a list, and building the steps that execute on each iteration. The node contains its own self-contained editor, modeled after the broader Workflow Builder.

![Iterations\_3.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-iterations-2.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=98f7f07319d8529b582fcaa9fe40842d)

### Iterate Over a List Input

An Iteration node requires a list as your Workflow input. To configure a list input, select the input node, add a new input, and choose **List** from the dropdown.

![Iterations\_2.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-iterations-3.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=a08f95d8f6866ef8cad9bcb46ac429f0)

### Configure your Iteration node

Select the node to open the node settings menu, then select the list input as the value to iterate on.

The output of an Iteration node defaults to the outputs of the last node within it.

### Add your Iteration steps

Add nodes to the Iteration node the same way you add nodes to the Workflow Builder. Select the **+** icon in the Iteration editor and select a node, or drag a node in from the node selection menu.

> [!TIP]
>
> Each step repeats for every item in the list. Place any node that should run only once outside the Iteration editor.

### Structure your output

Add output fields to map your final results when using the Prompt LLM node. In your LLM prompt, specify what information should populate each field, for example, "*Return each URL in the* `urls` *output field.*" Then add those same fields as structured output fields in the node settings.

![Screen Shot 2026-04-26 at 8.24.04 PM.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/how-tos/add-iterations-4.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=e7053239f50b1ea112f7912b2ee43b6b)

## Iteration node requirements

There are several requirements to ensure a Workflow executes as expected:

* The list variable used as input must exist and must be created as a Workflow input using the list type.
* The current item variable is only accessible within the steps inside the loop and cannot be referenced in steps outside the Iteration node.
* Steps inside the loop execute once per item in the list.
* The Iteration node must connect to an End step or additional steps to complete the Workflow run.


# Migrate a Workflow to an Agent
Source: https://docs.tryprofound.com/agents/how-tos/migrate-workflow-to-agent

Turn an existing Workflow into an Agent with AI Marketer (Aim)

Migration turns an existing node-based Workflow into an Agent. AI Marketer (Aim) reads the nodes in your Workflow and writes an Agent with equivalent instructions, tools, and outputs. Your Workflow isn't changed, so you can compare the two and switch over when you're ready.

## Why migrate?

Migrate your Workflows to Agents if you want:

* **Easier updates:** describe the change in the Agent's chat and Aim makes it.
* **Triggers beyond a schedule:** web monitors, Profound data changes, and app events such as Slack messages. Learn more in [Triggers in Agents](/agents/triggers-in-agents).
* **Flexibility with unexpected inputs and fewer broken runs:** the Agent reasons through the run and works around common issues.

Migration is optional. Agents don't affect your existing Workflows; they keep running as they are.

## Before you migrate

* **Check the tools.** Some integrations available to Workflows aren't available to Agents as tools yet. Note the tools your Workflow uses and compare them against the list in [Tools in Agents](/agents/tools-in-agents). If a tool is missing, consider running the Workflow from an Agent instead.
* **Check the credit cost.** Once credit billing for Agents starts, an Agent run uses more credits than the same Workflow run, because Aim reasons through each step. Keep high-volume, fixed tasks as Workflows.
* **Save an example of good output.** Use it to judge the migrated Agent's first run.

## Migration steps

**1. Start the migration**

1. In Profound, go to **Agents** in the left sidebar and filter by **Type** to find the Workflow you want to migrate.
2. Select the Workflow to open it.
3. Select **Migrate** at the top right of the screen. In the **Migrate to the new Agents experience?** dialog, select **Migrate**.

Profound creates a new Agent and opens its page, where Aim starts the migration in the chat.

**2. Review the Agent**

After Aim drafts the Agent, review what it did in the settings panel and in the chat:

* **Reasoning:** the Agent is Aim's interpretation of the Workflow, so the instructions and steps won't map one-to-one. Read Aim's reasoning and decisions in the chat.
* **Instructions:** read the new Agent instructions end to end. Check that filters, date ranges, and prompt wording from the Workflow stayed the same. Fix anything that doesn't sound right in the instructions editor or by asking Aim.
* **Tools:** Aim adds a tool for each integration it can match. If it can't find a match, it offers workaround options.
* **Inputs and outputs:** confirm the inputs are unchanged and the output type (Document, Sheet, HTML, CSV, or PDF) matches what you expect.

**3. Run the Agent**

Select **New run** and give the Agent the same inputs you'd give the Workflow. Compare the result with your saved output example. Expect differences in wording and order, because an Agent reasons through the steps instead of following nodes. Ask Aim in the same chat for changes, and run the Agent again until the output is right. Changes apply right away.

**4. Add triggers**

1. In the settings panel, select **Add** next to **Run triggers** and add a Schedule trigger that matches your Workflow's schedule, or try another trigger if it suits the task better.
2. Make sure the switch next to **Run triggers** is on.

The Agent now runs on its own. When you're ready, remove the schedule from the Workflow so the task doesn't run twice.

## FAQ

> **Can I migrate several Workflows at once?**
>
> No. Migrate one at a time so you can check each result.

> **The migrated Agent gives different numbers from the Workflow**
>
> Check that the instructions state the same category, asset, date range, and filters as the original nodes. If your Workflow uses fixed calculations or tools that aren't available to Agents yet, keep it as a Workflow, and add it to an Agent's instructions if you want the Agent to run it as a step.

> **Can I migrate back?**
>
> No. Migration is one-way, but it keeps your original Workflow. If the migrated Agent doesn't work for you, keep using the Workflow.


# Prompt LLM Workflow node best practices
Source: https://docs.tryprofound.com/agents/how-tos/prompt-llm-best-practices

How to write prompts that get the most out of the Prompt LLM node in Workflows

The [**Prompt LLM node**](/agents/workflows/nodes/prompt-llm) is one of the most powerful tools in the Workflow Builder. It lets you turn raw inputs (research, scraped content, user data, or Profound insights) into clear, useful outputs such as summaries, outlines, rewrites, analyses, and structured responses.

This guide teaches you how to write highly effective prompts *without any background in prompt engineering*.

### Why good prompts matter

Large Language Models (LLMs) are excellent pattern recognizers, but they need direction. A well-written prompt helps the model understand:

* **What task** to perform
* **How** to perform it
* **What style or format** to use
* **What constraints** to follow

The clearer your instructions, the more accurate and repeatable your output will be. Think of prompts as creative briefs for the AI: the more context you give, the better the work.

### The core structure of an effective prompt

Most great prompts follow this simple structure (but don't have to):

1. **Role**: who or what should the model act as?
2. **Task**: what should it produce?
3. **Context**: what information should it use?
4. **Requirements**: rules, guidelines, constraints
5. **Output format**: how the result should be structured

Below is an example prompt written with this structure:

```text expandable wrap theme={null}
You are an expert content strategist. Summarize the following research into 5 concise insights for a B2B marketing audience.

Context:
{{Web Page Scrape Output}}

Requirements:
- Focus only on key trends
- Avoid duplicate points
- Keep each insight under 20 words

Output Format:
- Bullet points only
```

This structure works across nearly every type of prompt you write in Profound.

### The 5 elements of strong prompts

#### 1. Give the model a role (optional but helpful)

LLMs perform better when given a defined perspective to write from.

Examples:

* “You are a senior SEO strategist who writes in a professional, concise tone.”
* “Act as a technical explainer for a non-technical audience.”

Assigning a role shapes the voice and decision-making of the model.

#### 2. Be explicit about the task

Ambiguous instructions lead to unpredictable results.

Weak:

> “Write something about this.”

Strong:

> “Rewrite the following paragraph to be clearer and more concise, while keeping the core meaning.”

Even stronger:

> “Rewrite the following text for a C-suite B2B audience. Reduce fluff. Improve clarity. Keep all data accurate.”

#### 3. Provide context

The Prompt LLM node can reference any previous step with `/` variables.

Good context sources include:

* Web Page Scrape
* YouTube Transcript
* Research Snippets
* Perplexity Search results

Include only relevant context, too much noise reduces quality.

Example:

```text wrap theme={null}
Based only on this content:
{{YouTube Video Transcript Output}}
```

### 4. Add clear requirements and constraints

This is where you shape the output to your needs.

Common constraints marketers use:

* **Tone:** professional, friendly, authoritative, neutral
* **Length:** 3 bullets, 100 words, 1 paragraph
* **Audience:** B2B marketers, enterprise executives, small business owners
* **Content rules:** no jargon, avoid repetition, do not invent facts
* **Brand style:** simple, direct, value-driven

Example:

```text wrap theme={null}
Requirements:
- Keep introduction under 50 words
- Do not invent data or statistics
- Avoid phrases like “in today’s world”
```

### 5. Specify the format of the output

LLMs are much more accurate when you tell them exactly what shape the output should take.

Examples:

* Bullet list
* JSON
* Table
* One-paragraph summary
* Step-by-step instructions
* Title + 3 subtitles + CTA

Example prompt:

```text wrap theme={null}
Return your answer in JSON with fields:
- insight
- supporting_quote
- source_url
```

## Prompt patterns you can use right away

Below are reusable templates tailored for common marketing Workflows.

### 1. Summarize content

```text expandable wrap theme={null}
Act as a senior content strategist. Summarize the key points from the text below for a B2B marketing audience.

Context:
{{Perplexity Search Output}}

Requirements:
- Provide 5 bullet points
- No more than 20 words per point
- Avoid repetition
- No fluff

Output format: Bullet list only.
```

### 2. Rewrite for clarity or tone

```text expandable wrap theme={null}
Rewrite the following text to be clearer, more direct, and suitable for an enterprise executive audience.

Context:
{{Web Page Scrape Output}}

Requirements:
- Remove jargon
- Keep factual accuracy
- Shorten sentences where possible

Output format: One clean rewritten paragraph.
```

### 3. Generate titles or headlines

```text expandable wrap theme={null}
Generate 10 high-performing article titles based on the topic below.

Topic: {{topic}}

Guidelines:
- Clear, crisp, compelling
- Avoid clickbait
- Include a strong, specific angle
- Keep under 60 characters when possible

Output format: Numbered list.
```

### 4. Extract structured data

```text expandable wrap theme={null}
Extract key insights from the content below.

Context:
{{YouTube Video Transcript Output}}

Requirements:
- Identify 3 themes
- For each theme, provide a 1-sentence explanation and a supporting quote (if available)

Output format (JSON):
[
  { "theme": "", "summary": "", "quote": "" }
]
```

### 5. Compare or evaluate content

```text expandable wrap theme={null}
Act as a content analyst. Compare the strengths and weaknesses of two competitor pages.

Context A:
{{Web Page Scrape Output A}}

Context B:
{{Web Page Scrape Output B}}

Requirements:
- Identify 3 strengths and 3 weaknesses for each page
- Keep explanations under 15 words

Output format: A table with columns: Page, Strength, Weakness.
```

### Tips for getting consistently better outputs

#### ✔ Tell the AI what to avoid

Hallucinations decrease significantly when you include rules like:

* “Do not add facts that are not in the context.”
* “If information is missing, say ‘Not enough information’.”

#### ✔ Use examples

“Show, not tell” is extremely effective.

```text wrap theme={null}
For example, a good headline looks like:
- “How AI Improves Inventory Forecasting”
- “The Future of Predictive Supply Chains”
```

#### ✔ Keep prompts short, but not vague

Overly long prompts can confuse the model. Overly short prompts give it no direction.

Aim for **3–8 sentences** + **one format block**.

#### ✔ Iterate and refine

Try one prompt, review the output, and refine it. Prompt writing is like briefing a writer.

#### ✔ Use variables to make prompts dynamic

This is especially helpful when building scalable Workflows.

Example variable-driven prompt:

```text wrap theme={null}
Summarize the following content for an audience of {{audience_segment}}.

Context:
{{Web Page Scrape Output}}
```

### Common mistakes and how to avoid them

#### ✖ Asking the AI to “write something good”

→ Too vague. Always specify *what* makes it good.

#### ✖ Forgetting to restrict the source

→ AI will make things up unless you say “Use only the context provided.”

#### ✖ No format specified

→ Results become inconsistent and hard to work with in Workflows.

#### ✖ Too many goals in one prompt

→ Split complex tasks into sequential Prompt LLM nodes for cleaner outputs.

### Advanced prompt writing (optional)

For users who want more precision:

* **Chain of Thought:**\
  “Explain your reasoning step-by-step, but only return the final answer.”
* **Transform → Evaluate → Improve loops:**\
  Create a second Prompt LLM node to critique and refine the first output.
* **Zero-shot vs. few-shot prompting:**\
  Providing “example outputs” dramatically improves accuracy.

### Example: full high-quality prompt template

```text expandable wrap theme={null}
You are an expert B2B content strategist. Rewrite the article below so it is clearer,
more scannable, and optimized for AI-answer performance.

Context:
{{Generate Article Output}}

Goals:
- Improve clarity and remove fluff
- Strengthen section headers
- Ensure each section answers a real user intent
- Maintain accuracy; do not add unsupported facts

Output Format:
Return:
1. A revised article
2. A bullet list of the key improvements you made

Style:
- Professional, concise, and trustworthy
```

### Final thoughts

Prompt writing is a skill that compounds over time. You don't need to be a "prompt engineer." You simply need to:

* Be clear
* Be specific
* Provide context
* Give constraints
* Define the output format

With these techniques, you'll consistently generate strong, predictable results with the Prompt LLM node, unlocking everything from content summaries to fully automated content production pipelines.


# Call API node
Source: https://docs.tryprofound.com/agents/workflows/nodes/call-api

Connect a Workflow to any external API endpoint

The **Call API** node connects your Workflow to any external API endpoint. Use this step to retrieve data, trigger external services, enrich Workflow inputs, or integrate with your internal systems and tools.

This node supports both **GET** and **POST** requests, optional authentication, and (for POST requests) request bodies in JSON format.

![call api.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/call-api-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=d7ebced3d931337a4f07d806de54557f)

## When to use this node

Use Call API for tasks such as:

* Pulling data from third-party tools, CRMs, CMS systems, or analytics platforms
* Triggering external automations or Workflows
* Fetching structured data for use in downstream LLM steps
* Enriching Workflow execution with custom business logic hosted elsewhere
* Integrating Profound Workflows with internal APIs or microservices

## Node configuration

Selecting the Call API node opens its configuration panel on the right side of the Workflow Builder.

### Method (required)

Choose the HTTP method for the request:

* **GET** – Retrieve data from an API endpoint
* **POST** – Send data to an endpoint; allows including a request body

Choose the method that matches the API you are calling.

### URL (required)

Enter the full API endpoint URL.

You can:

* Paste a static URL
* Dynamically insert variables using `/`, referencing inputs or earlier Workflow steps

Examples:

* `https://api.example.com/users/{{user_id}}`
* `https://internal.mycompany.com/search?q={{topic}}`

## Additional configuration for POST requests

If the method is **POST**, additional fields appear.

### Request Body

Enter the JSON payload to send in the request. This is required for POST requests.

Example payload:

```text expandable wrap theme={null}
{
  "userId": 1234
  "action": "create",
  "payload": {
    "firstName": "Jack",
    "email":"jack@example.com"
  }
}
```

Variables can be inserted using `/`.

### Content Type (required for POST)

Choose the content type for the request body:

* **None**
* **Application/Json**

Most APIs expect `application/json`.

## Output Label

Provide a descriptive label for the API response. This becomes the variable name that downstream nodes reference.

Examples:

* `api_response`
* `user_record`
* `external_data`

## Advanced settings

Select **Advanced settings** to configure authentication for the request.\
Authentication is required for APIs that are not publicly accessible.

### Authentication Type (required)

Choose the authentication mechanism:

* **None** – no authentication
* **Basic** – username and password
* **Bearer** – bearer token
* **API Key** – custom API key header

Each option exposes different required fields.

#### Basic Authentication

Fields:

* Username
* Password

#### Bearer Authentication

Fields:

* Bearer Token

#### API Key Authentication

Fields:

* Header Name
* API Key Value

Example:

* Header Name: `X-API-Key`
* API Key Value: `<your_key_here>`

Use the field names expected by the API you are calling.

## Output

The Call API node returns the full response body from the API as text. This output can be passed into:

* LLM steps for parsing, extraction, or summarization
* Logic and transformation nodes
* Downstream API calls
* Any step requiring structured or semi-structured external data

The output may be JSON, HTML, plain text, or any content type returned by the API.

## Example usage

### 1. Fetch user profile data

1. Add a **Call API** node with:
   * Method: GET
   * URL: `https://api.example.com/users/{{user_id}}`
   * Output Label: `user_profile`
2. Pass \{\{user\_profile}} into an LLM step for analysis or transformation.

### 2. Submit Workflow data to an external service

1. Add a **Call API** node with:
   * Method: POST
   * URL: `https://hooks.example.com/events`
   * Content Type: Application/Json

```text wrap theme={null}
Request Body:{
  "transcript": "{{video_transcript}}", 
  "summary": "{{summary}}"
}
```

* Output Label: `event_response`

### 3. Use authenticated APIs

1. Under **Advanced settings**, choose the appropriate authentication type.
2. Provide credentials (e.g., Bearer Token or API Key).
3. Use GET or POST as needed to retrieve secure data.

## Best practices

* Always verify the API’s expected method, headers, payload structure, and authentication requirements before configuring the node.
* Use variables to dynamically build URLs or request bodies for more flexible Workflows.
* When sending JSON, ensure proper formatting: malformed JSON will cause the request to fail.
* Keep output labels descriptive, especially when chaining multiple API calls.

## Troubleshooting

### The API returns an error

Check:

* URL correctness
* Required query parameters
* Authentication credentials
* JSON body structure (for POST)

### The response is empty or unexpected

APIs may return HTML error pages or wrapper objects. Use a Prompt LLM step to interpret or extract relevant portions.

### Variables aren’t resolving

Insert variables by typing `/` and selecting from the variable picker.


# Get Sitemap node
Source: https://docs.tryprofound.com/agents/workflows/nodes/get-sitemap

Build a complete sitemap of any publicly accessible website

The **Get Sitemap** node builds a comprehensive sitemap for any publicly accessible website. It returns a structured list of all discoverable URLs, giving you a complete view of a site's content footprint.

This is useful for research, content audits, competitive analysis, content planning, and feeding large sets of URLs into downstream Workflow steps.

Behind the scenes, this node uses Firecrawl's sitemap and site-mapping capabilities to crawl a domain and return a clean, deduplicated list of URLs, but the process is fully abstracted for you inside Profound.

![Frame 2147259079 (3).png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/get-sitemap-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=8f28e6d38d90365edfb8bbac595c60d3)

## When to use this node

Use **Get Sitemap** when you want to:

* Audit all URLs on your own site or a competitor’s
* Identify content gaps or opportunities for AEO
* Build Workflows that automatically analyze or summarize entire sites
* Retrieve source URLs for bulk scraping, bulk insights, or large-scale content research
* Feed multiple URLs into LLM steps for comparison, clustering, or extraction
* Conduct technical/SEO reviews or map site hierarchy

This is a foundational node for building AEO-aware research pipelines.

## Node configuration

### Website URL (required)

Enter the base URL of the website you want to map.

Examples:

* `https://example.com`
* `https://www.competitor-site.com`
* `https://docs.example.com`

The node will crawl the domain starting from this URL and attempt to discover all reachable pages.

You can type the URL directly or use `/` to insert variables from earlier Workflow steps.

### Output Label

Provide a name for the node's result, such as:

* `sitemap`
* `site_urls`
* `mapped_pages`

This label will reference the sitemap output in later steps.

## Advanced settings

Expand **Advanced settings** to configure crawl depth and filtering.

### Search Query

Allows you to filter the sitemap results by keyword.

Examples:

* `blog`
* `pricing`
* `case-study`
* `ai`

This is helpful when you only want URLs matching a specific topic or section of the site.

### Maximum Results

Limits the number of URLs returned.\
Default: **100**

Set this higher or lower depending on your use case:

* **100–500** for competitive research
* **500–2,000** for large content audits
* **10–50** for quick sampling or lightweight Workflows

## Output

The node returns a structured list of URLs in a clean, machine-readable format.\
Each entry typically includes:

* The page URL
* Basic metadata discovered during crawling
* Normalized and deduplicated links

This output can be consumed directly by:

* [Web Page Scrape node](/agents/workflows/nodes/web-page-scrape)
* [Perplexity Search node](/agents/workflows/nodes/perplexity-search)
* [Prompt LLM node](/agents/workflows/nodes/prompt-llm)
* [Profound data nodes](/agents/workflows/nodes/data/visibility-score)
* Custom analysis or extraction steps

## Example usage

### 1. Build a full competitor content audit

1. Add **Get Sitemap** with `https://competitor.com`
2. Feed output into [Web Page Scrape](/agents/workflows/nodes/web-page-scrape) using a loop or batched Workflow
3. Analyze patterns using [Prompt LLM](/agents/workflows/nodes/prompt-llm)​ (e.g., topics, structure, gaps)
4. Generate a research report or summary

### 2. Identify content gaps for AEO

1. Use **Get Sitemap** to list all pages on your site
2. Use [Citation Pages](/agents/workflows/nodes/data/citation-pages) to see which pages are cited
3. Use [Prompt LLM](/agents/workflows/nodes/prompt-llm) to identify pages **not** appearing in answer engines
4. Generate recommendations or content improvements

### 3. Large-scale content generation

1. Retrieve a site’s URLs
2. Filter using the Search Query field (e.g., `/blog`)
3. Feed selected URLs into a [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) or [Generate Article](/agents/workflows/nodes/content-creation/generate-article) chain
4. Produce updated or derivative content at scale

## Best practices

* Start with a reasonable limit (e.g., 200 URLs) before scaling up, especially for large sites.
* Use **Search Query** to narrow results when you only need a subset of the content.
* Combine with [Prompt LLM](/agents/workflows/nodes/prompt-llm) to cluster or categorize URLs.
* For large enterprise websites, run multiple Get Sitemap nodes with different starting URLs (e.g., `/blog`, `/products`, `/docs`).
* Use descriptive output labels when chaining multiple sitemap operations.


# Human Review node
Source: https://docs.tryprofound.com/agents/workflows/nodes/human-review

Add a step where a person approves, rejects, or edits a Workflow's output before the run continues

The Human Review node adds a review step inside a Workflow. Use it when a person needs to approve, reject, or edit a Workflow's output before the run continues.

![Screenshot 2026-07-30 at 12.38.43 pm.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/human-review-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=6b72b0b3c9185fa9f0d23342fafcd87d)

## When to use this node

Use the Human Review node for tasks such as:

* Reviewing AI-generated content before publishing
* Sending generated ads, claims, or messaging to a legal or compliance reviewer before they go live
* Approving a content brief, then generating the article from the approved version

## Limitations

* The Human Review node **works only with text or string values** and doesn't support lists, numbers, or other variable types.
* The node doesn't work with the [Iteration node](/agents/how-tos/add-iterations-to-workflows). However, it can run in [Sheets](/sheets/getting-started), and multiple Human Review nodes can run sequentially within the same Workflow.

## Configuration

### Review mode (required)

Choose what the reviewer can do:

* **Review only**: the reviewer approves or rejects the output to continue or cancel the run.
* **Review with edits**: the reviewer can edit the output before approving.

### Instructions for Reviewer (optional)

Add guidance for the person reviewing this step.

### Content to Review (required)

In the dropdown menu, select the variable that holds the content you want reviewed.

### Reviewer (required)

Choose who can review the step:

* **Any member can review**: any workspace member can respond, and the first response completes the review.
* **Current user**: only the person who runs the Workflow can review. To other team members, the Workflow appears in a waiting state.

### Output label (required)

Enter a descriptive label for this step's output, such as `reviewed_output` or `approved_brief`, or keep the default.

## Output

The node returns the reviewed result as text. In **Review only** mode, this is the approved output. In **Review with edits** mode, this is the reviewer's edited version. The run pauses at this step until the reviewer responds, then continues on approval. On rejection, the run stops and appears as failed in **Past runs**.

## Example use case: Approve the article content before publishing it to your Wordpress site

1. Add the node that generates your content, such as [Generate Article](/agents/workflows/nodes/content-creation/generate-article).
2. Add a **Human Review** node. In **Content to review**, select the generated content from step 1. Set Output label to `approved_content`.
3. Add a [WordPress](/integrations/wordpress/overview) - Create Post node that publishes `approved_content` to your WordPress site if the changes are approved.

## Best practices

* **Place this node before publishing**. Add it before any step that publishes or applies changes, so nothing goes live without review.
* **Write clear reviewer instructions**. The reviewer may not know what the Workflow is doing. Clear instructions help them understand what's needed and respond quickly.
* **Chain Human Review nodes.** When you use **Review with edits** to request content edits from a team member, add a second Human Review node set to **Current user** so you can check the changes before handing control back to the Workflow.

## FAQ

### How do my team members know if they need to review something?

For now, the Workflow run page is the only place to check for pending reviews. Notifications for pending reviews are in development. As a workaround, add a Slack, Gmail, or other messaging node right before the Human Review node to tell your team to check the Workflow run.

### Does a Workflow run paused for review have a timeout?

No. The run stays paused until the review is complete.

### How do I stop or cancel a paused Workflow run?

If you created the Workflow that's asking for a review, stop the run by selecting **Stop** on the published Workflow page. If you're a team member who didn't create the Workflow, cancel the run by rejecting the review (**Start Review** > **Reject**) on the published Workflow page.

### Can I request more than one person for the review?

No, each node has only one reviewer. To collect reviews from multiple team members, chain Human Review nodes sequentially.


# Perplexity Search node
Source: https://docs.tryprofound.com/agents/workflows/nodes/perplexity-search

Run a Perplexity-powered web search and return a citation-backed answer

The **Perplexity Search** node runs a Perplexity-powered web search and returns a detailed, citation-backed answer. Use this step when you need synthesized research, verified facts, or aggregated insights from multiple reputable sources.

This node is useful for Workflows that require high-quality search results before passing the information into LLM steps, analyses, or downstream transformations.

![Perplexity Search Node](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/perplexity-search-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=421b2c1c2d20d28864b438bd13e1d8be)

## When to use this node

Use Perplexity Search for tasks such as:

* Gathering up-to-date information about a topic, product, company, or trend
* Performing competitive analysis or market research
* Extracting factual answers supported by citations
* Enriching Workflows that rely on current web data prior to LLM processing

## Node configuration

Selecting the Perplexity Search node opens its configuration panel on the right side of the Workflow Builder.

### Model

Choose the Perplexity model used to process the query. Available options may differ in speed, reasoning depth, and output quality:

* **Sonar**
* **Sonar Pro**
* **Sonar Reasoning**
* **Sonar Reasoning Pro**
* **Sonar Deep Research**

Select a model appropriate for the complexity and depth of the research needed.

### Prompt

Enter the query or instruction you want Perplexity to research.\
You can type text directly or insert variables from earlier Workflow steps by typing /.

Examples:

* “What are the latest pricing updates for HubSpot Enterprise?”
* “Summarize the top three competitors to \{\{brand\_name}} and provide source links.”

###

### Output Label

Assign a descriptive label to this step’s output. The label becomes the variable name available to later nodes.

Examples:

* `perplexity_answer`
* `research_summary`
* `competitive_insights`

## Output

This node returns a single structured text output containing Perplexity’s synthesized answer, including citations to the underlying sources. The output can be passed directly into steps such as:

* Prompt LLM
* Answer comparison or scoring Workflows
* Additional research steps
* API calls or data extraction logic

## Example usage

### 1. Run a research query and summarize with LLM

1. Add a **Perplexity Search** step with a prompt such as:\
   “Explain the latest changes to Google’s AI Overviews and cite sources.”
2. Add a [Prompt LLM node](/agents/workflows/nodes/prompt-llm) step:

```text wrap theme={null}
Summarize the following research into 5 bullet points and highlight any implications for SEO:

{{perplexity_search.output}}
```

### 2. Dynamically research competitor information

1. Pass a brand name into the Workflow as an input.
2. Use Perplexity Search:\
   “Provide a competitor overview for \{\{brand\_name}} with citations.”
3. Use the output to generate analysis or comparisons downstream.

## Best practices

* Provide clear, specific prompts to get more accurate citations and structured outputs.
* Use more advanced models (e.g., **Sonar Reasoning Pro** or **Sonar Deep Research**) for complex or multi-step research queries.
* Combine Perplexity Search with Prompt LLM steps to refine, structure, or format the research for your use case.

## Troubleshooting

### Output lacks depth or detail

Try switching to a higher-tier model such as Sonar Pro or Sonar Deep Research.

### Variables are not recognized in the prompt

Type `/` in the prompt field to insert supported Workflow variables.

### Returned information seems outdated

Reframe the query to explicitly request the most recent updates or year-specific results.


# Prompt LLM node
Source: https://docs.tryprofound.com/agents/workflows/nodes/prompt-llm

Generate text with a large language model from a custom prompt inside a Workflow

The **Prompt LLM** node allows you to generate text using a large language model (LLM). Use this step when you need an AI-generated summary, classification, rewrite, extraction, or any other transformation based on a custom prompt.

You can insert variables from previous steps into the prompt, enabling dynamic responses that adapt to your Workflow's inputs and outputs.

![Frame 2147259079 (1).png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/prompt-llm-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=5295eb815bbff82544e06594d595467c)

## When to use this node

Use the Prompt LLM node for tasks such as:

* Summarizing search results or scraped content
* Drafting responses or structured outputs for further processing
* Extracting entities or insights from text
* Rewriting content for a specific audience or format
* Converting unstructured data into structured fields (e.g., JSON, bullet points)

## Node configuration

Selecting the Prompt LLM node opens its settings panel on the right side of the Workflow Builder. The node includes the following configuration fields.

### AI Model

Choose which model you want to generate the output.

You can switch models at any time while designing the Workflow. The selected model determines output quality, speed, and cost.

### Prompt

Write the instruction you want the model to follow. This is the core input that determines how the LLM will behave.

You can:

* Type plain text instructions
* Reference Workflow inputs
* Insert variables from previous nodes by typing / to open the variable picker

Prompts can be short or detailed. For best results, provide explicit instructions about tone, structure, output format, or constraints.

> [!TIP]
>
> Check out our  [Prompt LLM best practices](/agents/how-tos/prompt-llm-best-practices) guide for tips on writing better prompts.

### Use Skill

Toggle this option to apply an existing product skill to text generation.

### Use Knowledge Base

Toggle this option if you want an LLM to use the information in your Knowledge Base for the task. Once toggled, select the entire Knowledge Base from the dropdown, or select specific folders in the retrieval settings menu.

> [!TIP]
>
> Learn more about Knowledge Bases in [Knowledge Bases and Brand Kits](/context-manager/knowledge-bases-and-brand-kits).

### Structured Output

Define a JSON schema for the LLM response: add fields, define their data types (string, number, boolean, or list), and give them concise descriptions.

This is particularly useful when working with Sheets. Learn more in the [Get started with Sheets](/sheets/getting-started) guide.

### Output Label

Enter a descriptive label for this step’s output. The label you provide becomes the variable name accessible in later Workflow steps.\
Examples:

* `analysis_summary`
* `buyer_intent_score`
* `cleaned_text`

## Output

The Prompt LLM node produces a single text or JSON output containing the LLM’s response. This output can be passed to downstream nodes such as:

* Additional Prompt LLM steps
* API calls
* Profound data processing nodes
* Iteration node

## Example usage

Below are common examples of how the Prompt LLM node may be used in Profound Workflows.

### 1. Summarize a scraped webpage

1. Add a **Web Page Scrape** step.
2. Add a **Prompt LLM** node.
3. In the prompt, reference the scraped page content:

```text wrap theme={null}
Summarize the key themes from the following webpage content in 5 bullet points: 
{{web_page_scrape.content}}
```

4. Set the Output Label to `summary`.

### 2. Extract entities from an answer engine result

1. Use [**Profound data nodes**](/agents/workflows/nodes/data/visibility-score) **​** to gather inputs.
2. Add a **Prompt LLM** step to extract structured information:

```text wrap theme={null}
Extract the brand mentioned in this answer and return JSON:
{"brand": "<value>"}

Content: {{answer.text}}
```

3. Use the resulting JSON output in downstream conditional logic.

### 3. Rewrite content for a specific audience

```text wrap theme={null}
Rewrite the following text for a B2B marketing leader, keeping it concise and professional: 

{{input_text}}
```

## Best practices

* **Be explicit.** LLMs respond more reliably to detailed instructions.
* **Define expected formats.** Use bullet points, JSON, tables, or specific phrasing to control output structure.
* **Reference only what you need.** Avoid passing unnecessarily large content blocks unless required for the task.
* **Name outputs clearly.** Good labels make Workflows easier to maintain and reuse.

## Troubleshooting

### The output is not in the format I expected

Add explicit instructions such as:

* “Reply only in JSON.”
* “Provide exactly three bullet points.”
* “Do not include explanations.”

### Variables do not appear in the prompt

Type `/` while your cursor is in the Prompt field to open the variable picker.

### Output seems incomplete or cut off

Try switching to a larger model, or revise the prompt to request shorter, more structured outputs.


# Article Research Report node
Source: https://docs.tryprofound.com/agents/workflows/nodes/research/article-research-report

Produce a structured, multi-source research report with citations for a topic

The **Article Research Report** node performs deep, multi-source research and produces a structured research report with citations, insights, and synthesized key findings for your chosen topic. This report is designed to serve as **reference material**—not an article itself—supporting strategists, writers, and downstream content-generation Workflow nodes with authoritative, verifiable information.

Unlike the [Research Snippets Generator node](/agents/workflows/nodes/research/research-snippets-generator)(which extracts factual snippets), the Article Research Report node creates a **full narrative research summary**, typically 800–1300 words, complete with inline citations and a reference list.

## When to use this node

Use the Article Research Report when you need:

* A comprehensive research foundation before writing a content brief or article
* A narrative-format summary of key findings across many sources
* A detailed explanation of trends, themes, frameworks, or definitions
* A fact-rich report used to educate subject-matter experts or content teams
* AEO-aligned supporting material to ground downstream content generation

This node is ideal for workflows where accuracy, context, and depth matter.

## Node configuration

Selecting the Article Research Report node opens its configuration panel on the right side of the Workflow Builder.

### Article Title (required)

The article title or topic the research should focus on, for example:

* “Emerging trends in healthcare technology”

* “How AI is transforming supply chain management”

* “Cybersecurity risks in financial services”

This title guides the search queries and synthesis.

### Audience Segment

The intended audience for the research output. This influences tone and depth:

* Executive summaries
* Technical explanations
* Practitioner-level insights

### Company Name

Used to ensure report framing aligns loosely with the brand context or vertical.

### Content Type

Defaults to **General**, but can influence which explanations or structures are emphasized (e.g., guides vs. comparisons vs. thought leadership).

### Target Prompt

If your article must support a specific user query used in AI systems, enter that here. For example: “What is predictive maintenance and how does it work?”

This helps your Workflow prioritize research aligned to key user intents.

### Output Label

Assign a descriptive variable name for downstream referencing.

Examples:

* `research_report`
* `background_research`
* `topic_report`

## How the node works behind the scenes

The Article Research Report node executes a multi-stage research and synthesis pipeline. The internal workflow includes:

### 1. Input evaluation and research setup

The workflow prepares your inputs (title, content type, audience, company context) and determines which research patterns to use.

For example:

* “How-to” topics may require step-by-step explanations
* Comparative topics may need entity lists
* Thought-leadership topics require key themes and debates

### 2. Multi-engine topic research

The node runs several rounds of focused research across tools like:

* Tavily
* Perplexity
* Specialized LLM-based research routines

Multiple queries are generated from your article title to capture:

* Definitions
* Trends
* Statistics
* Case studies
* Expert opinions
* Historical context
* Practical frameworks

This produces a large pool of raw research artifacts.

### 3. Deduplication and relevance filtering

The workflow discards:

* Low-authority sources
* Irrelevant tangents
* Redundant content
* Outdated material

It retains only high-quality, authoritative findings.

### 4. Structured evidence formatting

Research results are converted into a structured JSON dataset that contains:

* Summaries
* Key facts
* Source URLs
* Extracted insights

This dataset becomes the input to the synthesis model.

### 5. Research synthesis into a narrative report

An LLM configured as a **research synthesizer** composes an 800–1300 word report following strict requirements:

* Clear paragraph writing
* Inline numbered citations: \[1], \[2], \[3], …
* A reference list at the end
* A content-type–specific structure:
  * **How-to** → step-by-step
  * **Comparative** → list of entities to compare
  * **Thought leadership** → themes and debates
  * **Opinion** → insights backed by evidence

The synthesizer is instructed to:

* Be factual
* Avoid hallucinations
* Include only information supported by cited sources
* Be neutral and research-driven
* Produce *only* the research report (no article draft, no commentary)

### 6. Final formatting

The output is standardized into a structured, ready-to-use text block containing:

* Title
* Corresponding narrative content
* Inline citations
* References section

This makes the report suitable for editors, strategists, or downstream LLM article generation.

## Output

The final output is an **800–1300 word research report** containing:

* A narrative overview of the topic
* Key findings from multiple sources
* Definitions, models, or frameworks
* Trends, challenges, and future outlooks
* Inline citations like \[1], \[2], \[3]…
* A reference list at the end

This report is intended to be a foundation for:

* Content briefs
* Articles
* Thought-leadership pieces
* Educational summaries
* Editorial decision-making

## Example usage

### 1. Create a deeply researched content workflow

1. Article Research Report ​
2. [Research Snippets Generator](/agents/workflows/nodes/research/research-snippets-generator)
3. [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief)
4. [Generate Article](/agents/workflows/nodes/content-creation/generate-article)

This ensures every step is backed by trustworthy research.

### 2. Editorial preparation

Use this node to educate internal teams about a topic before writing.

### 3. Automated research pipelines

Trigger recurring research reports on topics like:

* Industry trends
* Product categories
* Seasonal themes
* Competitor verticals

## Best practices

* Include a **Target Prompt** if you want the report aligned to real user queries.
* Provide a clear, specific **Article Title** to guide more accurate research.
* Pair this workflow with the [Research Snippets Generator node](/agents/workflows/nodes/research/research-snippets-generator) for fact extraction.
* Use the output label consistently for downstream LLM steps.
* Treat this report as **research**, not as a final draft article.


# Web Page Scrape node
Source: https://docs.tryprofound.com/agents/workflows/nodes/web-page-scrape

Retrieve the content of a single webpage in a clean, structured format

The **Web Page Scrape** node retrieves content from a single webpage and returns it in a clean, structured format. Use this step when you need to analyze, summarize, or extract information from any publicly accessible URL.

This node supports both raw HTML and cleaned markdown outputs, making it useful for downstream AI processing in Prompt LLM steps or for building structured research Workflows.

![scrape.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/web-page-scrape-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=365c277e2c1a6cb0143dcb76091537ae)

## When to use this node

Use Web Page Scrape for tasks such as:

* Gathering content from articles, landing pages, blog posts, or documentation
* Extracting text for summarization or competitive analysis
* Pulling structured content for use in downstream LLM prompts
* Normalizing webpage content before ingestion into other Workflow steps

## Node configuration

Selecting the Web Page Scrape node opens its configuration panel on the right side of the Workflow Builder. This node includes the following fields.

### URL (required)

Enter the webpage URL you want to scrape.

You can either:

* Paste a static URL, or
* Insert a variable by typing `/` to reference Workflow inputs or outputs from previous steps

### Render Output As

Choose the format for the scraped content. Available options:

* **Markdown** – Returns cleaned, readable text suitable for LLM processing
* **HTML** – Returns the page’s HTML structure

Markdown is recommended for most AI-powered tasks, while HTML is useful when tags, structure, or metadata are important.

### Extract main content

When enabled, Profound attempts to remove navigation elements, ads, footers, boilerplate, and other non-core content. This generally results in a cleaner output focused on the primary article or body text.

Disable this option if you need the full page content without filtering.

### Output Label

Provide a descriptive label for the output of this step. The label becomes the variable name you will reference in later Workflow nodes.

Examples:

* `page_content`
* `scraped_markdown`
* `raw_html`

## Output

This node returns a single text output containing the scraped page content in the selected format. The output can be used in:

* Prompt LLM steps
* Additional research or parsing nodes
* API calls
* Any downstream transformation step

## Example usage

### 1. Scrape and summarize a webpage

1. Add a **Web Page Scrape** step and set the URL to the page you want to analyze.
2. Choose **Markdown** as the output format.
3. Add a [Prompt LLM](/agents/workflows/nodes/prompt-llm) step that references the scraped content:

```text wrap theme={null}
Summarize the key ideas from the following content:

{{web_page_scrape.output}}
```

### 2. Extract structured data from HTML

1. Scrape the page using the **html** format.
2. Pass the HTML into a [Prompt LLM](/agents/workflows/nodes/prompt-llm) step with extraction instructions:

```text wrap theme={null}
Extract all product names and prices from the following HTML. Return JSON.
```

## Best practices

* Use **Markdown output** for the cleanest AI-ready text.
* Enable **Extract main content** when targeting article-style pages.
* If scraping fails, confirm that the URL is publicly accessible and does not require authentication.
* Keep output labels clear and descriptive to simplify downstream references.

## Troubleshooting

### The scraped content is empty or incomplete

This can occur if the page relies heavily on client-side rendering or uses anti-scraping protection. Try switching to **html** output for more raw content.

### Boilerplate text appears in the output

Ensure **Extract main content** is enabled to reduce clutter.

### The URL contains variables that are not resolving

Type `/` to insert supported variables and avoid mismatched parameter names.


# YouTube Title & Description node
Source: https://docs.tryprofound.com/agents/workflows/nodes/youtube-title-description

Extract the title and description from a public YouTube video

The **YouTube Title & Description** node extracts the title and description from any publicly accessible YouTube video. Use this step when you need structured metadata for analysis, summarization, competitive research, or to feed into downstream LLM nodes.

This node provides a clean, text-based output containing both fields, making it easy to integrate YouTube content into broader research or automation pipelines.

![title and des.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/youtube-title-description-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=3e6fe390402d49fccd91f6627c31ea51)

## When to use this node

Use YouTube Title & Description for tasks such as:

* Analyzing how competitors position and describe video content
* Generating summaries, rewrites, or metadata optimizations using LLMs
* Pulling YouTube context into AEO, SEO, or content strategy Workflows
* Extracting structured information before running additional transformations

## Node configuration

Selecting the YouTube Title & Description node opens its configuration panel on the right side of the Workflow Builder.

### YouTube video URL (required)

Enter the URL of the YouTube video you want to extract data from.

You can:

* Paste a direct YouTube watch URL
* Insert a variable from earlier Workflow steps by typing `/`

Supported formats include typical watch URLs such as:\
`https://www.youtube.com/watch?v=VIDEO_ID`

### Output Label

Provide a descriptive label for the output generated by this step. The label becomes the variable name that downstream steps reference.

Examples:

* `youtube_metadata`
* `video_info`
* `yt_title_description`

## Output

This node returns a text output containing:

* **Video title**
* **Video description**

The output can be passed into:

* Prompt LLM steps (for summarization, rewriting, tone analysis, etc.)
* Content comparison Workflows
* Metadata optimization pipelines
* Any downstream step requiring structured text

## Example usage

### 1. Analyze a competitor video description

1. Add a **YouTube Title & Description** step with a competitor video link.
2. Add a [Prompt LLM](/agents/workflows/nodes/prompt-llm) step to interpret the content:

```text wrap theme={null}
Analyze the tone, target audience, and messaging strategy in the following video title and description:

{{youtube_metadata_output}}
```

### 2. Rewrite metadata for optimization

1. Add a Workflow input called `video_url`.
2. Use YouTube Title & Description with `video_url`.
3. Pass the output into a [Prompt LLM](/agents/workflows/nodes/prompt-llm) node:

```text wrap theme={null}
Rewrite the following YouTube title and description to improve clarity and increase click-through rate: {{video_info.output}}
```

## Best practices

* Ensure the video URL is publicly available; private or restricted videos cannot be accessed.
* Use descriptive output labels, especially when working with multiple videos.
* Combine with the **YouTube Video Transcript** node for deeper semantic analysis of video content.

## Troubleshooting

### The node returns no data

Verify that the YouTube URL is valid and not restricted or private.

### The output includes unexpected formatting

Use a Prompt LLM step to clean or reformat the extracted text as needed.

### Variables do not resolve when used as the URL

Insert variables by typing / and selecting from the available options in the picker.


# YouTube Video Transcript node
Source: https://docs.tryprofound.com/agents/workflows/nodes/youtube-video-transcript

Extract the full transcript from a public YouTube video

The **YouTube Video Transcript** node extracts the full transcript from any publicly accessible YouTube video. This includes auto-generated or manually uploaded transcripts. The node combines all transcript segments into a clean, readable text output that can be used for analysis, summarization, content extraction, or downstream LLM processing in your Workflows.

![transcript.png](https://mintcdn.com/profound-37face47/LU_swtbqvG5kIsMS/images/agents/nodes/youtube-video-transcript-1.png?fit=max\&auto=format\&n=LU_swtbqvG5kIsMS\&q=85\&s=0cc32f180034e9464bd22006a9f24a3c)

## When to use this node

Use YouTube Video Transcript for tasks such as:

* Analyzing spoken content from competitor or industry videos
* Creating summaries, highlights, or structured notes using an LLM
* Extracting keywords, entities, or insights from long-form video content
* Feeding transcript text into research, comparison, or content generation Workflows

## Node configuration

Selecting the YouTube Video Transcript node opens its configuration panel on the right side of the Workflow Builder.

### YouTube Video Link (required)

Enter the URL of the YouTube video whose transcript you want to retrieve.

You can:

* Paste a direct YouTube watch URL
* Insert a dynamic variable from a previous Workflow step by typing `/`

Supported formats include typical YouTube links:\
`https://www.youtube.com/watch?v=VIDEO_ID`

### Output Label

Assign a descriptive label to the transcript output. This becomes the variable name used in downstream nodes.

Examples:

* `video_transcript`
* `yt_transcript`
* `transcript_text`

## Output

This node returns a single text output containing the full transcript of the video. Profound automatically:

* Retrieves all transcript segments
* Orders them correctly
* Merges them into a continuous, readable text block

The result is ideal for use in:

* Prompt LLM analysis
* Summaries and content transformation
* Topic extraction or semantic clustering
* QA extraction or answer engine testing Workflows

## Example usage

### 1. Summarize a long YouTube video

1. Add a **YouTube Video Transcript** step with the link to the video.
2. Pass the transcript to a [Prompt LLM](/agents/workflows/nodes/prompt-llm) step:

```text wrap theme={null}
Summarize the following transcript into concise bullet points:

{{video_transcript.output}}
```

### 2. Extract insights for competitive analysis

1. Use YouTube Video Transcript to capture the full content of a competitor’s product demo.
2. Pass the output into a [Prompt LLM](/agents/workflows/nodes/prompt-llm) step to identify messaging themes, product claims, or feature positioning.

### 3. Generate social content from video text

1. Retrieve the transcript.
2. Use a [Prompt LLM](/agents/workflows/nodes/prompt-llm) step to generate snippets or social posts:

```text wrap theme={null}
Create three LinkedIn post variations using insights from this transcript:

{{yt_transcript.output}}
```

## Best practices

* Ensure the video is public or unlisted; transcripts cannot be retrieved from private or restricted videos.
* Auto-generated transcripts may include inaccuracies; consider post-processing with an LLM to improve clarity.
* Use clear output labels, especially when extracting multiple transcripts in a single Workflow.

## Troubleshooting

### No transcript is returned

This may occur if the video has no transcript available or is private. Try another video link or verify settings on the source video.

### Transcript content looks fragmented

Profound merges segments automatically, but heavily timestamped or irregular transcripts may benefit from cleanup via a Prompt LLM step.

### Variable insertion is not working

Use `/` inside the URL field to insert available Workflow variables.


# About Context Manager
Source: https://docs.tryprofound.com/context-manager/about

How Context Manager builds and maintains your brand's source of truth

> [!NOTE]
>
> **Available on:** the Enterprise and Agency plans
>
> **Find it in:** the Context section of the Profound platform

Context Manager gives Profound one place for everything it knows about you and your brand: your positioning, audience, strategy, and the way you like to work.

The context it holds powers the rest of the platform. AI Marketer draws on it to respond like someone on your team, and [Agents](/agents/about) and [Projects](/projects/about) use it to produce work that fits your strategy.

## Why your context matters

Without shared context, every AI conversation starts from zero. You re-explain the audience, the voice, and the product line, and the output still reads like it came from a generic model. The knowledge that would fix this already exists in documents, calls, and old chats, but nothing reuses it.

Context Manager extracts durable information from those sources, stores it as knowledge records, and supplies it to AI Marketer, [Agents](/agents/about), and [Projects](/projects/about) when they generate output. You state a preference once and it carries over to later work, so you are not restating it in every conversation.

## Key features

### Overview of your context

The **Overview** tab summarizes how your brand context and Personal Context work and what influences them. From the tab, ask AI Marketer to fix gaps or add missing context, ask it questions about your context, and browse and search your source files.

### Knowledge records that build themselves

Your brand context lives in knowledge records that Profound generates and maintains for you, covering your company, audience, competitors, offerings, strategy, and execution. They are built from your data and your edits, and they shape how AI Marketer and [Agents](/agents/about) understand your workspace. Find your knowledge records in **Brand Context** **> Generated by Profound**.

### Identified gaps

When Profound is missing a detail about your brand, it surfaces the gap as a card at the top of the **Brand Context** tab, tagged to the knowledge record it belongs to. Reviewing a gap starts a short chat where AI Marketer asks you one question at a time and applies your answers immediately.

### Source references

Every knowledge record shows its references, section by section and claim by claim. Select **References** in your knowledge record to see where a piece of context came from. Each paragraph also ends with citation pills that name the sources behind it. Trace any claim about your brand to the document, call, or conversation behind it.

### Keeping the knowledge current

Correct any section in the knowledge records, and that correction becomes the known answer: your own edits carry the highest authority of any source. When newer information arrives from a highly trusted source, such as a customer call or an approved Slack conversation, Profound favors the most recent information and updates the record automatically.

### Personal memory

**Personal Context** keeps memories about how you work. Tell Profound anything it should remember, and it personalizes your future chats without mixing your preferences into the shared brand context. Search, add, edit, and remove memories at any time.

### Knowledge Bases and Brand Kits

**Knowledge Base** is a repository of brand documents, webpages, and source materials that Profound uses as a source of truth about your brand. **Brand Kits** are brand identity profiles that define tone of voice, writing style, positioning, and other brand attributes for content generation.

Learn more about both features in [Knowledge Bases and Brand Kits](/context-manager/knowledge-bases-and-brand-kits).

> [!TIP]
>
> Brand Kits will become part of the Brand Context in Context Manager. If you are starting out with Profound, we recommend focusing on refining your brand voice in your knowledge records.


# Context Manager FAQ
Source: https://docs.tryprofound.com/context-manager/faq

Frequently asked questions about Context Manager and Knowledge Bases

## Context and knowledge records

> **Do I have to upload all the context manually?**
>
> No. Profound prepopulates knowledge records from the data you've already given it and from background collection about your brand, so files appear even before you upload anything. Adding documents to your Knowledge Base and connecting synced sources deepens the context further.

> **Can we use Context Manager to enforce brand voice and editorial rules?**
>
> Yes. Knowledge records capture voice, approved claims, and do-and-don't rules, and AI Marketer and Agents draw on them when generating content. Correct a section once, and the change applies wherever your context is retrieved.

> **How quickly do updates to context sources show up?**
>
> Knowledge Base uploads populate into knowledge records in about 5 minutes, and context from a Gong call appears about an hour after the call. Edits you approve through AI Marketer propagate almost immediately.

> **What if a knowledge record says something wrong about my business?**
>
> Open the section and fix the claim manually, or tell AI Marketer what to change. The fix applies everywhere that context is used.

> **How does context stay current, and how are gaps filled?**
>
> Context ingestion runs continuously: new Knowledge Base documents and synced source activity keep the files fresh. When something is missing, Profound prompts you to fill the gaps through Identified gap cards in the **Brand Context** tab.

> **How do I remove synced data sources?**
>
> To opt out of ingesting context from a source, select the source in the **Synced Data** table and toggle off the **Context ingestion** setting.
>
> Only organization admins can change these settings. Learn more in [Roles and permissions](/platform-config/people/roles-and-permissions).

> **Where are the Audience Segments?**
>
> Audience Segments is a legacy feature that is now a part of the Brand Context. To define or edit your audience, go to **Context Manager > Brand Context** and refine your **Audience** knowledge record.

## Knowledge Bases

> **What’s the difference between a Knowledge Base and a knowledge record?**
>
> A Knowledge Base holds the source material you give Profound: the documents, webpages, and files themselves. A knowledge record is what Context Manager distills from that material, organized into your company, audience, competitors, offerings, strategy, and execution.

> **What kinds of content should I add to my Knowledge Base?**
>
> Documents that describe how your brand works give the best results: brand guidelines, legal and compliance rules, product information, marketing strategies, content calendars, prior reports, sample outputs, and editorial standards. Anything that answers your priorities, your audience, your voice, and what to always do or avoid.
>
> See [What makes a good Knowledge Base for FactCheck](/answer-engine-insights/how-tos/factcheck-best-results#what-makes-a-good-knowledge-base-for-factcheck) for useful best practices for Knowledge Base content. What works for FactCheck also works well for Context Manager.

> **What types of files can I upload to a Knowledge Base?**
>
> Supported file types are: PDF, text-based formats (Markdown, plain text), CSV, Word documents and Excel spreadsheets. All uploaded files are parsed to raw text before indexing, so non-text-based files may not display optimally. This doesn’t affect search.

> **Does the Knowledge Base update automatically when I add new URLs?**
>
> An auto-sync feature is available for sitemap or URL crawls and external imports (like Notion and Google Drive). It's off (set to **Never sync**) by default. Enable it in the import dialog or in the folder settings on the Knowledge Base page.

> **Is there a limit on how much content I can store in a Knowledge Base?**
>
> Files uploaded to Knowledge Bases can't exceed 100 MB. Larger files need to be split into smaller ones and uploaded separately.
>
> Increasing this limit is on our roadmap. That said, given how Profound processes information in Knowledge Bases (separating data into smaller chunks to enable effective search), it’s always best practice to keep files reasonably sized.


# Get started with Context Manager
Source: https://docs.tryprofound.com/context-manager/getting-started

Add context sources, review knowledge records, and fill identified gaps

## Step 1: Open Context Manager

In Profound, go to **Context Manager** in the **Context** section of the left sidebar. The **Brand Context** tab opens with your generated knowledge records, suggested actions, and any identified gaps.

After you select **Skip** or **Get Started** in the Context Manager introduction, a **Set up your Context Manager** checklist appears in the bottom-right corner. It stays until you complete its steps: **Personalize**, **Brand**, and **Use context in AI Marketer**. Select **Personalize** or **Brand** to go to the page where you add that context. Select **Use context in AI Marketer** to start a chat that uses what you added. The **Brand** step appears only if your role can manage Brand Context.

## Step 2: Add Knowledge Base and Brand Kits

In the **Knowledge Base** and **Brand Kits** tabs, upload the source-of-truth documents for your brand and your brand identity profiles.

Learn more in [Create and manage Knowledge Bases](/context-manager/manage-knowledge-bases) and [Create and manage Brand Kits](/context-manager/manage-brand-kits).

## Step 3: Connect synced data sources

Open the **Synced Data** tab and connect data sources such as Slack, Email, or Gong. Add sources with a **+ Add data** option and speak to your Profound team to set them up.

Select a data source and toggle the **Context ingestion** setting to opt-in or opt-out of using each. You can change your choices anytime.

> [!NOTE]
>
> Changing Context ingestion settings requires the Admin role. Learn more in [Roles and permissions](/platform-config/people/roles-and-permissions).

## Step 4: Review your knowledge records

Scroll down to the **Generated by Profound** section. Open each knowledge record and check it for accuracy. Use the **References** option on any section to see where a claim came from, or use AI Marketer to change anything that reads wrong.

## Step 5: Close your identified gaps

Select **Review** on a gap card in the **Identified gaps** section and answer the questions AI Marketer asks. Each answer flows into the right knowledge record right away.

## Step 6: Add personal memories

On the **Personal Context** tab, add anything Profound should remember about how you work, such as "Keep my briefs under one page" or "I focus on the EMEA market." These memories personalize your future chats.

## Step 7: Connect your workspace tools

Personal Context can also learn from the tools you work in. Go to **Context** > **Personal Context**, select **+ Add context**, then select **Connect to workspaces**. Connect an account and choose what to include:

* **Asana, Jira, and monday.com**: bring in project and task context from the workspaces, sites, and boards you select.
* **Google Drive, SharePoint, Notion, and Granola**: add the files, pages, and meeting notes you choose.

Profound syncs the sources you select daily. With this context, Aim gets a better understanding of your role, priorities, and active work.


# Knowledge Bases and Brand Kits
Source: https://docs.tryprofound.com/context-manager/knowledge-bases-and-brand-kits

What Knowledge Bases and Brand Kits are and how Profound uses them

> [!NOTE]
>
> **Available on:** the Enterprise plan
>
> **Find it in:** the Context Manager in the Context section of Profound

Knowledge Bases and Brand Kits are the sources of knowledge and context Profound has about your brand.

Knowledge Bases are repositories of documents, webpages, and source materials that Profound uses as a source of truth about your brand. Brand Kits are brand identity profiles that define tone of voice, writing style, positioning, and other brand attributes for content generation.

Both features are part of Profound's [Context Manager](/context-manager/about).

## What's the difference?

Both features feed Profound's context about your brand, but they do it in different ways.

Brand Kits define how Profound should communicate on behalf of your brand. Knowledge Bases provide Profound with information it should reference when generating content, performing [FactCheck analysis](/answer-engine-insights/factcheck/about), or any other action that needs to be grounded in facts.

| **Component** | **What it defines** |
| - | - |
| Knowledge Bases | What to say |
| Brand Kits | How to say it |

## How Knowledge Bases work

Knowledge Bases are the foundation the rest of the platform reads from:

* [Context Manager](/context-manager/about) distills them into knowledge records that ground AI Marketer, [Agents](/agents/about), [Projects](/projects/about), and anything else Profound does.
* [FactCheck](/answer-engine-insights/factcheck/about) checks AI answers about your brand against them.
* In [Workflows](/agents/workflows/about-workflows), a Knowledge Base node uses retrieval-augmented generation (RAG) to run a semantic search across your uploaded files and return the most relevant results, so the subsequent nodes' outputs rest on verified brand material rather than the model's assumptions.

### Information management

Knowledge Bases use a folder structure modeled after a standard filesystem. Create folders and subfolders to organize content by product line, team, campaign, or any other grouping. Within each Knowledge Base, view all indexed files, add tags, create new files, delete outdated material, download previously uploaded content, and rename Knowledge Bases and folders directly from the UI.

### Content auto-sync

An auto-sync feature is available for content created via URL or website crawls, as well as external imports. By default, it's disabled. Enable it per folder during import setup, or in the folder settings on the Knowledge Base page. Set the data update cadence (weekly, monthly, and so on) in the same place.

## How Brand Kits work

> [!TIP]
>
> Brand Kits will become part of the **Brand Context** in **Context Manager**. If you are starting out with Profound, we recommend focusing on refining your brand voice in your knowledge records.

Brand Kits analyze your brand's website to generate a brand identity profile covering brand information and writing style, which you can review and edit at any time. Anything that generates content on your behalf can apply a Brand Kit, so output matches your brand's voice and positioning.

### What a Brand Kit contains

When you create a Brand Kit by entering a brand's website URL, Profound automatically generates a profile broken down into three categories:

1. **Brand information:** A high-level overview of the brand, including its mission and values, ideal customer profile, main competitors, and brand point of view.
2. **Brand writing style:** An overview of how the brand communicates, including author persona, tone of voice, and specific writing rules.
3. **Brand writing sample:** A writing sample extracted from the brand's website and analyzed to calibrate the writing style.

All generated information can be reviewed and edited at any time from the Edit Brand Kit menu.

## Example use cases

### 1. Protect the brand's voice and style

Upload style guides, messaging frameworks, and approved published content to a Knowledge Base, and create a Brand Kit from your website. AI Marketer and [Agents](/agents/about) draw on both, so briefs, blog posts, and ad copy match your voice and positioning without you restating it in every request.

### 2. Check what AI says about your brand

Upload your source-of-truth documentation to a Knowledge Base and run [FactCheck](/answer-engine-insights/factcheck/about). It compares AI-generated answers about your brand against your documents, flags inaccurate claims, and shows which sources are behind them.

### 3. Ground generated content in facts

Connect content Workflows to your Knowledge Base to automatically cross-reference outputs against source documents before publishing to your CMS. This prevents errors in key details such as specs, pricing, and messaging.


# Create and manage Brand Kits
Source: https://docs.tryprofound.com/context-manager/manage-brand-kits

Create and edit Brand Kits in Context Manager

## Create a Brand Kit

Access Brand Kits from the **Context Manager** section in the left navigation bar, Brand Kits tab. To create a Brand Kit, click **+ Create new Brand Kit**.

This opens a prompt to input the URL of the brand's landing page. Profound analyzes the domain and **automatically** generates a brand identity broken down into two main categories: Brand Information and Writing Style. It also extracts a writing sample and generates a description of its style.

### Brand Information

Brand Information provides a high-level overview of the brand and includes:

* **About the brand:** Describes the brand's mission, values, and unique character.
* **Ideal customer profile:** Summary of key characteristics of target customers.
* **Competitors:** A list of the brand's main competitors.
* **Brand point of view:** The brand's unique perspective and stance.

### Brand Writing Style

Brand Writing Style provides an overview of how the brand communicates. It includes:

* **Author persona:** The character the brand writes as.
* **Tone of voice:** A summary of the brand tone.
* **Writing rules:** Specific writing rules the brand follows.

### Brand Writing Sample

A Brand Writing Sample is extracted from the brand's website and analyzed to help define the writing style.

## Edit a Brand Kit

Select any Brand Kit from the Brand Kits list to make edits. All generated information can be updated by modifying the text fields in the **Edit Brand Kit** window. Click **Save** at the bottom of the window to store your changes.


# Create and manage Knowledge Bases
Source: https://docs.tryprofound.com/context-manager/manage-knowledge-bases

Add documents, sync sources, and manage Knowledge Bases in Context Manager

Add and manage Knowledge Bases from the [Context Manager](/context-manager/about) section of the platform.

## Create a Knowledge Base

To add a new Knowledge Base:

1. In Profound platform, go to the **Context Manager** section in the left navigation bar. Select **Knowledge Base** tab at the top of the screen.

2. Select **+ Add Knowledge Base**.

3. Enter a name for the Knowledge Base.

4. Turn on **Sync with Brand Context** to let the documents in the Knowledge Base contribute to your Brand Context. Leave it off if you want to use this Knowledge Base only for file storage and retrieval.

5. Turn on **Limit to specific categories** to assign the contents to one or more specific [categories](/getting-started/categories), or leave it off to apply it's contents to all categories.

6. Select **Create** to confirm.

Once created, upload content and organize your files.

## Add files to your Knowledge Base

There are multiple ways to upload content to a Knowledge Base. Once uploaded, materials are indexed and made available to every part of Profound.

### Upload files

To add a file to your Knowledge Base:

1. Select an existing Knowledge Base.
2. Select **+ Add Files**.
3. Select the type of upload.

The available upload types include a URL scrape, a direct file upload, a website crawl, a file import from an external source, and pasting text into an editor.

### Scrape Specific URLs

Paste individual URLs to extract and store content from live webpages. This method is useful for importing content from sources such as blog posts or public product pages.

To scrape a single page:

1. Select **Scrape Page**.
2. Enter the full URL of the page you want to import.
3. Select the sync cadence: how often do you want Profound to refresh the information from this source.
4. Click **Import URL** to begin the process.

> [!NOTE]
>
> This option only scrapes the specific page referenced by the URL. It does **not** scrape every page within the domain. Use the **Crawl Website** option to scrape a full site.

### Upload files directly

Upload files directly into your Knowledge Base. This method works well for documentation such as guides, policies, or internal reference materials.

To upload a file:

1. Select **Upload Files**.
2. Use **Browse Files** to select a file from your device, or drag and drop a file into the upload area.

### Crawl Website

Index an entire site using a sitemap, or by crawling pages across a domain. Best used for documentation sites, help centers, or product catalogs.

To crawl a website:

1. Select **Crawl Website**.
2. Choose one of the following options:

* **Sitemap** – Points to a sitemap XML file, allowing the system to import all pages listed in the sitemap.
* **Crawl** – Progressively indexes pages across a domain. When selecting this option, configure additional options:
  * **maximum number of pages** to crawl, the default is 100
  * **maximum crawl depth**: the number of levels of the site structure to crawl (1 = crawl just the root page, 2 = root page and other pages linked from it, and so on). The default crawl depth is 3

Pages must be **publicly accessible** and not protected by login, authorization, or other credentials to be crawled.

> [!TIP]
>
> Crawling pages rendered with JavaScript has some known limitations. Some sites (usually belonging to large enterprise companies) block crawlers. If this is your case, consider adding knowledge in other ways.

### Import from an external source (Notion, Google Drive, SharePoint, Confluence, Granola)

Integrate your workspace, select files in the file picker, and import them to your Knowledge Base. Best used for internal documentation, similar use cases to direct file upload.

To import files from an external source:

1. Select **Import from ...** for your import source.
2. If prompted, sign in to authorize the integration.
3. Select files in the file picker.
4. Select **Import**.

To import from Confluence, sign in with your Atlassian account, then select the spaces and pages you want to import.

> [!NOTE]
>
> Granola import requires a Granola's personal API key. Learn how to create it in [Granola's documentation](https://docs.granola.ai/help-center/sharing/integrations/granola-api#getting-started).

### Paste text

Manually create a document by pasting or entering text directly in the text editor. Both structured and unstructured text are supported. Best used for creating new documents from scratch, or as a workaround when other methods don't work.

To paste text as a document:

1. Select **Paste Text**.
2. Provide the following information:
   * **Document name**
   * **Document format** (Markdown or Plain Text)
   * **Content**
3. Select **Add Document** to save the file to the Knowledge Base.

## Manage files and content

Knowledge Bases follow a folder structure modeled after a standard filesystem hierarchy. This lets you organize files the same way you would in a typical operating system, by creating folders and subfolders.

Within each Knowledge Base, you can:

* View indexed files
* Add tags to files
* Create new files
* Delete outdated material
* Download previously uploaded content

This structure makes it easy to organize documentation by product line, brand, or other logical groupings. Well-organized content improves search relevance, enhances Agent performance, and supports more effective long-term knowledge management.

## Choose which categories a Knowledge Base applies to

> [!TIP]
>
> Learn more about categories in [Categories in Profound](/getting-started/categories).

By default, every Knowledge Base syncs with **Brand Context** and applies to all categories in your organization. If your organization has more than one category, you can assign each Knowledge Base to the categories it belongs to so context from one category doesn't appear in another category's Brand Context.

Only users with access to a category can view and edit the Knowledge Bases and Brand Context for that category.

### Limit a Knowledge Base to select categories

To map a Knowledge Base to specific categories:

1. In the Knowledge Base settings, turn on **Sync with Brand Context**. This gives Context Manager access to its contents.
2. Select **Limit to select categories**, then select the categories the Knowledge Base applies to.

![Knowledge Base Settings panel with the Limit to specific categories toggle highlighted above the category list](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/context-manager/manage-knowledge-bases-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=f780a3a3578ed55e62988781eb07802d)

Only Knowledge Bases mapped to a category feed that category's Brand Context.

To change the category mappings of several Knowledge Bases at once, select them in the Knowledge Bases table, then select **Settings** at the bottom of the screen. Brand Context reorganizes automatically to reflect the new mappings.

To view the Knowledge Bases assigned to a specific category, use the category filter above the Knowledge Bases table.

![Context Manager Knowledge Base tab with the All categories dropdown highlighted next to search and Add Knowledge Base](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/context-manager/manage-knowledge-bases-2.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=ee17eed4fa1079fefb72444a71a720a1)


# Use Knowledge Bases and Brand Kits with Workflows
Source: https://docs.tryprofound.com/context-manager/using-with-workflows

Add Knowledge Base and Brand Kit nodes to Workflows

Knowledge Bases and Brand Kits are available as dedicated nodes that can be added to any [Workflow](/agents/workflows/about-workflows).

## Knowledge Base Workflow nodes

The following Workflow nodes use Knowledge Bases: [Prompt LLM](/agents/workflows/nodes/prompt-llm), Search Knowledge Base, Get Knowledge Base File, and Upload to Knowledge Base. Each node addresses a different stage of a Workflow.

### Prompt LLM node (with Use Knowledge Base enabled)

When Knowledge Base access is enabled within a [Prompt LLM](/agents/workflows/nodes/prompt-llm) node, the model references the Knowledge Base content to generate its output.

**Best used when:** You want the LLM to ground the generated content in your brand's approved documentation.

### Search Knowledge Base node

When you’re unsure which documents are most relevant, use this node to perform a search across your Knowledge Base. Results are ranked by relevance and passed into the following nodes for summarization, analysis, or content generation.

**Best used when:** The Workflow needs to find the right source document before acting. For example, it might retrieve the most relevant product spec before generating a comparison article.

### Get Knowledge Base File node

This node retrieves a specific file from your Knowledge Base, such as a brand style guide, product specification, or compliance document. The file is passed directly into subsequent workflow steps.

**Best used when:** The Workflow always needs the same document at a fixed point in the workflow.

### Write to Knowledge Base node

Enables Workflows to upload newly generated documents directly back into your Knowledge Base in one automated motion.

In the node configuration, specify the Knowledge Base (upload location), content, and a document name. If a document with that name already exists in the Knowledge Base, its contents are overwritten. If it doesn’t, a new document is created.

**Best used when:** You want to reuse a valuable Workflow node output (like competitive research reports or performance summaries) in future workflows.

## Brand Kit Workflow nodes

Brand Kits connect to Workflows through one dedicated retrieval node and three content generation nodes that accept Brand Kit variables as inputs.

### Use Brand Kit node

The Use Brand Kit node retrieves a single field from any of your saved Brand Kits and outputs it as a named variable. Available fields cover all Brand Kit sections: About the Brand, Ideal Customer Profile, Competitors, Brand Point of View, Author Persona, Tone of Voice, and Writing Rules.

**Best used when:** You want to pass a specific piece of brand context, such as tone of voice or writing rules, into a content generation or LLM step.

> [!TIP]
>
> The node retrieves one field at a time. If your workflow requires multiple fields, add a separate Use Brand Kit node for each, then reference all output variables in your downstream node.

### Prompt LLM node

[Prompt LLM](/agents/workflows/nodes/prompt-llm) can reference output variables from upstream Use Brand Kit nodes, letting you inject any Brand Kit field directly into the model's instruction prompt. This is independent of the node's Knowledge Base access option: use both in the same step, or either on its own.

**Best used when:** You need full control over how Brand Kit data is applied in a custom instruction. For example, pass Tone of Voice and Writing Rules as variables into a Prompt LLM node to rewrite content in the brand's established voice.

### Generate Article node

The [Generate Article](/agents/workflows/nodes/content-creation/generate-article) node accepts Brand Kit output variables in its Additional Instructions field. It also includes a built-in brand alignment pass: a final refinement stage that checks the article for brand-safe language, correct terminology, and consistent voice. Supplying Brand Kit data from upstream nodes gives this pass the information it needs to apply your specific brand standards.

**Best used when:** Building end-to-end automated content workflows where every generated article must reflect brand-approved tone, style, and positioning.

### Create Content Brief node

Brand Kit variables from upstream Use Brand Kit nodes can be passed into the [Create Content Brief](/agents/workflows/nodes/content-creation/create-content-brief) node via the Additional Instructions field, carrying brand context through to any downstream [Generate Article](/agents/workflows/nodes/content-creation/generate-article) step.

**Best used when:** You want to produce a content brief that already incorporates brand voice guidelines, ensuring writers and downstream generation steps are working from a brief that reflects both research and brand standards.

## Combining node outputs

Like all Profound data nodes, Knowledge Base nodes produce a named output variable that can be passed into any subsequent node that accepts text or structured data inputs.

For example, a Search Knowledge Base node can surface the most relevant product documentation, a Use Brand Kit node can retrieve the brand's tone of voice, a [Prompt LLM](/agents/workflows/nodes/prompt-llm) node can combine both to draft an accurate, on-brand comparison article, and a CMS node can publish the result directly. The Upload to Knowledge Base node saves finished outputs back into your Knowledge Base for reuse in future workflows.

Don't be afraid to experiment with different node combinations. The workflows that work best for your team will depend on your content goals, publishing cadence, and the structure of your Knowledge Base.


# Profound Documents
Source: https://docs.tryprofound.com/documents/about

Store, refine, and collaborate on generated content in the built-in Documents editor

Documents give you a built-in editor to store, refine, and collaborate on generated content without leaving Profound.

![Document editor with the slash command menu listing headings and list blocks](https://mintcdn.com/profound-37face47/V1WOgGibXMwY4fKe/images/documents/about-1.png?fit=max\&auto=format\&n=V1WOgGibXMwY4fKe\&q=85\&s=a2b6f43dc23c115f28fac8c9b7010d7b)

It works like other document collaboration services: rich text editing, real-time co-editing, and more. The difference is that it lives inside Profound, directly connected to your ​[Agents](/agents/about) and other areas of the platform.

## A persistent home for generated content

Documents give every platform tool output a place to live inside Profound. Instead of exporting or copying, open the result directly in a Profound Document, edit it in place, and share it with your team. Your colleagues can join in real time, leave comments, and iterate on the content alongside you.

[Workflows](/agents/workflows/about-workflows) can also read and write to Documents through Profound Docs [Workflow nodes](/agents/workflows/about-workflows#workflow-components), so you can build Workflows where Documents act as the handoff point between one Workflow and the next.

## Key features

### Rich text editor

Write and format content using an inline toolbar or slash command menu. The editor supports the formatting you'd expect: headings, lists, bold text, links, and more.

### Real-time collaboration

Multiple people can work on the same Document at once. Collaborator avatars appear on the page so you always know who's viewing or editing.

### Comments with resolution

Leave inline comments on any part of a Document, reply to threads, and resolve them when the feedback is addressed.

### Integration with Agents and Aim

[Agents](/agents/about) and [AI Marketer](/aim/about) can create, read, update, and delete Documents.

### Sharing inside and outside your organization

Share a Document with colleagues in your organization, or with people outside it, such as clients, agencies, and partners. Everyone you invite to view the document needs a Profound account. People who don't have one yet are prompted to create a free account when you share with them.

## Getting started

### Step 1: Create a Document

To create a Document manually, go to [docs.profound.new](http://docs.profound.new) to open a blank Document instantly, or select Documents from the left navigation sidebar in Profound.

To save a [Workflow](/agents/workflows/about-workflows) output into a Profound Document, include **Profound Docs** [Workflow nodes](/agents/workflows/about-workflows#workflow-components) (Create Doc or Update Doc) in any content Workflow. Alternatively, go into any Workflow’s **Past runs** and open the output in Documents. To save an [Agent](/agents/about) output as a Document, declare a Document as one of the Agent’s [outputs](/agents/building-blocks-of-agent#outputs).

### Step 2: Iterate

Once your generated content is ready, review your Document manually, then optionally feed the refined version into an Agent for further optimization.

### Step 3: Collaborate

Share the Document with colleagues in your Profound organization. They'll see your real-time presence, can leave comments on any section, and can edit alongside you.

### Step 4: Share

Share the Document with other organizations or partners outside of Profound. Select **Share** in the top right of the Document and enter the person's email address. Choose whether they can view, comment, or edit.

Only the Document's owner can invite people from other organizations; editors can't share the Document further with external users. Documents shared across organizations are labeled as externally shared in your document list.

> [!NOTE]
>
> A Document you receive from another organization belongs to that organization: you can't add it to your Knowledge Base, and AI Marketer can't reference it. To use its content in your own organization, duplicate the Document.


# Adobe Experience Manager (S2S)
Source: https://docs.tryprofound.com/integrations/adobe-experience-manager/overview

Let Workflows find, read, create, and update Content Fragments in AEM as a Cloud Service

Connect Adobe Experience Manager (AEM) to Profound and let Workflows find, read, create, and update Content Fragments directly in your AEM as a Cloud Service instance. The integration authenticates with an Adobe OAuth server-to-server (S2S) credential, so Workflow work with your content without an individual user login.

## Why the Adobe Experience Manager (S2S) integration exists

Enterprise content teams manage thousands of structured Content Fragments in AEM. When Profound surfaces a content gap or an outdated claim in AI answers, acting on it means exporting the insight, finding the right fragment in AEM, and editing it by hand. This integration closes that loop: Workflows read your existing fragments, create new ones, and push updates back to AEM without leaving Profound.

## Using Adobe Experience Manager nodes in Profound Workflows

Build Workflows that audit, refresh, and update AEM content based on Profound insights.

> [!TIP]
>
> Follow Profound’s [AEM Integration Setup Guide](/integrations/adobe-experience-manager/setup/aem-setup-guide) to set up your Adobe Experience Manager (S2S) integration and make AEM nodes available to Workflows.

Workflow nodes available for Adobe Experience Manager include:

* **List Content Fragments** — lists the Content Fragments in a folder of your AEM instance
* **Get Content Fragment** — reads the full content of a specific Content Fragment
* **Create Content Fragment** — creates a new Content Fragment in your AEM instance
* **Update Content Fragment** — updates the content of an existing Content Fragment

Learn more about the nodes, configuration and troubleshooting in [Using Adobe Experience Manager in Workflows](/integrations/adobe-experience-manager/using-adobe-experience-manager-in-workflows).

## Example Workflow: product FAQ refresh

This Workflow keeps your AEM-hosted product FAQs aligned with your current messaging. It writes revised copy back to AEM, then posts a Slack summary so your content team can review and publish the changes.

1. **AEM - List Content Fragments**: Point the node at the AEM folder that holds the Content Fragments for the FAQ you want to review. The node returns the fragments with their identifiers.
2. ​**Code node**: Turn the output of the previous step into the list the Iteration node can use.
3. [**Iteration node**](/agents/how-tos/add-iterations-to-workflows): Loop over the list of fragments, run the following steps for each, and return the list of updated AEM Content Fragments
   1. **AEM - Get Content Fragment**: Read the fragment's full content.
   2. [Prompt LLM node](/agents/workflows/nodes/prompt-llm) with the **Use Knowledge Base** option enabled: Compare the fragment's copy against your Knowledge Base. Return revised copy and a flag indicating whether the fragment needs changes.
   3. [**Conditional node**](/agents/how-tos/add-conditionals-to-workflows): If the fragment needs no changes, move on to the next fragment. If it does, continue to the next step.
   4. **AEM - Update Content Fragment**: Write the revised copy back to the fragment in AEM.
4. [Slack](/integrations/slack/connect-slack-to-profound): After the loop finishes, post the list of updated fragments returned by the Iteration node to your content team's Slack channel so they can review and publish the changes in AEM.

Schedule this Workflow to run once or twice a week so your product FAQ content stays current.


# AEM Integration Setup Guide
Source: https://docs.tryprofound.com/integrations/adobe-experience-manager/setup/aem-setup-guide



This guide walks you through connecting Adobe Experience Manager (AEM) as a Cloud Service to Profound using an Adobe OAuth Server-to-Server credential. Once connected, Profound Workflows can access, create, and update the Content Fragment models, folders, and content allowed by the credential's AEM permissions.

## Before you start

You'll need:

* An [AEM as a Cloud Service](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/overview/introduction) program and **Author** environment
* [AEM Cloud Manager](https://experience.adobe.com/#/cloud-manager) access to commit to an Adobe-managed repository and run a config pipeline (**Deployment Manager** role required)
* Permissions to manage product profiles in the [Adobe Admin Console](https://adminconsole.adobe.com/) and to create OAuth credentials in the [Adobe Developer Console](https://developer.adobe.com/console)
* A [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Select the AEM environment**

* Open [Adobe Cloud Manager](https://experience.adobe.com/#/cloud-manager).
* Sign in under the organization that owns the AEM program you intend to use with Profound.
* Select and open the program, then select **Environments** in the left sidebar.
* Open the environment you intend to use and copy its base **Author URL**. It should look like: `https://author-p12345-e67890.adobeaemcloud.com`.

  ![Adobe Cloud Manager program overview page screenshot with Environments menu option and Author URL highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-author-url.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=42c8c2c078ee518a383da531daad117f)

**2. Prepare your AEM environment for API access**

To enable API access in your environment, follow [Adobe's AEM APIs setup guide](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/setup).

Once done, confirm that you have the correct product profile to use with the AEM API:

* Go back to your AEM program environment page in [Adobe Cloud Manager](https://experience.adobe.com/#/cloud-manager) (where you left off in [Step 1](#step-1)).
* Open the environment's action menu and select **Manage Access** > **Author Profiles**. [Adobe Admin Console](https://adminconsole.adobe.com/) opens your environment's page.

  ![Adobe Cloud Manager environment action menu screenshot with Manage Access and Author Profiles options highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-env-manage-access.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=1f68022443bb0a3201c9ff3ccbf60fba)
* Confirm that the **AEM Sites Content Managers - author...** profile exists for your program and environment.

  ![Adobe Admin Console Product profiles list screenshot with the AEM Sites Content Managers author profile highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-content-managers-profile.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=3fc3ef6e366cca3ed72730a47c2b2a25)

**3. Assign a Developer role to the credential creator**

To authenticate with Profound, create an OAuth credential. You, or anyone with a **Developer** role in your **AEM Sites Content Managers** product profile, can create it.

To assign a **Developer** role to the credential creator, continue where you left off in [Step 2](#step-2):

* On your environment's page in [Adobe Admin Console](https://adminconsole.adobe.com/), select the **AEM Sites Content Managers - author...** product profile.
* Open the **Developers** tab, then select **Add developers**.

  ![Adobe Admin Console product profile screenshot with the Developers tab and Add developers button highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-add-developer.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=344874f02a4457f4695ffb351d71bb55)
* Add the email or username of the person creating the OAuth credential, then select **Save**.

> [!TIP]
>
> Learn more about assigning developers to Adobe product profiles in Adobe's [Manage developers in Adobe Admin Console](https://helpx.adobe.com/enterprise/using/manage-developers.html) guide.

**4. Create the OAuth Server-to-Server credential**

* Open the [Adobe Developer Console](https://developer.adobe.com/console). If you're following these steps in sequence, you should already be signed in under the organization that owns your Author AEM environment (same as in [Step 1](#step-1)).
* Select **Create new project** in the **Quick start** section, or go to **Projects** in the top navigation menu and select **Create new project**.

  ![Adobe Developer Console home page quickstart section screenshot with Create new project option highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-dev-console-create-project.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=4976d8aa25b741a477d2cb3b50ba158d)
* On the project page, select **Edit project**, give your project a descriptive name (such as **Profound integration credentials**), then select **Save**.

  ![Adobe Developer Console Edit Project dialog screenshot with the Edit project button and the Project Title field highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-dev-console-edit-project.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=57dd5b9c83524aee97d71c8018117610)
* Select **Add to Project** > **API**.

  ![Adobe Developer Console project page screenshot with Add to Project and the API option highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-dev-console-project-add-api.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=dd25b6dfd1305627eb727fc9ea5f55be)
* On the **Add an API** page, filter the API list by **Experience Cloud**, select **AEM CS Sites Content Management**, then select **Next**.

  ![Add an API view screenshot with Experience Cloud filter option, AEM CS Sites Content Management option, and Next button highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-dev-console-cs-sites-content-management.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=e5e114839ec4ecfb03f113b765d3d5af)
* Select **Server-to-Server** > **OAuth Server-to-Server** as the authentication type.

  ![Configure API dialog screenshot with Server-to-Server Authentication and Next highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-dev-console-s2s-auth.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=63392365f06c8098d1a9d85e1b46687f)

  ![Configure API dialog screenshot with OAuth Server-to-Server, the Credential name field, and Next button highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-configure-api-add-s2s-oauth.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=468030a62ea0ee9e1b9cda0a86c4e351)
* Select the Author product profile you prepared in [Step 2](#step-2), then save the configuration.

  ![Configure API dialog screenshot with the AEM Sites Content Managers product profile selected and Save configured API option highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-configure-api-select-profile.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=63261c319a90c159ca5c1858d15e1c99)

> [!NOTE]
>
> The AEM Sites Content Management API also provides the folder access Profound uses. No additional credentials are required.

**5. Retrieve the connection values**

After you save the API configuration, [Adobe Developer Console](https://developer.adobe.com/console) returns you to the project page. In the **Credentials** section, select your new OAuth Server-to-Server credential. From the credential page, copy the values you need to create the AEM integration in Profound:

* **Client ID**
* **Client Secret**
* **Scopes** — copy the complete value exactly as Adobe displays it
* **Organization ID** — optional; copy it only if the AEM tenant requires the `x-gw-ims-org-id` header

  ![OAuth Server-to-Server credential details view screenshot with Client ID, Client Secret, Scopes, and Organization ID highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-connection-values.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=33a7ec41d13549bd3ae8ac6ed42993d9)

**6. Register the client ID with the AEM environment**

Before you can use the credential, register its client ID with your Author environment.

* Back in [Adobe Cloud Manager](https://experience.adobe.com/#/cloud-manager), open the program and select **Repositories** in the left sidebar.
* Identify the Adobe-managed repository and branch used by your environment's config pipeline.

  ![Adobe Cloud Manager Repositories screenshot with Repositories menu option, repository name, and the configured pipeline branch highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-repositories.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=485eedb44a7d7f739f739f236640ff63)
* Select **Access Repo Info** and copy the repository URL and access credentials.
* Using your regular development workflow, clone the repository onto your machine and create or update the `config/api.yaml` file in the root directory with the following configuration:
  ```yaml theme={null}
  kind: "API"
  version: "1"
  metadata:
    envTypes: ["dev", "stage", "prod"]
  data:
    allowedClientIDs:
      author:
        - "<CLIENT_ID>"
  ```

Replace `<CLIENT_ID>` with the client ID you copied from Developer Console in [Step 5](#step-5).

If you already have other API credentials configured, keep the existing configuration and add the client ID to the `author` list under `allowedClientIDs`.

> [!TIP]
>
> Learn more about managing repositories and configuring pipelines in the following Adobe resources:
>
> * [AEM API client ID registration guide](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/developing/open-api-based-apis#registering-a-client-id)
> * [Manage repositories in Cloud Manager](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/using-cloud-manager/managing-code/managing-repositories)
> * [Use config pipelines](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/operations/config-pipeline) guide
> * [Invoke OpenAPI-based AEM APIs using OAuth Server-to-Server authentication](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/invoke-api-using-oauth-s2s) tutorial

**7. Deploy API configuration with a config pipeline**

* Still on your program page in [Adobe Cloud Manager](https://experience.adobe.com/#/cloud-manager), select **Pipelines** in the left sidebar.
* Select **Add Pipeline**, then select **Add Non-Production Pipeline** for the development environment, or **Add Production Pipeline** for stage and production.

  ![Adobe Cloud Manager Pipelines screenshot with the Pipelines menu option and Add Pipeline button highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-add-pipeline.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=d5a8bd283871ac52c0ed6a4d9e5a068a)
* In the **Add Pipeline** dialog, select **Deployment Pipeline**, give your pipeline a descriptive name, then select **Continue**.
* On the next screen, fill in the values as follows:
  * Select code to deploy: **Targeted deployment**
  * Include: **Config**
  * Eligible Deployment Environments: select your AEM environment from the dropdown
  * Repository, Git Branch, Code Location: the repository, branch, and config directory from [Step 6](#step-6)
* Select **Save**.

Your new config pipeline appears in the list of pipelines in the **Pipelines** section. Select the pipeline, then select **Run selected**. Wait for the deployment to complete successfully.

![Adobe Cloud Manager Pipelines screenshot with a new config pipeline selected and a Run selected button highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-pipeline-run.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=f718730ffb2ea2526a06a6ce2b1cdfcd)

> [!TIP]
>
> Learn more about deploying config pipelines in the following Adobe resources:
>
> * [Use config pipelines](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/operations/config-pipeline) guide
> * [Cloud Manager CI/CD pipelines](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/using-cloud-manager/cicd-pipelines/introduction-ci-cd-pipelines) overview

**8. Add an AEM S2S integration in Profound**

* In [Profound](https://platform.tryprofound.com), select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Adobe Experience Manager (S2S)** and select it.

  ![Profound Integrations view screenshot with Adobe Experience Manager (S2S) integration card highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-s2s-integration-card.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=bbc744a5c70df84ac351a72fe10642ed)
* In the Adobe Experience Manager (S2S) integration view, select **+ Add account**.
* In the **Connect Adobe Experience Manager (S2S)** dialog, enter the Author URL from [Step 1](#step-1) and the client secret, client ID, scopes, and organization ID you retrieved in [Step 5](#step-5).
* Select **Connect account**.

If the connection succeeds, your AEM environment appears in the list of connected accounts.

![Profound Adobe Experience Manager integration view screenshot with a connected account highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-profound-integration.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=4170d3dfd15b7309d5a86171dd5bc7f0)

Revoke Profound's access to your AEM environment at any time by selecting **Revoke** in the integration account action menu.

![Profound connected account action menu screenshot with the three-dot menu and the Revoke option highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-integration-revoke.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=9ba55d38e6c349de6b38b98830e71bf1)

**9. Verify the connection**

To make sure Profound has all the permissions it needs to work with your AEM environment, verify that the AEM Workflow nodes run successfully.

* In [Profound](https://platform.tryprofound.com), [create a new Workflow](https://help.tryprofound.com/articles/2212787792-create-an-agent), or open an existing Workflow draft.
* Add the Adobe Experience Manager **List Content Fragments** and **Get Content Fragment** nodes, then select the connected AEM instance.
* Select **Test node** (a play icon) in the node configuration panel and confirm that the AEM output contains the expected data.

  ![List Content Fragments node configuration panel screenshot with the play icon highlighted](https://mintcdn.com/profound-37face47/VadJiYMKj-yl6ums/images/integrations/aem/aem-test-run-agent.png?fit=max\&auto=format\&n=VadJiYMKj-yl6ums\&q=85\&s=f17bb8614765c9ad4aa26a037e4b21ac)
* If you intend to create or update content fragments with Profound Workflows, test the **Create Content Fragment** and **Update Content Fragment** nodes the same way.

> [!NOTE]
>
> Delete test content directly in the AEM environment. Current Profound Workflow nodes don't include a delete operation.

## Rotating AEM credentials

### Rotate a client secret only

Adobe supports two client secrets on one OAuth Server-to-Server credential, which allows rotation without changing the client ID.

Rotate your client secret as instructed in [Adobe developer documentation](https://developer.adobe.com/developer-console/docs/guides/authentication/ServerToServerAuthentication/implementation#rotating-client-secrets), then update the credentials in Profound (**Integrations** > **Adobe Experience Manager (S2S)** > **Edit credentials** in a connected account's action menu).

The client ID doesn't change during this process, so you don't need to update or redeploy `api.yaml`.

### Rotate the entire OAuth credential

To replace the entire AEM OAuth credential:

1. Generate a new credential as described in [Step 4](#step-4) of the setup guide.
2. Deploy the new client ID in `config/api.yaml` as instructed in [Step 6](#step-6) of the setup guide.
3. Update the credentials in Profound.

> [!TIP]
>
> Keep the old client ID registered until you've verified the new connection.

## Best practices

* Use an **Author** environment only. Profound doesn't connect to AEM Publish or Preview. Learn more about environments in [AEM as a Cloud Service tutorials](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/cloud-manager/environments).
* Test the read and write Profound Workflow nodes separately to verify their folder and repository permissions.
* Always store your client secret and other credentials in an approved secret manager, and never include them in screenshots, tickets, or source control.


# Troubleshooting AEM Integration Setup
Source: https://docs.tryprofound.com/integrations/adobe-experience-manager/setup/aem-setup-troubleshooting



## OAuth Server-to-Server authentication method is not available

Confirm that the credential creator is assigned as a **Developer** on the intended product profile and that the profile includes the AEM Sites service, as described in [Step 2](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-2) and [Step 3](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-3) of the setup guide.

## AEM CS Sites Content Management product profile is missing

Go back to [Step 2](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-2) of the setup guide and verify the following:

* Your Adobe organization owns the AEM environment you intend to use with Profound.
* The environment is modernized for API access as instructed in Adobe's [Set up OpenAPI-based AEM APIs](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/setup) guide.

## Profound can't save the AEM credentials

Confirm the following:

* All connection values are correct and match the values of your OAuth credential from [Step 5](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-5) of the setup guide.

* Your OAuth credential uses the **AEM Sites Content Managers - author...** product profile. Check this in the [Adobe Developer Console](https://developer.adobe.com/console). If it doesn't, repeat [Step 2](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-2) of the setup guide.

* The config pipeline that deploys your API configuration (`api.yaml`) ran successfully. Check the **Pipelines** section of your AEM program page in [Adobe Cloud Manager](https://experience.adobe.com/#/cloud-manager). If it didn't, repeat [Step 6](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-6) and [Step 7](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-7) of the setup guide.

* The client ID of the OAuth credential is in the correct place in the `api.yaml` configuration file, as instructed in [Step 6](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-6) of the setup guide.

* An Author IP allowlist isn't blocking Profound. Learn more in Adobe's [Manage IP Allow Lists](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/using-cloud-manager/ip-allow-lists/managing-ip-allow-lists) guide.
  > [!TIP]
  >
  > If you have to apply an Author IP allowlist, contact Profound Support to obtain the current Profound outbound IP addresses rather than reusing values from an old guide or ticket.

* If you're using a sandbox environment, it's online and not hibernated. Learn more in Adobe's [Hibernate and De-Hibernate Sandbox Environments](https://experienceleague.adobe.com/en/docs/experience-manager-cloud-service/content/implementing/using-cloud-manager/programs/hibernating-environments) guide.

## Content Fragment Models don't appear in Profound

This usually means either the Content Fragment Model isn't allowed for the folders you use with Profound, or your OAuth credential's technical account lacks the permissions to manage Content Fragment Model configuration.

Confirm the following:

* The Content Fragment Model is enabled and allowed on your assets folder. Learn more in Adobe's [Content Fragment Models](https://experienceleague.adobe.com/en/docs/experience-manager-65/content/assets/content-fragments/content-fragments-models#enabling-disabling-a-content-fragment-model) documentation.
* The technical account's service group has permissions to manage the Content Fragment Model configuration. With the setup described in the [setup guide](/integrations/adobe-experience-manager/setup/aem-setup-guide#setup-guide), your technical account should already have the necessary permissions to read and manage your AEM content. If it doesn't, manage permissions for specific folders and files in your AEM environment as described in Adobe's [Product Profile and Services user group permission management](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/how-to/services-user-group-permission-management) guide.

> [!TIP]
>
> Grant the AEM OAuth credential's technical account the narrowest permissions that support the Workflow nodes you intend to use in Profound. Scope create and update access to approved folders only.

## Folders don't appear in Profound

* Go back to [Step 5](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-5) of the setup guide and confirm that the **Scopes** connection value for your Profound AEM integration matches the scopes value of your AEM OAuth credential.
* Confirm that the scopes value includes `aem.fragments.management` and `aem.folders`. If it doesn't, don't add the missing scopes by hand. Confirm instead that you completed all prerequisites in [Step 2](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-2) and [Step 3](/integrations/adobe-experience-manager/setup/aem-setup-guide#step-3) of the setup guide.
* Confirm that the technical account's service group has the permissions it needs to manage content in the `/content/dam` folder in your AEM environment. To add permissions for specific folders and files, follow Adobe's [Product Profile and Services user group permission management](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/how-to/services-user-group-permission-management) guide.

## Create or Update Content Fragment node returns a permissions error

This usually means your AEM OAuth credential is authenticated but doesn't have the required repository access. Follow Adobe's [Product Profile and Services user group permission management](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/how-to/services-user-group-permission-management) guide to grant the service group that contains your AEM technical account the create or update permissions it needs for the folder you use with Profound.


# Using Adobe Experience Manager in Workflows
Source: https://docs.tryprofound.com/integrations/adobe-experience-manager/using-adobe-experience-manager-in-workflows



After you connect your Adobe Experience Manager (AEM) account, the following nodes become available in Profound Workflows. Use them to find, read, create, and update AEM Content Fragments. All operations run against the connected AEM Author environment and use the permissions assigned to the OAuth credential's technical account.

## Available nodes

**List Content Fragments**

### List Content Fragments

Retrieve Content Fragments from the connected Author environment.

#### Required inputs

* **AEM Instance** — the connected Author environment

#### Optional inputs

* **Folder** — limits the list to an approved folder below `/content/dam`
* **References** — controls whether linked fragments are returned and hydrated
* **Limit** — number of results to return; the default is `25` and the maximum is `50`
* **Cursor** — a pagination cursor from a previous List or Search node's output, used to fetch the next page of results
* **Projection** — use `summary` when you need summary data only

#### Output

The output contains the matching Content Fragments and, when more results are available, a pagination cursor for the next page.

**Search Content Fragments**

### Search Content Fragments

Find Content Fragments matching one or more criteria.

#### Required inputs

* **AEM Instance** — the connected Author environment

> [!NOTE]
>
> Profound starts the search with a **Folder** filter of `/content/dam`. Change that folder or add another filter as needed.

#### Search filters

* **Search text** — searches the AEM-supported fragment content for the selected text
* **Folder** — scopes the search to a DAM folder path
* **Models** — limits results to selected Content Fragment Models
* **Statuses** — New, Draft, Published, Modified, or Unpublished
* **Locale** — limits results to a locale
* **Direct children only** — excludes fragments in nested folders
* **Created after** and **Created before** — constrain the creation date

#### Optional inputs

* **Raw Query** — an AEM-supported search-query JSON; use it for filters or sorting the node configuration form doesn't include
* **Limit** — number of results to return; the default is `25` and the maximum is `50`
* **Cursor** — a pagination cursor from a previous List or Search node's output, used to fetch the next page of results
* **Projection** — use `summary` when you need summary data only

#### Output

The output contains the matching Content Fragments and, when more results are available, a pagination cursor for the next page.

**Get Content Fragment**

### Get Content Fragment

Retrieve the complete data for one Content Fragment.

#### Required inputs

* **AEM Instance** — the connected Author environment
* **Content Fragment** — select a fragment from the dropdown, or use a variable from a previous node's output

#### Optional inputs

* **Content Fragment ID** — the fragment UUID as text, which overrides the fragment selected in **Required inputs**
* **References** — controls whether linked fragments are returned and hydrated

#### Output

The output includes the Content Fragment data and its ETag when available. Model-specific fields follow the response AEM returns.

**Create Content Fragment**

### Create Content Fragment

Add Author content based on an enabled Content Fragment Model.

#### Required inputs

* **AEM Instance** — the connected Author environment
* **Parent Folder** — the destination folder below `/content/dam`
* **Content Fragment Model** — the model that defines the fragment fields
* **Title** — the fragment title
* **Data format**
  * **Structured** — loads the selected model and displays its fields in the configuration panel
  * **Raw JSON** — accepts an object whose keys match the AEM model's field names
* **Fields** / **Raw JSON** — the structure for your input data, based on the selected **Data format**

> [!TIP]
>
> Use Raw JSON when the Content Fragment Model is supplied with a variable, because the node editor can't load that model's field definition in advance.

#### Optional inputs

* **Name** — URL-safe Content Fragment name; derived from the **Title** when left blank
* **Description** — fragment description

#### Output

The output includes the created Content Fragment, its ETag, and its Location when AEM provides those values.

> [!NOTE]
>
> Creating or updating a fragment doesn't publish it. Review and publish the content in AEM.

**Update Content Fragment**

### Update Content Fragment

Change an existing fragment. The node modifies only the fields you provide.

#### Required inputs

* **AEM Instance** — the connected Author environment
* **Content Fragment** — select a fragment from the dropdown, or use a variable from a previous node's output
* **Data format**
  * **Structured** — loads the selected model and displays its fields in the configuration panel
  * **Raw JSON** — accepts an object whose keys match the AEM model's field names
* At least one change to **Title**, **Description**, or **Fields** / **Raw JSON**

> [!TIP]
>
> Use Raw JSON when the Content Fragment Model is supplied with a variable, because the node editor can't load that model's field definition in advance.

#### Optional inputs

* **Title** — leave empty to keep the current title
* **Description** — leave empty to keep the current description
* **Fields** / **Raw JSON** — the structure for your input data, based on the selected **Data format**
* **References** — controls whether linked fragments are returned and hydrated
* **ETag** — optimistic-lock value from an earlier Get or Update output. When left blank, Profound reads the current ETag immediately before sending the update. If the provided ETag is stale, the node reports a conflict.

#### Output

The output includes the updated Content Fragment and its new ETag when AEM provides one.

> [!NOTE]
>
> Creating or updating a fragment doesn't publish it. Review and publish the content in AEM.

***

## Key concepts

### AEM Instance

The connected AEM Author environment. Select it before choosing a model, folder, or fragment.

### Content Fragment Model

The AEM schema that defines a fragment's fields, types, required values, and allowed content. The model must be enabled and allowed by the destination folder's policy.

### DAM paths

Content Fragment folders use repository paths below `/content/dam`. Folder permissions and policies control which content Profound can read or change.

### Content Fragment ID

AEM API operations identify a Content Fragment by UUID. Profound's picker displays searchable fragment information and supplies the UUID to the node.

### ETag

An ETag is a unique identifier for a specific version of a Content Fragment's data. Pass it from a **Get Content Fragment** node to the **Update Content Fragment** node to prevent the workflow from silently overwriting a change made after the fragment was read.

### Pagination

List and Search nodes return up to `50` fragments per request. When the output contains a cursor, provide it to the next List or Search node to retrieve the next page of results.

### Hydrated Content Fragment

A Content Fragment that contains its full field values, rather than only basic IDs and URLs.

### References

References control how linked fragments are included. Hydrated and all-reference modes can return larger responses, so use them only when the workflow needs linked content.

***

## Troubleshooting

### A folder, fragment, or model picker is disabled or empty

Select an **AEM Instance** first. If the picker stays empty, verify that your OAuth credential's technical account can read the corresponding model, folder, or fragment. Learn more in the [setup troubleshooting](/integrations/adobe-experience-manager/setup/aem-setup-troubleshooting) guide.

### Search returns an invalid-query error

Restore the **Folder** filter to `/content/dam` or add another filter. When using **Raw Query**, include a non-empty JSON object and use Adobe-supported filter and sort fields.

### Create node reports an unknown field or invalid value

Select the model and use **Structured** format when possible. If you have to use **Raw JSON** format, make sure each key matches the AEM model field name and provide a value with the model's expected type and cardinality.

### Update node reports that there are no changes

Enter at least one **Title**, **Description**, or model field value. Empty inputs preserve the current content.

### Update node reports a conflict

The fragment changed after AEM issued the ETag. Run the **Get Content Fragment** node again, review the latest content, then apply the update with the new ETag.

### A node returns a permissions error

Confirm that the technical account's service group has permission for that exact operation and DAM path. Read access can succeed while Create or Update remains blocked. To add permissions for specific folders and files in your AEM environment, follow Adobe's [Product Profile and Services user group permission management](https://experienceleague.adobe.com/en/docs/experience-manager-learn/cloud-service/aem-apis/openapis/how-to/services-user-group-permission-management) guide.

### Only the first page of results is returned

Use the cursor returned by the List or Search node as the next request's **Cursor**. Each page can contain no more than `50` results.

***

## Additional resources

* [AEM Sites Content Management API reference](https://developer.adobe.com/experience-cloud/experience-manager-apis/api/stable/sites/)
* [AEM Folders API reference](https://developer.adobe.com/experience-cloud/experience-manager-apis/api/stable/folders/)


# Connect Contentful to Profound
Source: https://docs.tryprofound.com/integrations/contentful/connect-contentful-to-profound



## Prerequisites

Before connecting Contentful to Profound, ensure:

* You have access to your Contentful organization
* You have **Admin** or **Owner** permissions for the space you want to connect
* Your Contentful account is active

## Connect Your Account

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **Contentful**, select it, then select **+ Add account**.
4. You'll be redirected to Contentful to authorize Profound.
5. Sign in to your Contentful account and grant the requested permissions.
6. Select the organization and space you want to connect.
7. Once authorized, you'll be redirected back to Profound.

Once connected, Contentful will appear as an available integration and can be selected inside Profound Workflows.

> [!NOTE]
>
> Profound requests **content\_management\_manage** scope, which allows creating, updating, and publishing content in your connected spaces.


# Using Contentful in Workflows
Source: https://docs.tryprofound.com/integrations/contentful/using-contentful-in-workflows



After connecting your Contentful account, the following actions become available as Workflow steps. Each step requires you to select a connected **Contentful Account** from a dropdown.

**Create entry**

#### **Create Entry**

Create a new entry in your Contentful space.

**Required inputs**

* **Contentful Account**
* **Space**
* **Environment**
* **Content Type**
* **Fields** (dynamic based on content type schema)

**Optional inputs**

* **Locale** (defaults to space's default locale)
* **Publish** (publish immediately or save as draft)

The entry is created based on your content type's schema. Rich text fields accept Markdown which is automatically converted to Contentful's rich text format.

**Tip:** Leave **Publish** unchecked to create drafts for review before publishing.

**Update entry**

#### **Update Entry**

Update an existing entry in your Contentful space.

**Required inputs**

* **Contentful Account**
* **Space**
* **Environment**
* **Entry ID**

**Optional inputs**

* **Fields** (only fields you want to update)
* **Locale** (defaults to space's default locale)
* **Publish** (publish changes immediately)

**Important:** Only the fields you provide will be modified; all other fields remain unchanged.

**Get entry**

#### **Get Entry**

Retrieve a single entry by ID.

**Required inputs**

* **Contentful Account**
* **Space**
* **Environment**
* **Entry ID**

The output includes the entry's full content with all fields and metadata.

**List entries**

#### **List Entries**

Retrieve a list of entries from your Contentful space.

**Required inputs**

* **Contentful Account**
* **Space**
* **Environment**

**Optional inputs**

* **Content Type** (filter by content type)
* **Limit** (number of entries to return)
* **Skip** (pagination offset)

The output is a structured list of entries that can be used in downstream Workflow steps.

***

## Supported Field Types

Contentful's dynamic schema means Profound automatically adapts to your content model. All field types are supported:

| Field Type | Description | Input Format |
| - | - | - |
| **Short text** | Single-line text | Plain text |
| **Long text** | Multi-line text | Plain text |
| **Rich text** | Formatted content | Markdown (auto-converted) |
| **Number** | Decimal numbers | Number |
| **Integer** | Whole numbers | Integer |
| **Boolean** | True/false | Boolean |
| **Date & time** | ISO 8601 dates | Date string |
| **Location** | Geographic coordinates | `{lat, lon}` object |
| **JSON object** | Arbitrary JSON | JSON object |
| **Media** | Asset references | Asset ID |
| **Reference** | Entry references | Entry ID or array of IDs |

***

## Working with Rich Text

Rich text fields in Contentful use a structured document format (AST). Profound accepts **Markdown**, **HTML**, or **Contentful's native rich text JSON** format.

**Supported Markdown:**

* Headings (`# H1` through `###### H6`)
* Paragraphs
* **Bold**, *italic*, and other formatting
* Bullet and numbered lists
* Blockquotes
* Code blocks
* Links

**Supported HTML:**

* `<h1>` through `<h6>` headings
* `<p>` paragraphs
* `<strong>`, `<em>`, `<u>` formatting
* `<ul>`, `<ol>`, `<li>` lists
* `<blockquote>` blockquotes
* `<code>`, `<pre>` code blocks
* `<a>` links

**Native Rich Text JSON:**

If you already have Contentful rich text documents, you can pass them directly:

```json theme={null}
{
  "nodeType": "document",
  "data": {},
  "content": [
    {
      "nodeType": "paragraph",
      "data": {},
      "content": [
        { "nodeType": "text", "value": "Hello world", "marks": [] }
      ]
    }
  ]
}
```

**Tip:** Use whichever format is most convenient for your workflow. Markdown, HTML, and native rich text JSON all work seamlessly.

***

## Working with References

Reference fields link entries together. You can provide:

* **Single reference**: Just the entry ID as a string
* **Multiple references**: An array of entry IDs

Profound automatically compiles these into Contentful's required link format.

***

## Understanding Locales

Contentful uses a locale-based content model where every field value is keyed by locale (for example, `en-US`, `fr-FR`). Profound handles this automatically:

* **Default behavior**: If you don't specify a locale, Profound uses your space's default locale
* **Single locale**: Specify a locale to write content in that language
* **Multi-locale**: Use the raw JSON format to write multiple locales at once

> **How locale wrapping works**
>
> When you provide a simple field value like:
>
> ```json theme={null}
> {
>   "title": "My Blog Post"
> }
> ```
>
> Profound automatically wraps it for Contentful's API:
>
> ```json theme={null}
> {
>   "title": {
>     "en-US": "My Blog Post"
>   }
> }
> ```
>
> This locale wrapping happens for all fields before submission to Contentful.

***

## Input Formats

Profound accepts two input formats for creating and updating entries:

**Simple format**

The simple format is human-friendly and handles locale wrapping automatically:

```json theme={null}
{
  "space": "your-space-id",
  "environment": "master",
  "content_type_id": "blogPost",
  "locale": "en-US",
  "fields": {
    "title": "My Blog Post",
    "body": "# Hello\n\nThis is **markdown**.",
    "heroImage": "asset-123",
    "relatedPosts": ["post-1", "post-2"]
  },
  "publish": false
}
```

**Features:**

* Flat field structure (no locale nesting)
* Simple string IDs for references
* Markdown for rich text fields
* Automatic locale wrapping

**Raw JSON format**

The raw format gives you full control over locale-specific content:

```json theme={null}
{
  "space": "your-space-id",
  "environment": "master",
  "content_type_id": "blogPost",
  "fields": {
    "title": {
      "en-US": "My Blog Post",
      "fr-FR": "Mon Article de Blog"
    },
    "body": {
      "en-US": "# Hello\n\nEnglish content.",
      "fr-FR": "# Bonjour\n\nContenu français."
    }
  },
  "publish": false
}
```

**Use this format when:**

* Writing content in multiple locales simultaneously
* You need precise control over locale-specific values
* Migrating content from another system

***

## Common Use Cases

* **Content Publishing**: Generate and publish blog posts, articles, or product descriptions
* **Content Updates**: Bulk update existing entries with new information
* **Content Migration**: Move content between environments or spaces
* **Automated Workflows**: Create entries based on external triggers or data sources

***

## Additional Resources

Learn more about Contentful's content model and API:

* [Contentful Content Model Concepts](https://www.contentful.com/developers/docs/concepts/data-model/) - Understanding spaces, environments, and content types
* [Contentful Localization](https://www.contentful.com/developers/docs/concepts/locales/) - How locales work in Contentful
* [Rich Text Field Type](https://www.contentful.com/developers/docs/concepts/rich-text/) - Understanding Contentful's rich text format
* [Content Management API](https://www.contentful.com/developers/docs/references/content-management-api/) - Full API reference


# Connect Drupal to Profound
Source: https://docs.tryprofound.com/integrations/drupal/connect-drupal-to-profound



This guide walks you through connecting your Drupal account to Profound. Once connected, Profound [Workflows](/agents/workflows/about-workflows) can access, edit, and manage your Drupal content directly.

## Before you start

You'll need:

* [Drupal](https://www.drupal.org/) account with an administrator role, or any other role that allows you to administer users and permissions (learn more in Drupal's [Roles and Permissions documentation](https://www.drupal.org/docs/roles-and-permissions))
* A public Drupal HTTPS website
* [JSON:API](https://www.drupal.org/docs/core-modules-and-themes/core-modules/jsonapi-module) Drupal core module: it powers every action Profound performs
* [JSON:API Permission Access](https://www.drupal.org/project/jsonapi_permission_access) module to enable API authentication access for select user roles
* [Key auth](https://www.drupal.org/project/key_auth) module to enable API key authentication that allows Profound to take actions on your behalf
* [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Allow write operations for your JSON:API module**

* In Drupal, navigate to **Configuration > Web Services > JSON:API**.
* Enable **Accept all JSON:API create, read, update, and delete operations**.
* Select **Save configuration**.

  ![Drupal JSON:API module configuration view screenshot with the write mode option and save configuration button highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-write-mode.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=7ff235b4470145a47c434d968e5c8986)

**2. Create a Profound user in Drupal**

* In the Drupal platform, navigate to **People** in the left sidebar and select **+ Add user**.

  ![Drupal platform screenshot with People tab and an Add user button highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-people.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=d633700a378e755d4f5d9f62f89450a7)
* Name the user `profound`, give it a **Content editor** role or any other role with content management permissions, and select **Create new account**.

  ![Drupal Add user form screenshot with the Username and Role fields, and a Create new account button highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-add-user.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=cfe1b4f15641c475dff1dca4a7606207)
* Once the user is created, it appears on the user list.

  ![Drupal People view screenshot with the Profound user visible and highlighted in the users list](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-user-created.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=0929a896683f768b3cbf1a312818ccad)

**3. Enable necessary permissions for the Profound user**

To allow Profound to authenticate with Drupal using an API key, enable **Access JSON:API** and **Use key authentication** permissions for the role you assigned to your Profound user.

* In Drupal, navigate to **People > Permissions**.
* Search for the relevant permission and enable it for the role your Profound user has.
* Select **Save permissions**.

> [!NOTE]
>
> A standard Drupal content editor role has sufficient content permissions for Profound to perform the necessary actions. If you want to review or customize content permissions for Profound, you can do so anytime in the **Permissions** view in Drupal.

**4. Generate an API key**

* Navigate to **People** and select `profound` user you created in [Step 2](#step-2).
* Append `/key-auth` to the user page URL. For example:
  ```text theme={null}
    https://your-site.com/user/5/key-auth
  ```
* On the key authentication page for the user, select **Generate new key**.

  ![Drupal key authentication page screenshot with the Generate new key button highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-generate-key.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=7ad2783f940381b342b2a58349d2e2eb)
* Save the API key in a secure place, you'll need it to set up the integration in Profound.

**5. Connect Drupal to Profound**

Now that you have everything you need, create a Drupal integration in the Profound platform:

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Drupal** and select it.

  ![Profound Integrations view screenshot with "Drupal" typed in the search bar and a Drupal integration card both highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-integration.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=a66f360f62cd07c292c16ebbf8cdad2a)
* In the Drupal integration view, select **+ Add account**.
* In the **Connect Drupal** dialog, enter the API key you generated in [Step 4](#step-4) and your Drupal website's base URL.
* Select **Connect Account**.

  ![Connect Drupal dialog screenshot with the Connect account button highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-connect.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=715a8bf7577aca43ea9b296fe0d85e4f)

**6. Verify the connection**

If your Drupal account appears in the list of connected accounts, the integration was successful and is ready to use with Profound.

Revoke Profound's access to your Drupal account anytime by selecting **Revoke** in the integration account action menu.

![Profound account list item from the list of connected accounts with the three-dot action menu and the Revoke option highlighted](https://mintcdn.com/profound-37face47/Lvba4O-pjtEK-wzg/images/integrations/drupal/drupal-revoke.png?fit=max\&auto=format\&n=Lvba4O-pjtEK-wzg\&q=85\&s=37be72150c73b57b402b470d8b58404c)

> [!NOTE]
>
> You can also connect your Drupal account to Profound directly in the Workflow Builder view. [Create a Workflow](/agents/workflows/create-workflow), add one of the Drupal nodes to the canvas, select it to open its configuration panel, and follow the **Connect account** flow.

## Best practices

* Make sure your Drupal website is public and uses HTTPS. Profound rejects non-HTTPS websites.
* To customize permissions specifically for the Profound Drupal user, consider creating a custom role for it. Learn more in the Drupal [Roles and Permissions](https://www.drupal.org/docs/roles-and-permissions#s-create-a-custom-role) documentation.
* When reviewing Profound user permissions, grant it permissions only for the content you intend Profound to use.
* Always store your API keys in a secure place, such as secure notes in your password manager.

## Troubleshooting

### There is no JSON:API permission in my Drupal Permissions view

Check that the JSON:API Permission Access module is installed in your Drupal platform:

* Navigate to **Extend** in the left sidebar.
* If the module is installed, it appears on the module list with a selected checkbox.
* If it's not on the list, install it via the **Browse modules** tab.

### I see "We couldn't save these credentials" error in Profound

When connecting your Drupal account to Profound, you may see this error:

```txt wrap theme={null}
We couldn't save these credentials

Drupal JSON:API is reachable without a valid API key. Configure JSON:API to require key authentication.
```

It likely means that you're missing either the [JSON:API Permission Access](https://www.drupal.org/project/jsonapi_permission_access) module in Drupal or the **Access JSON:API** permission for your Profound user.

If that's the case, install the module and enable the permission as instructed in [Step 3](#step-3).


# Using Drupal in Workflows
Source: https://docs.tryprofound.com/integrations/drupal/using-drupal-in-workflows



After connecting your Drupal account, the following nodes become available in Profound Workflows.

The nodes work with Drupal's [JSON:API](https://www.drupal.org/docs/core-modules-and-themes/core-modules/jsonapi-module) module. Content is accessed by a **resource type** (the entity type, such as Basic, Image, or Article) and by a **resource UUID**.

## Available nodes

**Create Item**

### Create Item

Create a new item of a resource type.

#### Required inputs

* **Drupal Site** — Select a site from the dropdown
* **Resource type** — The JSON:API resource type, such as `node--article`
* **Attributes** or **Relationships** (or both) — A JSON object of field values or references to other resources.

> [!WARNING]
>
> A request with neither Attributes nor Relationships is rejected; include at least one.

**Update Item**

### Update Item

Update an existing item.

#### Required inputs

* **Drupal Site** — Select a site from the dropdown
* **Resource type** — The JSON:API resource type, such as `node--article`
* **Resource UUID**
* **Attributes** or **Relationships** (or both) — A JSON object of field values or references to update. At least one is required.

> [!NOTE]
>
> Only the attributes and relationships you provide are modified.

**Get Item**

### Get Item

Retrieve a single item.

#### Required inputs

* **Drupal Site** — Select a site from the dropdown
* **Resource type** — The JSON:API resource type, such as `node--article`
* **Resource UUID**

The output includes the item's full content with attributes, relationships, and metadata from the JSON:API response.

**List Items**

### List Items

List items of a given resource type.

#### Required inputs

* **Drupal Site** — Select a site from the dropdown
* **Resource type** — The JSON:API resource type, such as `node--article`

#### Optional inputs: Query options

* **Filter** — A JSON object using Drupal JSON:API filter syntax
* **Fields** — The fields to return per object in the node output; if nothing is specified, all fields are returned
* **Include** — A list of the related entities to include in the output (learn more in the [JSON:API documentation](https://www.drupal.org/docs/core-modules-and-themes/core-modules/jsonapi-module/includes))
* **Sort** — Sort expression for the result set
* **Limit** — Number of items to return (default 25, up to 50)
* **Offset** — Pagination offset (default 0)

The output is a structured list of items that can be used in downstream workflow steps.

***

## Key concepts

### Resource types

Resource types follow Drupal's JSON:API naming convention: the entity type and bundle joined by two hyphens.

| Resource type | Description |
| - | - |
| `node--article` | Article content type nodes |
| `node--page` | Basic page nodes |
| `taxonomy_term--tags` | Tags taxonomy terms |

> [!TIP]
>
> Use **List Items** to discover available resource types and inspect the structure of items returned for your Drupal site.

### Resource UUIDs

Items are referenced by **UUID**. When you create or list items, use the UUID from the JSON:API response as the **Resource UUID** in later steps.

### Attributes and relationships

**Create Item** and **Update Item** require at least one of **Attributes** or **Relationships**.

* **Attributes** — Field values for the item, such as title, body, or status. Example attributes for a `node--article`:
  ```json theme={null}
  {
    "title": "My new article",
    "body": {
      "value": "<p>Article content here.</p>",
      "format": "basic_html"
    },
    "status": true
  }
  ```
* **Relationships** — References to other resources, such as taxonomy terms, authors, or related nodes.

  Example relationships linking tags to an article:

  ```json theme={null}
  {
    "field_tags": {
      "data": [
        { "type": "taxonomy_term--tags",
          "id": "9b2c3d4e-5f6a-7b8c-9d0e-1f2a3b4c5d6e"
        }
      ]
    }
  }
  ```

***

## Additional resources

* [Drupal's JSON:API module documentation](https://www.drupal.org/docs/core-modules-and-themes/core-modules/jsonapi-module)
* [JSON:API specification](https://jsonapi.org/format/)


# Connect Framer to Profound
Source: https://docs.tryprofound.com/integrations/framer/connect-framer-to-profound



## Prerequisites

Before connecting Framer to Profound, ensure:

* You have access to the Framer project you want Profound to manage
* You have generated a Framer API key for that project
* You have the Framer project URL for the project you want to connect

> **How to create a Framer API key**
>
> 1. Open the Framer project you want to connect.
> 2. Go to **Site Settings → General**.
> 3. Create a new API key for the project.
> 4. Copy the API key and store it securely.

> **How to find your Framer project URL**
>
> 1. Open the Framer project you want to connect.
> 2. While viewing the Framer editor/project page, copy the project URL from your browser's address bar.
>
> Use a project URL like:
>
> ```text theme={null}
> https://framer.com/projects/<project-id>
> ```
>
> Do **not** use your deployed or published site URL, for example:
>
> ```text theme={null}
> https://your-site-name.framer.website
> ```

## Connect Your Account

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **Framer**, select it, then select **+ Add account**.
4. In the **Connect Framer** dialog, enter your Framer **Project API Key**.
5. Enter your Framer **Project URL**.
6. Select **Connect account**.

Once connected, Framer will appear as an available integration and can be selected inside Profound Workflows.

> [!NOTE]
>
> Framer API keys are bound to a single project. If you want to use multiple Framer sites in Profound, connect each project separately.


# Using Framer in Workflows
Source: https://docs.tryprofound.com/integrations/framer/using-framer-in-workflows



This page covers the **Framer CMS** integration in Profound.

After connecting your Framer project, the following actions become available as Workflow steps for working with Framer CMS collections and items. Each step requires you to select a connected **Framer Project** from a dropdown.

**Create item**

#### **Create Item**

Create a new CMS item in a Framer collection.

**Required inputs**

* **Framer Project** — Select the Framer project you connected in Profound from the dropdown
* **Collection** — Select the collection from the dropdown
* **Field Data** — The available fields are generated from the selected collection's schema. See [Working with Create Item Inputs](#working-with-create-item-inputs).

The output includes the created item's `id`, `slug`, `draft`, and [`field_data`](#understanding-item-output).

**Tip:** Use **Get Collection** first to inspect the collection schema and confirm which fields are required before creating an item.

**List collections**

#### **List Collections**

Retrieve the CMS collections available in your connected Framer project.

**Required inputs**

* **Framer Project** — Select the Framer project you connected in Profound from the dropdown

The output includes a structured list of collections with each collection's `id`, `name`, and `readonly` values.

**Get collection**

#### **Get Collection**

Retrieve a single Framer collection and its schema.

**Required inputs**

* **Framer Project** — Select the Framer project you connected in Profound from the dropdown
* **Collection** — Select the collection from the dropdown

The output includes the collection's `id`, `name`, `readonly`, `managed_by`, and [`field_definitions`](#understanding-collection-schemas), which describe each field's `id`, `name`, `type`, and whether it is required.

**List items**

#### **List Items**

Retrieve CMS items from a Framer collection.

**Required inputs**

* **Framer Project** — Select the Framer project you connected in Profound from the dropdown
* **Collection** — Select the collection from the dropdown

The output includes each item's `id`, `slug`, `draft`, and [`field_data`](#understanding-item-output).

**Get item**

#### **Get Item**

Retrieve a single CMS item by ID.

**Required inputs**

* **Framer Project** — Select the Framer project you connected in Profound from the dropdown
* **Collection** — Select the collection from the dropdown
* **Item ID** — Enter the item ID directly

The output includes the item's `id`, `slug`, `draft`, and [`field_data`](#understanding-item-output) so it can be inspected or passed to later steps.

***

## Supported Field Types

Framer's dynamic schema means Profound adapts to your collection model. The following Framer CMS field types are currently supported when creating items:

| Field Type | Description | Supported |
| - | - | - |
| **Plain Text** | Single-line text | Yes |
| **Formatted Text** | Rich text content | Yes |
| **Date** | Date field | No |
| **Link** | URL field | Yes |
| **Image** | Image asset | No |
| **Gallery** | Array of images | No |
| **Color** | Color value | No |
| **Toggle** | True/false value | Yes |
| **Number** | Numeric value | Yes |
| **Option** | Single-select option | Yes |
| **File** | File asset | No |
| **Reference** | Reference to another item | No |
| **Multi-Reference** | Reference to multiple items | No |

***

## Understanding Collection Schemas

Use **Get Collection** to inspect a collection's schema before working with its items.

The output includes:

* **Collection ID** and **Collection Name**
* **readonly** — whether the collection is read-only
* **managed\_by** — how the collection is managed
* **field\_definitions** — the fields configured on that collection

Each field definition includes:

* **id** — the field's internal Framer ID
* **name** — the label shown in Framer
* **type** — the field's primary Framer type, such as `string`, `formattedText`, or `number`
* **required** — whether the field is required
* **itemType** — additional item information for compound field types when Framer provides it

Example collection output:

```json theme={null}
{
  "collection": {
    "field_definitions": [
      {
        "id": "slug",
        "name": "Slug",
        "required": true,
        "type": "string",
        "itemType": null
      },
      {
        "id": "111111",
        "name": "Title",
        "required": true,
        "type": "string",
        "itemType": null
      },
      {
        "id": "222222",
        "name": "Content",
        "required": true,
        "type": "formattedText",
        "itemType": null
      }
    ],
    "id": "333333",
    "managed_by": "user",
    "name": "Example Collection",
    "readonly": false
  }
}
```

***

## Working with Create Item Inputs

When you select a Framer collection in **Create Item**, Profound renders inputs dynamically from that collection's schema.

For supported field types, you can fill in the values directly in the step UI. For example, a collection may expose fields such as:

* **Slug**
* **Title**
* **Content**
* **Link**
* **Name**
* **Option**
* **Count**
* **Featured**

If a collection includes required fields with unsupported types, the **Create Item** request will fail because Profound cannot provide values for those fields.

**Note:** Use **Get Collection** to confirm the field definitions before creating an item. If you need support for a Framer field type that is not currently supported, contact us and we can help evaluate your use case.

***

## Understanding Item Output

Framer item responses include top-level item fields such as:

* **id**
* **slug**
* **draft**
* **field\_data**

The `field_data` object is keyed by field ID. Each entry includes the field's human-readable name, type, and value.

Example item output:

```json theme={null}
{
  "item": {
    "draft": true,
    "field_data": {
      "111111": {
        "name": "Title",
        "type": "string",
        "value": "Example title",
        "valueByLocale": {}
      },
      "222222": {
        "name": "Featured",
        "type": "boolean",
        "value": true
      }
    },
    "id": "333333",
    "slug": "example-title"
  }
}
```

***

## Common Use Cases

* **Content Publishing**: Create new Framer CMS items from AI-generated copy or structured data
* **Content Retrieval**: Pull structured content from Framer collections for review, summarization, or downstream workflows
* **Workflow Preparation**: Inspect collection schemas before building automations that depend on specific Framer fields


# Connect G2 to Profound
Source: https://docs.tryprofound.com/integrations/g2/connect-g2-to-profound



This guide walks you through connecting your G2 account to Profound. Once connected, Profound [Workflows](/agents/workflows/about-workflows), Agents and Aim can access, edit, and manage your G2 data zdirectly.

## Before you start

You'll need:

* [G2](https://www.g2.com/) vendor account
* [Profound](https://www.tryprofound.com/) account

## Setup guide

To locate the G2 integration in the Profound platform:

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **G2** and select it.

  ![Profound Integrations view screenshot with "G2" typed in the search bar and a G2 integration card both highlighted](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-integration.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=51171af16800ae2c042181a9fc3e026a)

Depending on your needs, set up a G2 integration, a G2 connector, or both.

> [!NOTE]
>
> This is a one-time setup. After you connect your G2 account, it remains available for future use until you [revoke access](#revoking-access).

### G2 integration

The G2 integration allows Profound Workflows and Sheets to work with your G2 data.

**1. Connect your G2 account**

* In the G2 integration view, navigate to the **Integration** tab.
* Select **+ Add account**.

  ![G2 Integration view screenshot with the Integration tab and the Add account button highlighted](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-integration-add-account.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=69d47958c4c2775ac71e8b80674c065f)

**2. Follow the authorization dialog**

Follow the pop-up dialog that prompts you to sign in to your G2 account and authorize Profound to access your account.

![G2 authentication dialog screenshot with the Sign Up button highlighted](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-integration-auth.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=e6643a4a1775440486812dc7ea9c2476)

**3. Verify the connection**

If your G2 account appears in the list of connected accounts, the connection was successful and is ready to use with Profound.

> [!TIP]
>
> You can also connect your G2 account to Profound directly in the Workflow Builder view. [Create a Workflow](/agents/workflows/create-workflow), add one of the G2 nodes to the Workflow Builder canvas, select it to open its configuration panel, and follow the **Connect account** flow.

### G2 connector

The G2 connector lets Aim and Agents access your G2 data and use it in conversations live.

**1. Connect your G2 account**

* In the G2 integration view, navigate to the **Connector** tab.
* Select **+ Add account**.

  ![G2 Integration view screenshot with the Connector tab and the Add account button highlighted](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-connector-add-account.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=591c6a21f76a0acfba71829e55842bd6)

**2. Follow the authorization dialog**

Follow the pop-up dialog that prompts you to sign in to your G2 account and share authentication with Profound.

**3. Verify the connection**

If your G2 account appears in the list of connected accounts, the connection was successful and is ready to use with Profound.

![G2 account on list of connected accounts screenshot](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-connector-success.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=d3af6089988810bec10cf6681bba66d1)

**4. Review the permissions**

Review the permissions you want to give Profound Aim for the tools available with the G2 connector in the **Tools & permissions** table.

Each tool has three options:

* **Allow**: Aim can use the tool anytime
* **Ask**: Aim will ask you before using the tool
* **Block**: Aim can't use the tool

  ![G2 connector Tools & permissions table screenshot with permission options highlighted](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-connector-tools-permissions.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=4aad3598d9f87bb599225df4a1a46fff)

> [!TIP]
>
> The tools that can cause irreversible deletions or edits are listed in the **Destructive tools** section. Block these tools by default or set them to **Ask**.

## Revoking access

Revoke Profound's access to your G2 account anytime by selecting Revoke in the integration account action menu.

![G2 account on list of connected accounts screenshot with the three-dot action menu and the Revoke option highlighted](https://mintcdn.com/profound-37face47/I70sdA07a9xh3pui/images/integrations/g2/g2-revoke.png?fit=max\&auto=format\&n=I70sdA07a9xh3pui\&q=85\&s=561c81ddf6045a08309bd3b70402c316)

Revoke access separately for the integration and the connector.


# G2 integration overview
Source: https://docs.tryprofound.com/integrations/g2/overview

Combine G2 buyer intent, reviews, and ratings with your Profound data in one Workflow

Use G2 nodes to combine G2's buyer intent, reviews, ratings, and review snippets with your Profound data in one Workflow: spot accounts actively in market, track category-level demand, and turn customer language into reports or content, all without leaving Profound.

## Why the G2 integration exists

Buyer intent, reviews, and AI visibility usually live in separate tools. You see who is researching your category on G2, and separately, how your brand shows up in AI search. Combining the two means exporting data from one dashboard and cross-referencing it in another by hand, or waiting on a report that's already stale by the time it's built.

G2 Workflow nodes let you build repeatable workflows that read G2 and Profound data together, for cases where you want the same report or list on a schedule.

## Using G2 nodes in Profound Workflows

> [!TIP]
>
> Follow Profound’s [Connect G2 to Profound](/integrations/g2/connect-g2-to-profound) to set up your G2 integration and make G2 nodes available to Workflows.

Workflow nodes available for G2 include:

* **Browse Buyer Intent** — lists G2 accounts showing active buying intent for one or more products
* **Get Market Signals** — returns market intent activity for one or more G2 categories over a period of time
* **Get Reviews** — pulls G2 reviews for a product
* **Get Product Rating** — gets a product’s G2 rating scores

Learn more about the nodes and configuration in the [Using G2 in Workflows](/integrations/g2/using-g2-in-workflows) guide in Profound developer documentation.

## Example Workflow: G2 snapshot report

Get started with G2 Workflow nodes by building a Workflow that creates a snapshot report across a product's G2 metrics.

1. **G2 - Get Market Signals**: pull how much buyer interest a category is seeing right now, so you know whether demand is picking up or cooling off.
2. **G2 - Browse Buyer Intent**: surface the accounts actively researching the product, the ones worth a follow-up call today.
3. **G2 - Get Product Rating**: grab the product's current overall rating on G2, an instant read on reputation.
4. **G2 - Get Reviews**: pull a few quotable lines straight from reviewers, ready to drop into a deck or email.
5. [Prompt LLM](/agents/workflows/nodes/prompt-llm): turn all of it into one short summary: current rating, what reviewers keep bringing up, and any accounts worth flagging to sales.
6. [Profound Documents](/documents/about) - **Create Doc**: save that summary as a Profound document, ready to share with sales or product teams.

Schedule this Workflow to run weekly so G2 standing and demand data is ready in time for your weekly sales sync.

> [!NOTE]
>
> Try this Workflow in Profound: download the Workflow definition file below, import it into Profound (select **...** > **Import an Agent** in the Workflow Builder), connect your G2 account, and select **Run Test**.
>
> [g2-snapshot-report-agent.json](/files/g2-snapshot-report-agent.json)


# Using G2 in Workflows
Source: https://docs.tryprofound.com/integrations/g2/using-g2-in-workflows



After connecting your G2 account, the following nodes become available in Profound Workflows. G2 Workflow nodes bring buyer intent, market signals, reviews, ratings, and review excerpts into Workflows.

## Available nodes

**Browse Buyer Intent**

### Browse Buyer Intent

Surfaces G2 accounts showing active buying intent for one or more products.

#### Required inputs

* **G2 Account** — Select an account from the dropdown

#### Optional inputs

* **Products** — One or more products to track, entered as a product slug, product ID, or G2 product URL
* **Dimensions** — Which intent dimensions to include. **Company Name** and **Company Domain** are selected by default. Other options include Company Country, Employees, Industry, Signal Type, Provider, as well as other product, category, vendor, and visitor dimensions
* **Start date** / **End date** — The date range to pull intent data for (`YYYY-MM-DD`)
* **Company name** — Filter results to a specific company, for example `Acme`
* **Company domain** — Filter results to a specific company by domain, for example `example.com`
* **Signal type** — Narrow results to one kind of intent signal: Profile, Category, Compare, Competitors, Ad, Product Reference Page, Licensed Content, Vendors, Pricing, Competitive Profile, Competitive Compare, or Competitive Pricing
* **Minimum intent score** — Only return results with an intent score at or above this value (`0`–`100`)
* **Limit** — Maximum number of results to return. Defaults to `100` (max `100`)
* **Pagination cursor** — Pass a cursor from a previous response's `next_cursor` to fetch the next page of results

#### Output

The output is a JSON object with accounts showing buyer intent signals on G2, plus metadata that includes the next pagination cursor.

**Get Market Signals**

### Get Market Signals

Returns market intent activity for one or more G2 categories over a date range.

#### Required inputs

* **G2 Account** — Select an account from the dropdown
* **Categories** — One or more G2 categories to pull market signals for

#### Optional inputs

* **Start date** / **End date** — The date range to pull signals for (`YYYY-MM-DD`)
* **Limit** — Maximum number of results to return. Defaults to `100` (max `100`)
* **Pagination cursor** — Pass a cursor from a previous response's `next_cursor` to fetch the next page of results

#### Output

The output is a JSON object with market intent signals for the selected categories, plus metadata that includes the next pagination cursor.

**Get Reviews**

### Get Reviews

Pulls G2 reviews for a product.

#### Required inputs

* **G2 Account** — Select an account from the dropdown
* **Product** — A single product, entered as a product slug, product ID, or G2 product URL

#### Optional inputs

* **Submitted after** / **Submitted before** — Limit results to reviews submitted in a date range
* **Minimum star rating** — Only return reviews with a rating at or above this value (`1`–`5`)
* **Limit** — Maximum number of results to return. Defaults to `100` (max `100`)
* **Pagination cursor** — Pass a cursor from a previous response's `next_cursor` to fetch the next page of results

#### Output

The output is a JSON object with reviews for the product, plus metadata that includes the next pagination cursor.

**Get Product Rating**

### Get Product Rating

Returns a product's G2 rating scores.

#### Required inputs

* **G2 Account** — Select an account from the dropdown
* **Product** — A single product, entered as a product slug, product ID, or G2 product URL

#### Output

The output is a JSON object with G2 rating records for the product, including ease of use, ease of setup, direction, and other satisfaction sub-scores.

**Get Review Snippets**

### Get Review Snippets

Returns short, quotable excerpts from G2 reviews.

#### Required inputs

* **G2 Account** — Select an account from the dropdown
* **Product** — A single product, entered as a product slug, product ID, or G2 product URL

#### Optional inputs

* **NPS bucket** — Net Promoter Score category: `promoter`, `passive`, or `detractor`
* **Company segment** — For example, `Mid-Market`
* **Industry** — For example, `SaaS`
* **Categories** — Narrow snippets to one or more G2 categories
* **Role** — Narrow snippets to reviewers with a specific role, for example `CTO`
* **Tags** — Narrow snippets to one or more tags, for example `onboarding`

#### Output

The output is a JSON object with short review excerpts for the product.

> [!TIP]
>
> Snippets are separate from the full reviews feed. G2 may return zero snippets for a product that still has reviews. Use **Get Reviews** when you need the full review text.

***

## Key concepts

### Product identifiers

Product fields accept a **product slug**, **product ID**, or **G2 product URL**. Use the same identifier in every G2 node in a Workflow so results describe the same product throughout the run.

### Filters and optional fields

Optional fields act as filters. When left empty, the node returns unfiltered data.

### Pagination

Nodes that return lists default to a maximum of `100` results. To fetch more, pass the previous response's `next_cursor` into **Pagination cursor** on the next run or step.


# Connect Gamma to Profound
Source: https://docs.tryprofound.com/integrations/gamma/connect-gamma-to-profound



## Prerequisites

Before connecting Gamma to Profound, ensure:

* You have a [Gamma](https://gamma.app) account with API access
* You have generated an API key from Gamma

> **How to get a Gamma API key**
>
> 1. Visit the [Gamma developer documentation](https://developers.gamma.app/docs/get-access) for instructions on requesting API access
> 2. Once approved, generate an API key from your Gamma account settings
> 3. Copy and store the API key securely: you'll need it when connecting to Profound

## Connect Your Account

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **Gamma**, select it, then select **+ Add account**.
4. In the **Connect Gamma** dialog, enter your Gamma **API Key**.
5. Provide a **Workspace Name**, a friendly label to identify this Gamma connection.
6. Select **Connect account**.

Once connected, Gamma will appear as an available integration and can be selected inside Profound Workflows.


# Gamma integration overview
Source: https://docs.tryprofound.com/integrations/gamma/overview

Turn Profound Workflow's output into Gamma presentations, documents, and web pages

The [Gamma](https://gamma.app/products/presentations) node generates an AI-powered presentation using Gamma based on a structured prompt. Use this step when you want your Workflow to transform  outputs into a presentation-ready format, for example to generate client-ready visibility reports, turn AEO performance data into executive decks, or convert structured research into slide-based narratives.

## Connect Gamma

Connect your own Gamma account before using this node.

### Authorize Gamma Account

1. Add the Gamma node to your Workflow.
2. Open the node's configuration panel on the right side of the Workflow Builder.
3. Click **Add integration**.
4. Click **+ Connect account**.
5. Enter your Gamma API key and workspace name. Refer to [Gamma API key access instructions](https://help.gamma.app/en/articles/11962420-does-gamma-have-an-api#h_07981c54e0).
6. Once connected, select the account from the dropdown menu.

> [!TIP]
>
> You only need to connect an account once. After authorization, it remains available for future Workflows.

## Gamma Node Types

There are two Gamma node types:

* **Create new Gamma presentation** — Generates a brand-new presentation from scratch within Gamma. [Learn more](https://help.gamma.app/en/articles/7838093-how-do-i-create-a-new-presentation-document-or-webpage-in-gamma) in Gamma documentation.

* **Remix existing Gamma presentation** — Repurposes an existing Gamma presentation or template to create new content based on prior material. [Learn more](https://help.gamma.app/en/articles/12601672-what-is-remix-and-how-do-i-use-it) in Gamma documentation.

## Node configuration

Selecting the Gamma node opens its configuration panel on the right side of the Workflow Builder.

### Gamma integration (required)

Select the connected Gamma account where the presentation will be created.

### Template ID (required for Remix)

If using the Remix node, you must select the Gamma template to remix.

### Presentation prompt (required)

Enter the prompt to instruct Gamma's presentation generation. Write a static prompt, insert dynamic variables from earlier Workflow steps, or combine static instructions with dynamic output.

Provide clear structure in your prompt to produce predictable slide output.

Example:

```text expandable wrap theme={null}
Create a 10-slide executive summary presentation based on the following visibility report.

Audience: Marketing leadership
Goal: Highlight key changes in Answer Engine visibility and recommend next steps.

Data:
Visibility Score: /visibility_score
Share of Voice: /share_of_voice
Top Performing Topics: /top_topics

Include a title slide, summary slide, data breakdown slides, and a recommendations slide.
```

## Advanced settings

### Text mode

Controls how Profound processes the input text before submitting to Gamma, refining and adding focus to the prompt. Options include:

* **Generate** – Create new content based on the input.
* **Condense** – Summarize or shorten the input text.
* **Preserve** – Keep the original wording and structure.

### Format

Defines the output format of the final result: Presentation, Document, Webpage, or Social.

### Theme ID

The ID of an existing Gamma theme to apply to the output. [Learn more about Gamma Themes](https://help.gamma.app/en/articles/11962420-does-gamma-have-an-api#h_07981c54e0).

### Number of cards

Specifies the number of slides (cards) in the final presentation. [Learn more about Cards in Gamma](https://help.gamma.app/en/articles/11016396-what-are-cards-in-gamma-and-how-to-do-they-work).

### Image source

Determines the source of images included in the presentation: AI Generated, Pictographic, Unsplash, Giphy, Web (All), Web (Free), Web (Commercial), Placeholder, or No Images. [Learn more about Gamma Image generation](https://help.gamma.app/en/articles/11856101-how-do-i-use-the-visuals-menu-in-gamma#h_d7e25a630f).

### Text amount

Controls the volume of text per slide: Brief, Medium, Detailed, or Extensive.

### Tone

Sets the tone of the generated content (for example, **Professional** for a more formal style).

### Audience

Defines the intended audience (for example, **Marketers**).

### Language

Specifies the language of the output content.

## Output

The Gamma node returns two URLs:

* **Export URL** — A link to download the generated presentation. Only present if you select an export format in Sharing Settings.
* **Presentation URL** — A link to the presentation hosted in your Gamma workspace.


# Using Gamma in Workflows
Source: https://docs.tryprofound.com/integrations/gamma/using-gamma-in-workflows



After connecting your Gamma account, the following actions become available as Workflow steps. Each step requires you to select a connected **Gamma Account** from a dropdown.

**Generate content**

#### **Generate Content**

Create a new AI-powered presentation, document, or webpage from scratch.

**Required inputs**

* **Gamma Account**
* **Input Text** (the content or topic to generate from)

**Optional inputs**

* **Format** — `presentation`, `document`, `webpage`, or `social`
* **Text Mode** — how Gamma handles your input text:
  * `generate` (default) — AI generates content based on your input
  * `condense` — AI condenses your input into fewer cards
  * `preserve` — AI preserves your input text as-is
* **Number of Cards** — how many slides/pages to generate
* **Additional Instructions** — extra guidance for the AI
* **Theme ID** — a Gamma theme to apply
* **Export As** — optionally export as `pdf` or `pptx`

**Text options**

* **Text Amount** — level of detail (e.g., `detailed`, `concise`)
* **Text Tone** — comma-separated tone descriptors (e.g., `friendly, professional`)
* **Text Audience** — target audience (e.g., `executives`, `students`)
* **Text Language** — output language (e.g., `en`, `es`, `fr`)

**Image options**

* **Image Source** — where to source images from: `ai_generated`, `unsplash`, `giphy`, `web_free_to_use`, `web_free_to_use_commercially`, `pictographic`, `placeholder`, or `no_images`

**Sharing options**

* **Workspace Access** — access level for workspace members:
  * `workspace_default` — uses your workspace's default sharing settings
  * `view` — members can view the content
  * `comment` — members can view and leave comments
  * `edit` — members can view, comment, and edit the content
  * `full_access` — members have full control, including sharing and deletion
* **External Access** — access level for people outside the workspace (defaults to `no_access`):
  * `no_access` — external users cannot access the content
  * `view` — external users can view the content via a shared link
  * `comment` — external users can view and leave comments
  * `edit` — external users can view, comment, and edit the content

**Tip:** Generation is asynchronous: Profound automatically polls for completion before returning the result to subsequent Workflow steps.

**Generate from template**

#### **Generate from Template**

Remix an existing Gamma template with new content.

**Required inputs**

* **Gamma Account**
* **Template ID** (the Gamma ID of the template to remix)
* **Prompt** (instructions and content to adapt the template with)

**Optional inputs**

* **Theme ID** — override the template's theme
* **Export As** — optionally export as `pdf` or `pptx`
* **Workspace Access** — access level for workspace members (see sharing options in Generate Content above)
* **External Access** — access level for external users (see sharing options in Generate Content above, defaults to `no_access`)

**Tip:** Templates are great for maintaining consistent branding. Create a template in Gamma, then use this action to generate variations with different content.

**Get generation status**

#### **Get Generation Status**

Check the status of a content generation.

**Required inputs**

* **Gamma Account**
* **Generation ID**

The output includes:

* **Status** — `pending`, `completed`, or `failed`
* **Gamma URL** — link to the generated content (available when completed)
* **Export URL** — link to the exported file if PDF/PPTX export was requested

**Note:** You typically don't need to use this step directly: Profound handles polling automatically when you use the Generate actions.

***

## Output Formats

Gamma can generate content in several formats:

| Format | Description |
| - | - |
| **Presentation** | Slide-based format, similar to PowerPoint or Google Slides |
| **Document** | Long-form scrollable content |
| **Webpage** | Standalone shareable webpage |
| **Social** | Social media-optimized content |

You can also export the result as **PDF** or **PPTX** for offline use.

***

## Common Use Cases

* **Sales Decks**: Generate personalized presentations from CRM data or call transcripts
* **Reports**: Turn data and analysis into polished documents or presentations
* **Training Materials**: Create training decks from knowledge base content
* **Client Deliverables**: Generate branded documents from templates with client-specific content
* **Social Content**: Create social media visuals from blog posts or product updates


# Connect Gong to Profound
Source: https://docs.tryprofound.com/integrations/gong/connect-gong-to-profound



## Prerequisites

Before connecting Gong to Profound, ensure:

* You have **Technical Administrator** permissions in Gong
* Your Gong account is active and in good standing

> **How to check if you're a Technical Administrator**
>
> 1. Log in to Gong and click your name in the top right corner
> 2. Select **My Settings**
> 3. Scroll to the bottom of the page to see the list of Technical Administrators
> 4. Alternatively, go to **Company Settings → Team Members** and check your profile
>
> If you're not listed as a Technical Administrator, contact someone who is to complete the integration setup.

## Connect Your Account

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **Gong**, select it, then select **+ Add account**.
4. You'll be redirected to Gong to authorize Profound.
5. Sign in with your **Technical Administrator** account and grant the requested permissions.
6. Once authorized, you'll be redirected back to Profound.

Once connected, Gong will appear as an available integration and can be selected inside Workflows.


# Using Gong in Workflows
Source: https://docs.tryprofound.com/integrations/gong/using-gong-in-workflows



After connecting your Gong account, the following actions become available as Workflow steps. Each step requires you to select a connected **Gong Account** from a dropdown.

**Get call**

#### **Get Call**

Retrieve a single call by ID with full details.

**Required inputs**

* **Gong Account**
* **Call ID**

**Optional inputs**

* **Include Transcript** (include full transcript structure)

The output includes comprehensive call data: metadata, participants, topics discussed, trackers detected, and optionally the full transcript.

**Get transcript**

#### **Get Transcript**

Retrieve the full transcript for a specific call.

**Required inputs**

* **Gong Account**
* **Call ID**

The output includes the complete transcript with:

* Speaker identification
* Timestamps for each segment
* Topic segmentation
* Full text of the conversation

**Tip:** Transcripts can be large for long calls. Use this action when you need the full conversation text for analysis or content generation.

***

## What Data is Available

Gong provides rich AI-analyzed conversation data:

| Data Type | Description |
| - | - |
| **Call Metadata** | Title, date, duration, participants, direction (inbound/outbound) |
| **Transcript** | Full text with speaker identification and timestamps |
| **Topics** | AI-detected topics discussed during the call |
| **Trackers** | Custom keyword and phrase detection (competitors, pricing, etc.) |
| **Talk Ratios** | Speaking time distribution between participants |
| **Questions** | Questions asked during the conversation |

***

## Common Use Cases

* **Content Generation**: Use call transcripts to generate blog posts, case studies, or training materials
* **Competitive Intelligence**: Monitor calls for competitor mentions and extract insights
* **Sales Coaching**: Analyze top-performing calls and create training content
* **CRM Updates**: Extract action items and next steps from calls to update your CRM
* **Knowledge Base**: Convert call transcripts into searchable documentation


# Ads node
Source: https://docs.tryprofound.com/integrations/google-ads/ads

Pull Responsive Search Ad creatives and performance from Google Ads into a Workflow

The **Ads node** pulls your Responsive Search Ad (RSA) creatives directly from Google Ads into a Workflow. It returns headlines, descriptions, final URLs, and performance metrics per ad.

Ads node is the creative layer of the Google Ads nodes. Where the other nodes cover which queries triggered your ads ([Search Terms](/integrations/google-ads/search-terms)), which keywords you bid on ([Keywords](/integrations/google-ads/keywords)) and so on, this node focuses on the ad creatives themselves and how they're performing.

Use this node when you need to audit ad copy at scale, identify creatives that need a rewrite, or check whether your messaging aligns with how AI engines currently describe your category.

## When to use this node

Use the Ads node for tasks such as:

* **RSA creative audits:** pull all live headlines and descriptions and evaluate them at scale
* **Copy iteration:** identify which creatives are underperforming and generate rewrite candidates
* **Brand voice scoring:** score ad copy against brand guidelines or tone-of-voice rules
* **AI search alignment:** check whether your ad messaging aligns with how AI engines currently describe your category
* **Weekly ad digests:** include ad performance alongside search term and keyword data in a single weekly summary for your team

## Node configuration

### Google Integration (required)

Select a previously connected Google account, or click **+ Add integration** and follow the prompts to connect one. This is a one-time setup. Once you add an integration, it's available in all future sessions until you remove it in Profound **Settings > Integrations**.

### Google Ads Account (required)

Select the Google Ads customer account from the dropdown. The picker handles both Manager accounts and sub-account selection.

### Campaigns

Select the ad campaigns from the dropdown. To include every active campaign, leave the field empty.

### Ad Groups

Select the ad groups from the dropdown. To include every ad group within selected campaigns, leave the field empty. To see the ad group options in the dropdown, select at least one campaign in the **Campaigns** field.

### Date Range (required)

Select the date range to pull ad performance over. Choose from available presets (today, yesterday, last 7 days, and others) or set a custom range.

### Output Fields

Select the fields to include in the output. The options include metrics and campaign details, such as Impressions, Clicks, Conversions, Campaign Type, and Bid Strategy Type. To return output unfiltered, leave this field empty.

### Output Label

Enter a descriptive label for this step's output, for example `ad_data` or `rsa_performance_last_7_days`. The label becomes the variable name you reference in later nodes.

## Output

The node outputs your ad data in JSON format, including:

* headlines, descriptions, and final URLs
* performance metrics: impressions, clicks, cost, and others

**Pass the output to:**

* [Prompt LLM](/agents/workflows/nodes/prompt-llm) for AI analysis: creative audits, brand voice scoring, copy rewrite suggestions, or AI search trend alignment checks
* [Slack](/integrations/slack/connect-slack-to-profound) to push an automated weekly ad creative digest to your team channel
* **Google Sheets (Write Row) node** to log ad performance data over time
* **Code node** or **[Conditional node](/agents/how-tos/add-conditionals-to-workflows)** to filter, sort, or slice the data (for example, keep only the top 10 creatives by CTR) before passing a clean dataset to the LLM

## Example usage: RSA creative audit

Set up a Workflow to score every live ad creative against brand guidelines and surface rewrite candidates automatically. Here is the example structure:

1. **[Ads node](/integrations/google-ads/ads)**

* Select your **Google Integration** and set **Google Ads Account** to your Google Ads account
* Set **Date Range** to Last 7 days
* Set **Campaigns**, **Ad Groups** and **Output Fields** filters as needed
* Set **Output label** to `ad_data`

2. **Prompt Responses node**

* Set **Date range** to Last 7 days
* Select relevant topics from the **Topics** dropdown
* After selecting a topic, relevant prompts are selected automatically; edit the selection in the **Prompts** dropdown
* Set **Output label** to `ai_prompt_responses`

> [!TIP]
>
> The data from both Ads and Prompt Responses nodes can be too large to send directly to the Prompt LLM node. Use AI Marketer to build the preprocessing step into your Workflow.

**3.** [Prompt LLM](/agents/workflows/nodes/prompt-llm) **node**

* Enter the prompt:

  ```text expandable wrap theme={null}
  You are a brand copywriter. Review these datasets:
  - RSA creatives: {{ad_data}}
  - AI prompt response data: {{ai_prompt_responses}}

  For each ad, score each headline and description 1–5 against:
  - Brand voice alignment (clear, confident, not generic)
  - Alignment with the language and themes AI engines are actually
  using to describe this category (from {{ai_prompt_responses}})

  Flag any headline or description scoring 1–2 as a rewrite candidate.

  For each flagged item, write one improved alternative that mirrors
  the language patterns winning in AI answers.
  ```
* Set **Output label** to `creative_audit`

> [!TIP]
>
> Check out our [prompt best practices](/agents/how-tos/prompt-llm-best-practices) guide for the Prompt LLM node.

**4.** [Slack](/integrations/slack/connect-slack-to-profound) **node**

* Select the Slack workspace and channel to receive updates
* Input the message text, or use the `creative_audit` variable in the **Message** field

## Best practices

### Scope by campaign and ad group

Pulling all ads data from all campaigns and ad groups at once can return an unmanageably large dataset. Use the built-in campaign and ad group filtering to scope results before passing them to subsequent nodes.

Use [List Campaigns](/integrations/google-ads/list-campaigns) node to discover the active campaign set first, map campaign names to IDs, or fan out the workflow across every campaign dynamically.

### Pair with other Google Ads nodes

Pair with [Search Terms](/integrations/google-ads/search-terms) to see what users searched, what ad they saw, and how that ad performed.

Pair with [Keywords](/integrations/google-ads/keywords) to see if your highest-bid keywords appear in your RSA headline copy.

### Combine with Profound data

Ads node is useful on its own, but it becomes most powerful when combined with Profound data nodes:

* Add a **Prompt Responses** node to score ad copy against real AI engine language
* Add a ​  [Citation Pages](/agents/workflows/nodes/data/citation-pages) node to check if your landing page (final URL) is already being cited by AI engines

### Process the output data

Ads node can return hundreds of rows for larger date ranges. Passing a large dataset directly to a [Prompt LLM](/agents/workflows/nodes/prompt-llm) node risks hitting token limits and introducing noise from low-volume, irrelevant terms.

A good way to pre-process the data is to use AI Marketer: describe what you want done with the data, and let the Assistant select and insert appropriate nodes into your Workflow.


# Keywords
Source: https://docs.tryprofound.com/integrations/google-ads/keywords

Return the keywords you bid on in Google Ads with quality and efficiency metrics

The **Keywords node** returns the keywords you're actively bidding on in Google Ads, paired with quality score, post-click experience signals, and derived efficiency metrics over a configurable date range.

## When to use this node

Use the Keywords node for tasks such as:

* **Bid management:** identify which keywords are spending efficiently and which are wasting budget
* **Quality score audits:** surface keywords with low quality scores dragging down overall ad performance
* **Identifying underperformers:** flag keywords with high cost but low CVR (conversion rate), CPA (cost per acquisition), or ROAS (return on ad spend)
* **Cross-channel visibility audits:** cross-reference keywords you're bidding on against Profound AI Visibility data to find where you're paying for clicks you already win organically

## Node configuration

### Google Integration (required)

Select a previously connected Google account, or click **+ Add integration** and follow the prompts to connect one. This is a one-time setup. Once you add an integration, it's available in all future sessions until you remove it in Profound **Settings > Integrations**.

### Google Ads Account (required)

Select the Google Ads customer account from the dropdown. The picker handles both Manager accounts and sub-account selection.

### Campaigns

Select the ad campaigns from the dropdown. To include every active campaign, leave the field empty.

### Ad Groups

Select the ad groups from the dropdown. To include every ad group within selected campaigns, leave the field empty. To see the ad group options in the dropdown, select at least one campaign in the **Campaigns** field.

### Date Range (required)

Select the date range to pull keyword performance over. Choose from available presets (today, yesterday, last 7 days, and others) or set a custom range.

### Output Fields

Select the fields to include in the output. The options include metrics and campaign details, such as Impressions, Clicks, Conversions, Campaign Type, and Bid Strategy Type. To return output unfiltered, leave this field empty.

### Output Label

Enter a descriptive label for this step's output, for example `keyword_data` or `keyword_performance_last_7_days`. The label becomes the variable name you reference in later nodes.

## Output

The node outputs JSON data containing your active keywords, along with:

* the keyword text

* keyword quality score

* post-click experience

* performance metrics: CTR, CPC, CPA, and others

**Pass the output to:**

* [Prompt LLM](/agents/workflows/nodes/prompt-llm) for AI analysis: quality score audit reports, bid management recommendations, underperformer identification
* **​Prompt Responses node** for the most powerful use: cross-reference keywords you're bidding on against AI prompt answer data to find queries where you're paying for traffic you already win organically, or topics with strong AI presence where you're not yet bidding
* **Google Sheets (Write Row) node** to build a keyword performance tracker and log results over time for trend analysis

## Example usage: Cross-channel keyword visibility audit

Set up a Workflow to find where you're paying for clicks you already win organically in AI answers, and save the report to a Profound Document. Here is the example structure:

1. [**Keywords node**](/integrations/google-ads/keywords)

* Select your **Google Integration** and set **Google Ads Account** to your Google Ads account
* Set **Date Range** to Last 7 days
* Set **Campaigns**, **Ad Groups,** and **Output Fields** filters as needed
* Set **Output label** to `bidded_keywords`

2. **[Visibility Score node](/agents/workflows/nodes/data/visibility-score)**

* Set the **Metric** to **Visibility Score**
* Set **Date range** to Last 4 weeks
* Click **+ Add filter > Topics**, and select the topics you want to measure your visibility by

> [!TIP]
>
> Use **Agent Assistant** to generate the list of keywords from the Keyword node output data and pass it on to Topics field as a variable.

* Add **Topics as a Dimension** option to make the output a separate visibility score per topic rather than a single aggregated value
* Set **Output label** to `ai_visibility_data`

3. **Code node**

* Use **Agent Assistant** to add custom logic to cross-reference `bidded_keywords` with `ai_visibility_data`, match paid keywords whose text contains an AI-visible asset name, sum the overlap spend, and rank the underperforming keywords by cost
* Set **Output label** to `cross_reference_result`

4. **[Prompt LLM node](/agents/workflows/nodes/prompt-llm)**

* Enter the prompt:

  ```text expandable wrap theme={null}
  Produce a cross-channel visibility audit report from {{cross_reference_result}}.

  Sections:
  - Executive Summary — overlap % of paid spend vs. organic AI 
  wins + headline recommendation.
  - Spend at Risk — table: Total Paid Spend | Overlap Spend | Overlap %.
  - Top Overlap Keywords — table of top 15: Keyword | Cost | Clicks | 
  Conversions | Matched Asset | AI Visibility. Follow with 3–5 callouts for strongest 
  cut candidates (high cost + high AI visibility + low conversions).
  - Recommended Actions — 4–6 prioritized bullets with specifics 
  (e.g. "Pause exact-match on [keyword] — savings: $X/period").
  ```
* Set **Output label** to `cross_channel_audit`

> [!TIP]
>
> Check out our [prompt best practices](/agents/how-tos/prompt-llm-best-practices) guide for the Prompt LLM node.

5. **[Profound Docs](/documents/about.mdx) - Create Doc node**

* Enter any document text you need in the **Initial Content** field, and use `cross_channel_audit` variable
* Enter **Document Name**, for example “2026-05-01 Cross-channel keyword audit”

## Best practices

### Scope by campaign and ad group

Pulling all keywords data from all campaigns and ad groups at once can return an unmanageably large dataset. Use the built-in campaign and ad group filtering to scope results before passing them to subsequent nodes.

Use [List Campaigns](/integrations/google-ads/list-campaigns) node to discover the active campaign set first, map campaign names to IDs, or fan out the workflow across every campaign dynamically.

### Pair with Profound data

Keyword Performance is useful on its own, but combined with Profound data nodes (such as **Prompt Responses**, [Visibility Score](/agents/workflows/nodes/data/visibility-score), and so on), it becomes a cross-channel strategy tool. There's no clear way to tell which of your paid search terms are also surfacing in AI answers without connecting both data sources.

### Use Agent Assistant

Working with both Keywords data and Profound AI visibility data can have a learning curve. Use **Agent Assistant** to build a working Workflow faster, and then edit the Workflow as needed to better match your use case.


# Landing Pages node
Source: https://docs.tryprofound.com/integrations/google-ads/landing-pages

Connect Google Ads performance to the pages users land on after clicking

The **Landing Pages node** connects paid search performance to the final URLs users land on after clicking your ads.

## When to use this node

Use the Landing Pages node for tasks such as:

* **Prioritize Profound optimization**: identify which landing pages are actively receiving Google Ads traffic, then prioritize Profound optimization work on those pages
* **Find visibility gaps**: surface high-spend or high-conversion landing pages with weak AI visibility, citation coverage, or answer-engine performance
* **Audit destination URLs**: check whether ad traffic is going to the right pages, especially if campaigns have destination URL sprawl
* **Spot content mismatches**: compare landing page messaging against the keywords and ad groups driving traffic to identify gaps
* **Feed Profound workflows**: send landing pages into downstream Workflow nodes for content refreshes, page analysis, or competitive visibility checks

## Node configuration

### Google Integration (required)

Select a previously connected Google account, or click **+ Add integration** and follow the prompts to connect one. This is a one-time setup. Once you add an integration, it's available in all future sessions until you remove it in Profound **Settings > Integrations**.

### Google Ads Account (required)

Select the Google Ads customer account from the dropdown. The picker handles both Manager accounts and sub-account selection.

### Campaigns

Select the ad campaigns from the dropdown. To include every active campaign, leave the field empty.

### Ad Groups

Select the ad groups from the dropdown. To include every ad group within selected campaigns, leave the field empty. To see the ad group options in the dropdown, select at least one campaign in the **Campaigns** field.

### Date Range (required)

Select the date range to pull search term performance over. Choose from available presets (today, yesterday, last 7 days, and others) or set a custom range.

### Output Fields

Select the fields to include in the output. The options include metrics and campaign details, such as Impressions, Clicks, Conversions, Campaign Type, and Bid Strategy Type. To return output unfiltered, leave this field empty.

### Output label

Enter a descriptive label for this step's output, for example `landing_pages_data_past_month`. The label becomes the variable name you reference in later nodes.

## Output

The node outputs landing page data in JSON format, including:

* Campaign and ad group attribution
* Performance metrics: clicks, conversions, and others
* Page-experience metrics: mobile-friendly clicks %, speed score, and valid AMP clicks %

**Pass the output to:**

* [Prompt LLM](/agents/workflows/nodes/prompt-llm) node to analyze underperforming pages and generate optimization recommendations
* [Web Page Scrape](/agents/workflows/nodes/web-page-scrape) node to scrape content from flagged pages for deeper LLM analysis
* [Slack](/integrations/slack/connect-slack-to-profound) to deliver the optimization report to your team

## Example usage: Weekly paid landing page health report

Set up a Workflow to surface landing pages with page-experience problems or poor conversion efficiency, and post a weekly digest to your paid search channel.

1. **[Landing Pages node](/integrations/google-ads/landing-pages)**

* Select your **Google Integration** and set **Google Ads Account** to your Google Ads account
* Set **Date Range** to Last 7 days
* Set **Campaigns**, **Ad Groups,** and **Output Fields** filters as needed
* Set **Output label** to `landing_page_data`

2. **[Prompt LLM node](/agents/workflows/nodes/prompt-llm)**

* Enter the prompt:

  ```text expandable wrap theme={null}
  You are a paid search analyst. Review this landing page data:
  {{landing_page_data}}:

  - Identify pages with speed score below 50, poor mobile-friendly 
  rate, or high spend but low conversions.
  - Summarize optimization priorities, ordered by estimated impact.

  Keep the report short and actionable.
  ```

* Set **Output label** to `landing_page_report`

3. **[Slack node](/integrations/slack/connect-slack-to-profound)**

* Select the Slack workspace and channel to receive updates
* Input the message text and use the `landing_page_report` variable in the **Message** field

Schedule the Workflow to run weekly.

## Best practices

### Scope by campaign and ad group

Pulling all landing page data from all campaigns and ad groups at once can return an unmanageably large dataset. Use the built-in campaign and ad group filtering to scope results before passing them to subsequent nodes.

Use [List Campaigns node](/integrations/google-ads/list-campaigns) to discover the active campaign set first, map campaign names to IDs, or fan out the workflow across every campaign dynamically.

### Pair with Web Page Scrape node

Scraping the actual content of flagged pages using [Web Page Scrape node](/agents/workflows/nodes/web-page-scrape) gives the LLM what it needs to generate specific, actionable recommendations rather than generic ones.


# List Campaigns node
Source: https://docs.tryprofound.com/integrations/google-ads/list-campaigns

Pull the active Search campaigns from a connected Google Ads account

The **List Campaigns node** pulls the active Search campaigns from a connected Google Ads customer account.

It's designed as an upstream lookup step: its job is to provide a campaign list that subsequent Google Ads nodes use to filter, label, and scope their data pulls before you run performance analysis.

## When to use this node

Use the List Campaigns node for tasks such as:

* **Scoping downstream nodes to specific campaigns:** get campaign IDs and names first, then pass them into [Search Terms](/integrations/google-ads/search-terms), [Keywords](/integrations/google-ads/keywords), [Landing Pages](/integrations/google-ads/landing-pages), or [Ads](/integrations/google-ads/ads) as a filter rather than pulling data across all campaigns at once
* **Labeling campaign-level performance data:** enrich downstream node outputs with campaign names so LLM-generated summaries are human-readable
* **Building campaign-aware Workflows at scale:** especially useful for Manager accounts (MCC) with many campaigns, where structured campaign discovery is needed before diving into metrics

## Node configuration

### Google Integration (required)

Select a previously connected Google account, or click **+ Add integration** and follow the prompts to connect one. This is a one-time setup. Once you add an integration, it's available in all future sessions until you remove it in Profound **Settings > Integrations**.

### Google Ads Account (required)

Select the Google Ads customer account from the dropdown. The picker handles both Manager accounts and sub-account selection.

### Output label

Enter a descriptive label for this step's output, for example `campaigns` or `active_campaigns`. The label becomes the variable name you reference in later nodes.

## Output

The node outputs your campaigns data in JSON format, including campaign names, dates, and statuses.

**Pass the output to:**

* [Search Terms](/integrations/google-ads/search-terms): Filter by campaign ID to get actual user queries for that campaign, with CTR, CPC, CVR, CPA, and ROAS
* [Keywords](/integrations/google-ads/keywords)**:** Scope the keyword quality score and efficiency audit to specific campaigns
* [Ads](/integrations/google-ads/ads):​ Pull only the RSA creative assets (headlines and descriptions) for a specific campaign
* [Prompt LLM](/agents/workflows/nodes/prompt-llm): Feed the campaign list for summarization, naming convention audits, or campaign structure analysis
* **Code node** or **[Conditional](/agents/how-tos/add-conditionals-to-workflows) node** to filter, sort, or slice campaigns data, or branch workflow logic based on factors such as status or presence of a campaign

## Example usage: Weekly negative keyword discovery

Set up a Workflow to scan last week's search terms across all campaigns, identify wasted spend, and post actionable negative keyword recommendations to Slack. Here is the example structure:

1. **[List Campaigns node](/integrations/google-ads/list-campaigns)**

* Select your **Google Integration** and set **Google Ads Account** to your Google Ads account
* Set **Output label** to `campaigns`

2. **[Search Terms node](/integrations/google-ads/search-terms)**

* Repeat the **Google Integration** and **Ads Account** from the previous step
* Set **Date Range** to Last 7 days
* Set **Campaigns**, **Ad Groups,** and **Output Fields** filters as needed
* Set **Output** label to `search_term_data`

> [!TIP]
>
> The data from both List Campaigns and Search Terms nodes can be too large to send directly to the Prompt LLM node. Use AI Marketer to build the data preprocessing steps, such as the Code node for trimming search term data, into your Workflow.

3. **[Prompt LLM node](/agents/workflows/nodes/prompt-llm)**

* Enter the prompt:

  ```text expandable wrap theme={null}
  You are a paid search analyst. Review last week's Google Ads search term
  data: {{search_term_data}}.

  Campaigns context: {{campaigns}}.

  Flag terms as negative keyword candidates if they have:
  - Zero conversions and cost over $100
  - CVR below 1% and cost over $50
  - Informational or off-target intent (e.g. "free", "what is", "certification")
  - Competitor branded terms we are not intentionally targeting

  For each flagged term: name, campaign, cost, conversions, reason to negate,
  recommended match type (Exact or Phrase).

  End with: total wasted spend, top 3 negatives to add today.

  Format as a scannable Slack message with bullet points.
  ```

* Set **Output label** to `negative_keyword_report`

> [!TIP]
>
> Check out our [prompt best practices](/agents/how-tos/prompt-llm-best-practices) guide for the Prompt LLM node.

**4.** [Slack](/integrations/slack/connect-slack-to-profound) **node**

* Select the Slack workspace and channel to receive updates
* Input the message text, or use the `negative_keyword_report` variable in the **Message** field

Schedule the Workflow to run weekly, for example every Monday at 9 AM.

## Best practices

### Use to support other Google Ads nodes

List Campaigns is a lookup step. Its full value comes through when its output flows into [Search Terms](/integrations/google-ads/search-terms), [Keywords](/integrations/google-ads/keywords), [Landing Pages](/integrations/google-ads/landing-pages) or [Ads](/integrations/google-ads/ads) as a scoping filter.

### Use Iterations for multi-campaign Workflows

To run downstream nodes for each campaign individually, wrap them in an  [Iteration block](/agents/how-tos/add-iterations-to-workflows) that loops over each campaign ID returned by List Campaigns.

### Test with a single campaign first

On accounts with many campaigns, scope to one campaign ID in your test run to validate the pipeline before scaling.


# About Google Ads integration
Source: https://docs.tryprofound.com/integrations/google-ads/overview

Connect your Google Ads account and pull live paid search data into Profound Workflows

The Google Ads integration connects your Google paid search account directly to Profound. The Workflow nodes (Search Terms, Keywords, Ads, Landing Pages, and List Campaigns) pull live Google Ads data into any Workflow.

## Why Google Ads integration exists

Before Google Ads integration, paid search teams had no way to get Google Ads data into the same workflow as Profound's AI visibility insights without exporting CSVs and rebuilding reports manually. The Google Ads integration fixes this: connect your account once and pull campaigns, search terms, keywords, or ad creatives directly into Workflow Builder, and combine them with Profound's visibility data for powerful cross-channel research.

## Google Ads Workflow nodes

* [List Campaigns](/integrations/google-ads/list-campaigns) — pulls active Search campaigns from your Google Ads account.
* [Search Terms](/integrations/google-ads/search-terms) **—** returns the actual queries users typed that triggered your ads, with click-through rate (CTR), cost per click (CPC), conversion rate (CVR), cost per acquisition (CPA), and return on ad spend (ROAS).
* [Keywords](/integrations/google-ads/keywords) **—** returns the keywords you're bidding on with quality score, post-click experience signals, and efficiency metrics.
* [Landing Pages](/integrations/google-ads/landing-pages) — returns performance and page-experience data for the final URLs users land on after clicking your ads.
* [Ads](/integrations/google-ads/ads) - returns your Responsive Search Ad (RSA) creatives (headlines, descriptions, and final URLs) alongside performance metrics.

## What you can build

Google Ads nodes are designed to be combined. Here are some examples of multi-node Workflow structures to build or adapt using Agent Assistant.

> [!TIP]
>
> Head to individual node description sections for more detailed use case examples.

### Cross-channel visibility audit

Find queries you're paying for that you already win in AI answers, and identify topics with strong AI visibility where you're not yet bidding.

* [Keywords](/integrations/google-ads/keywords) — pulls your active bid keywords for the last 30 days.
* [Visibility Score](/agents/workflows/nodes/data/visibility-score) **—** returns visibility metrics for topics where your brand appears in AI answers .
* **Code node** **—** custom logic to analyze both datasets and cross-reference paid keywords with AI visibility
* [Prompt LLM](/agents/workflows/nodes/prompt-llm) — generate the audit that includes data analysis and recommended actions, such as what to bid or not to bid on
* **Profound Docs node** — saves the audit as a formatted document for stakeholders.

### Weekly paid search digest

Build a Workflow that pulls the last seven days of performance, highlights movers, and posts a summary to Slack every Monday.

* [Search Terms](/integrations/google-ads/search-terms) — returns last seven days of actual user queries with CTR, CPC, CVR, CPA, and ROAS.
* [Ads](/integrations/google-ads/ads) — pulls RSA creative performance over the same window.
* [Prompt LLM](/agents/workflows/nodes/prompt-llm) — summarizes top performers, worst performers, and terms to consider adding as negatives.
* [Slack](/integrations/slack/connect-slack-to-profound) — posts the digest to your team channel. Schedule the Workflow to run weekly.

### Negative keyword discovery

Scan last week's search terms across all campaigns, flag wasted spend, and get actionable recommendations.

* [List Campaigns](/integrations/google-ads/list-campaigns) — pulls active campaign IDs and names to scope the analysis.
* [Search Terms](/integrations/google-ads/search-terms) — returns search term data for your Google Ads account.
* **Code node** — filters and trims the search term data to active campaigns.
* [Prompt LLM](/agents/workflows/nodes/prompt-llm) — flags terms with zero conversions and high cost, low CVR, or off-intent signals (informational queries, competitor brands). Returns a scannable Slack-ready report.
* [Slack](/integrations/slack/connect-slack-to-profound) — posts the negative keyword report to your team channel.

### Paid landing page optimization audit

Identify which landing pages are getting the most Google Ads traffic, check their AI visibility, and surface a prioritized list for Profound optimization work.

* [Landing Pages](/integrations/google-ads/landing-pages) — returns paid landing pages with spend, conversions, and page-experience metrics.
* [Visibility Score](/agents/workflows/nodes/data/visibility-score) — returns AI visibility metrics for topics and pages where your brand appears in AI answers.
* **Code node** — cross-references both datasets to flag pages with high paid spend but low AI visibility.
* [Prompt LLM](/agents/workflows/nodes/prompt-llm) — generates a prioritized list of pages to optimize in Profound, with a one-line reason for each.
* [Slack](/integrations/slack/connect-slack-to-profound) — posts the optimization priorities to your team channel.


# Search Terms
Source: https://docs.tryprofound.com/integrations/google-ads/search-terms

Return the queries that triggered your Google Ads with performance metrics

The **Search Terms node** returns the actual queries users typed that triggered your ads, paired with derived performance metrics: click-through rate (CTR), cost per click (CPC), conversion rate (CVR), cost per acquisition (CPA), and return on ad spend (ROAS) over a configurable date range.

## When to use this node

Use the Search Terms node for tasks such as:

* **Audit cross-channel search coverage:** find queries you're paying for in paid search that you already win in AI answers, and the reverse
* **Identify negative keyword opportunities:** flag high-cost, low-CVR queries that look off-intent
* **Build automated weekly paid search digests:** pull the last seven days of search term performance and highlight movers
* **Complement Profound AEO data:** layer paid query intent on top of AI visibility data in a single workflow

## Node configuration

### Google Integration (required)

Select a previously connected Google account, or click **+ Add integration** and follow the prompts to connect one. This is a one-time setup. Once you add an integration, it's available in all future sessions until you remove it in Profound **Settings > Integrations**.

### Google Ads Account (required)

Select the Google Ads customer account from the dropdown. The picker handles both Manager accounts and sub-account selection.

### Campaigns

Select the ad campaigns from the dropdown. To include every active campaign, leave the field empty.

### Ad Groups

Select the ad groups from the dropdown. To include every ad group within selected campaigns, leave the field empty. To see the ad group options in the dropdown, select at least one campaign in the **Campaigns** field.

### Date Range (required)

Select the date range to pull search term performance over. Choose from available presets (today, yesterday, last 7 days, and others) or set a custom range.

### Output Fields

Select the fields to include in the output. The options include metrics and campaign details, such as Impressions, Clicks, Conversions, Campaign Type, and Bid Strategy Type. To return output unfiltered, leave this field empty.

### Output Label

Enter a descriptive label for this step's output, for example `search_term_data` or `paid_search_last_7_days`. The label becomes the variable name you reference in later nodes.

## Output

The node outputs JSON data containing real search terms paired other data, such as the search query text, performance metrics (CTR, CPC, CPA, and so on), campaign and ad group details.

**Pass the output to:**

* [Prompt LLM node](/agents/workflows/nodes/prompt-llm) for AI analysis: negative keyword suggestions, intent categorization, budget reallocation recommendations, or cross-channel gap analysis
* [Slack node](/integrations/slack/connect-slack-to-profound) to push a weekly paid search summary to a channel
* **Google Sheets (Write Row) node** to log performance data for ongoing tracking
* **Code node** to write custom logic, extract variables, and pass results to a [**Conditional node**](/agents/how-tos/add-conditionals-to-workflows) for further branching

## Example usage: Weekly paid search digest

Set up a Workflow for your team to get a weekly performance summary report in Slack every Monday. Here is the example structure:

**1. Search Terms node**

* Select your **Google Integration** and set **Google Ads Account** to your Google Ads account
* Set **Date Range** to Last 7 days
* Set **Output** label to `search_term_data`
* Set **Campaign**, **Ad Groups**, **Output Fields** filters as needed

**2.** [Prompt LLM](/agents/workflows/nodes/prompt-llm) **node**

* Enter the prompt:

  ```text expandable wrap theme={null}
  You are a paid search analyst. Review this week's {{search term data}}.

  In plain English, summarize:
  - Top 5 best-performing terms (high CVR, low CPA)
  - Top 5 worst-performing terms (high spend, low CVR)
  - 3 terms to consider adding as negatives

  Keep the summary short and actionable.
  ```

* Set **Output label** to `weekly_summary`

> [!TIP]
>
> Check out our [prompt best practices](/agents/how-tos/prompt-llm-best-practices) guide for the Prompt LLM node.

**3.** [Slack](/integrations/slack/connect-slack-to-profound) **node**

* Select the Slack workspace and channel to receive updates
* Input the message text and use the `weekly_summary` variable in the **Message** field

Schedule the Workflow to run weekly, for example every Monday at 9 AM.

## Best practices

### Scope by campaign and ad group

Pulling all search terms data from all campaigns and ad groups at once can return an unmanageably large dataset. Use the built-in campaign and ad group filtering to scope results before passing them to subsequent nodes.

Use [List Campaigns](/integrations/google-ads/list-campaigns) node to discover the active campaign set first, map campaign names to IDs, or fan out the workflow across every campaign dynamically.

### Pair with Profound data

Search Terms node is useful on its own, but combined with Profound data nodes, such as [Visibility Score](/agents/workflows/nodes/data/visibility-score) node, it becomes a cross-channel strategy tool. There's no clear way to tell which of your paid search terms are also surfacing in AI answers without connecting both data sources.

### Use AI Marketer

Working with both search term data and Profound AI visibility data can have a learning curve. Use AI Marketer (Aim) to build a Workflow faster, and then edit the Workflow as needed to better match your use case.


# Connect Google Drive to Profound
Source: https://docs.tryprofound.com/integrations/google-drive/connect-google-drive-to-profound

A guide on how to connect your Google account to Profound and give Workflows access to your Google Drive files

This guide walks you through connecting your Google account to Profound. Once connected, Profound [Workflows](/agents/workflows/about-workflows) can read and create Google Docs and Google Sheets, read Google Slides, and you can import Drive files into a [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits.mdx).

> [!NOTE]
>
> Google Drive is a personal connection. Each user connects their own Google account, and only that user can see the connection and use it in Workflows and Knowledge Bases. Other people in your organization connect their own accounts.

## How Profound accesses your files

Profound requests Google's `drive.file` permission, so it can only reach the files you select in Google's file picker and the files Profound creates for you. Profound can't list, search, or read anything else in your Drive.

That's why connecting takes two steps: sign in to Google, then select the files or folders that Profound is allowed to use. You can add more files at any time.

## Before you start

You'll need:

* A Google account with access to the files you want to use
* A [Profound](https://www.tryprofound.com/) account with the **Admin** role. Only admins can connect accounts (learn more in [Roles and permissions](/platform-config/people/roles-and-permissions))
* A browser that allows pop-ups for Profound. The Google sign-in opens in a pop-up window

## Setup guide

**1. Open the Google Drive integration**

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Google Drive** and select it.

**2. Connect your Google account**

* In the **Connected accounts** card, select **+ Add account**.
* In the pop-up window, sign in to Google and allow the access Profound requests.
* When the window closes, Profound shows an **Account connected** message, and your Google account email appears in the **Connected accounts** list with a **Personal** badge.

**3. Give Profound access to files**

* On the connected account row, select **Add files from Google Drive** (the folder icon).
* In Google's file picker, select the files you want to use in Profound. You can select several files at once and browse shared drives.

Repeat this step whenever you need Profound to reach more files. Files that Profound creates through Workflows are available automatically.

**4. Verify the connection**

To confirm that Profound can reach a file:

* In Profound, [create a new Workflow](/agents/workflows/create-workflow), or open an existing Workflow draft.
* Add the [**Get Document**](/integrations/google-drive/using-google-drive-in-workflows#get-document) node from the **Google Docs** group, then select your account under **Google Account**.
* Under **Document**, select one of the files you added in [Step 3](#step-3).
* Select **Test node** (the play  icon) and check that the output contains the document text.

> [!TIP]
>
> You can also connect Google Drive directly in the Workflow Builder. Add a Google Docs, Google Sheets, or Google Slides node, open the **Google Account** dropdown in the node settings, and select **+ Add integration**. In the **Manage integrations** dialog, select **+ Add account** and complete the Google sign-in. Google's file picker then opens so you can select files right away.
>
> Every file dropdown in these nodes also has an **Add files from Google Drive** action that opens the picker for the selected account.

## Revoking access

To disconnect your Google account, go back to **Integrations** > **Google Drive** in Profound settings, open the connected account's **...** menu, and select **Revoke**.

Revoking deletes the tokens Profound stores for the account. Profound also revokes its access with Google, unless the same Google account has other active Google connections in your organization, such as Gmail or Google Search Console. Workflow nodes and Knowledge Base imports that use the connection stop working until you connect the account again.

If you remove Profound from your Google account's [third-party access page](https://myaccount.google.com/permissions) instead, the connection in Profound stops working and reports that it needs to be re-authorized. Revoke it in Profound and connect the account again.

## Troubleshooting

> **Profound reports that pop-ups are blocked**
>
> The Google sign-in opens in a pop-up window. Allow pop-ups for Profound in your browser, then select **+ Add account** again.

> **Profound can't connect the account**
>
> Closing the Google window before you finish, or declining the requested access, ends the connection attempt. Select **+ Add account** again and complete the sign-in.
>
> If a node later reports that the integration is missing required Google permissions, the connection was made without all the requested access. Revoke it and connect the account again, allowing everything Profound requests.

> **A node's file dropdown says no files are granted yet**
>
> Profound can only reach the files you selected in Google's file picker. Select **Add files from Google Drive** in the dropdown or on the account row in **Integrations**, then select the file.

> **A node reports that the integration needs to be re-authorized**
>
> Google no longer accepts Profound's token for the account. This happens when Profound's access is removed on the Google side. Revoke the connection in Profound and connect the account again.

> **A node reports that your Google Workspace domain doesn't allow Profound**
>
> Your Google Workspace administrator controls which third-party apps can access Workspace data. Ask them to allow Profound. Google documents the setting in [Control which third-party and internal apps access Google Workspace data](https://support.google.com/a/answer/7281227).


# Permissions and security
Source: https://docs.tryprofound.com/integrations/google-drive/google-drive-permissions-and-security

OAuth scopes, what Profound can and can't access in Drive, and how it stores, syncs, and deletes your tokens and files

Profound uses the Google Drive connection in [Knowledge Bases](/context-manager/knowledge-bases-and-brand-kits) and in [Workflows](/agents/workflows/about-workflows). Every request runs with the connected user's token, and every request first checks that the connection belongs to the requesting user's organization and to that user.

## Connection model

* **Each user connects their own Google account.** The connection goes through Google's standard [OAuth consent screen](https://developers.google.com/workspace/guides/configure-oauth-consent) and is private to the user who initiated it. Other users in your Profound organization can't use the same connection.
* **No service account, domain-wide delegation, or admin consent.** Profound doesn't use a [service account](https://cloud.google.com/iam/docs/service-account-overview) or [domain-wide delegation](https://support.google.com/a/answer/162106), and a Workspace administrator doesn't need to grant any permissions for a user to connect. Profound never accesses Drive with an identity other than the connected user's, and it can't reach anything that user can't open.
* **Access is narrower than the user's own access.** Profound requests the `drive.file` scope, so it can only read files the user picks in the [Google Picker](https://developers.google.com/workspace/drive/picker/guides/overview) and files Profound creates on the user's behalf. Profound can't list, search, or read anything else in the user's Drive, even files that user can open.

## OAuth scopes

> [!NOTE]
>
> Profound's Google Drive integration doesn't request any [restricted](https://developers.google.com/identity/protocols/oauth2/production-readiness/restricted-scope-verification#understand-restricted-scopes) Drive scope. See [Choose Google Drive API scopes](https://developers.google.com/workspace/drive/api/guides/api-specific-auth) for Google's classification of every Drive scope.

Profound requests these scopes when a user connects Google Drive:

| Scope | [Google's classification](https://developers.google.com/workspace/drive/api/guides/api-specific-auth#drive-scopes) | Why Profound requests it |
| - | - | - |
| `openid` | Non-sensitive | Returns a verified [ID token](https://developers.google.com/identity/openid-connect/openid-connect#obtainuserinfo) so Profound can tie the connection to one Google account and detect when a user reconnects the same account instead of creating a duplicate. Profound [validates the token](https://developers.google.com/identity/openid-connect/openid-connect#validatinganidtoken) (signature, audience, expiry, and issuer) before using it. |
| `https://www.googleapis.com/auth/userinfo.email` | Non-sensitive | Shows the connected account's email in Profound, so you can see which Google account a Knowledge Base import or Workflow run uses. Learn more in Google's [OAuth 2.0 scopes documentation](https://developers.google.com/identity/protocols/oauth2/scopes). |
| `https://www.googleapis.com/auth/drive.file` | Non-sensitive | Grants per-file access to files the user selects in the [Google Picker](https://developers.google.com/workspace/drive/picker/guides/overview) in the user's browser and to files Profound creates. Profound doesn't control the dialog and only learns which files the user picked. Learn more in Google's [Drive API scopes documentation](https://developers.google.com/workspace/drive/api/guides/api-specific-auth#nonsensitive-scopes). |

## User flows

### Knowledge Base imports

* A user with edit access to a [Knowledge Base](https://help.tryprofound.com/articles/5775538586-knowledge-bases-overview) selects files in the Google Picker. Profound receives the ID, name, and file type of each selected file.
* Profound reads each selected file's metadata and downloads its content. Google Docs export as `.docx`, Google Sheets as `.xlsx`, and Google Slides and Drawings as PDF. PDF, Word, Excel, plain text, Markdown, and CSV files download as they are. Files larger than 100 MB aren't imported.
* Profound resolves the parent folder names of the selected files so it can recreate the folder structure inside the Knowledge Base.

To open the Google Picker, Profound's web app requests the connected account's access token from Profound's backend and passes it to Google's [Picker library](https://developers.google.com/workspace/drive/picker/guides/overview) in the user's browser. The response carrying the token is marked `Cache-Control: no-store`.

#### Auto-sync

If a user turns on [auto-sync](https://help.tryprofound.com/articles/5775538586-knowledge-bases-overview#content-auto-sync) for an imported Google Drive folder, Profound checks for changes on the chosen cadence using Google's [Changes API](https://developers.google.com/workspace/drive/api/guides/manage-changes). On each run Profound:

* re-downloads imported files whose content changed, and skips the ones whose content is unchanged
* removes the Knowledge Base copy of a file that was deleted from Drive, or that Profound can no longer access or download

Profound performs these actions as the user who turned auto-sync on, using that user's connection. Auto-sync only tracks files that were imported. It can't see files the user didn't select.

### Workflow nodes

The Google Docs, Google Sheets, and Google Slides nodes in Workflows use the same Google Drive connection and the same `drive.file` grant. The read nodes read only the files the user selected. The create and write nodes (for example, creating a Google Doc or writing rows to a Google Sheet) create new files in the connected user's Drive or modify files the user selected.

Nodes that accept a file by name search only within the files the `drive.file` grant covers, and ask for a file ID or URL if the name matches more than one file.

## Operations Profound performs with the Drive API

| Operation | Drive API | Used for |
| - | - | - |
| Search files by name, optionally filtered by type | [`files.list`](https://developers.google.com/workspace/drive/api/reference/rest/v3/files/list) | File selection in Workflow node settings, resolving file names |
| Read file metadata, individually or in batches | [`files.get`](https://developers.google.com/workspace/drive/api/reference/rest/v3/files/get) | Imports, sync, folder-path reconstruction |
| List the children of a folder | [`files.list`](https://developers.google.com/workspace/drive/api/reference/rest/v3/files/list) | Available in Profound's Drive service. Not used by Knowledge Bases or Workflow nodes today |
| Download or export file content | [`files.get`](https://developers.google.com/workspace/drive/api/reference/rest/v3/files/get) (media), [`files.export`](https://developers.google.com/workspace/drive/api/reference/rest/v3/files/export), export links | Imports, sync, read nodes |
| List changes since the last sync | [`changes.getStartPageToken`](https://developers.google.com/workspace/drive/api/reference/rest/v3/changes/getStartPageToken), [`changes.list`](https://developers.google.com/workspace/drive/api/reference/rest/v3/changes/list) | Auto-sync |

These are the only Drive API operations the Google Drive connection performs. Writes to Drive happen only through the Google Docs and Google Sheets Workflow nodes.

## Where imported content is stored

Profound stores only the files that users import into a Knowledge Base. Workflow read nodes also download the content of the files they read from, but only for that run: the content becomes the node's output, which Profound keeps in the Workflow run's history like every other node output. For each imported file, Profound stores:

* the file content in an Amazon S3 bucket, logically separated in Profound's AWS account
* the parsed text in the same S3 bucket, and its search embeddings in a managed vector search service, so Knowledge Base search can find it
* the file's name, path, size, type, source link, and content hash in Profound's database

Profound uses imported content only as a source of information for the features you use it in, such as [Knowledge Base](/context-manager/knowledge-bases-and-brand-kits) search, [Workflows](/agents/workflows/about-workflows), and [FactCheck](/answer-engine-insights/factcheck/about). Profound doesn't use your files to train models.

## Deletion and retention

Profound keeps imported content until one of these happens:

* a user deletes the document, the folder, or the Knowledge Base in Profound
* auto-sync detects that the file was deleted from Drive or is no longer accessible

Deleting a document removes the stored file, its parsed text, and its search embeddings. Revoking the Google Drive connection doesn't delete files that were already imported. Delete them from the Knowledge Base if you need them removed.

## Token storage and handling

* **Encryption:** Profound envelope-encrypts the OAuth access and refresh tokens with an [AWS Key Management Service (KMS)](https://docs.aws.amazon.com/kms/latest/developerguide/overview.html) key reserved for customer data and stores the ciphertext in Amazon DynamoDB. The encryption context binds each record to one organization, one provider, and one user, so a record can't be decrypted under another organization's or user's identity. Profound's main database holds only a reference to the encrypted record, never the tokens.
* **Scoping:** Every credential read checks that the connection belongs to the requesting user's organization, and, for personal connections like Google Drive, to the requesting user. Profound records every credential read in an internal audit table, with the integration, the requesting endpoint, the user the read ran for, and a timestamp. This record isn't visible in the platform.
* **Refresh:** Profound requests [offline access](https://developers.google.com/identity/protocols/oauth2/web-server#offline) from Google, so it receives a refresh token and can keep Knowledge Base auto-sync running without asking the user to sign in again. When an access token expires, Profound's backend exchanges the refresh token for a new one. If several requests need the token at the same moment, only one of them contacts Google. The others wait for that refresh to finish and reuse its result.
* **What's stored about the account:** Profound stores the Google account's email address and its subject identifier (the `sub` key in the [ID token's payload](https://developers.google.com/identity/openid-connect/openid-connect#an-id-tokens-payload)) alongside the connection. The subject identifier never changes, even when the account's email address does, so Profound uses it to recognize when a user reconnects the same account.

## Revoking access

Users revoke the connection in Profound under **Settings** > **Integrations** > **Google Drive**, from the **...** action menu on the connected account, by selecting **Revoke**. When a user revokes:

* Profound deletes the encrypted token record and marks the connection as revoked. The deletion isn't reversible.
* If this is the account's last remaining Google connection in your organization, Profound also calls Google's [token revocation endpoint](https://developers.google.com/identity/protocols/oauth2/web-server#tokenrevoke) for the access token and the refresh token. Otherwise it defers that call, as explained below.
* Knowledge Base auto-sync and Workflow nodes that use the connection stop working until the user connects again.

The Google Drive, Gmail, Google Search Console, and Google Ads integrations identify themselves to Google with the same OAuth client (Google Analytics uses a separate one). Google treats all permissions a Google account has granted to that client as a single grant, and revoking any token invalidates every token Google issued to that account for Profound. This is why revoking one connection deletes its stored tokens in Profound immediately, but Profound asks Google to revoke the grant only when the account's last remaining connection is revoked. Connections made with other Google accounts aren't affected.

> [!NOTE]
>
> Users can also revoke Profound's access from their Google account's [third-party access page](https://support.google.com/accounts/answer/13533235). Profound's stored refresh token then stops working, and the user needs to reconnect the account in Profound before using the integration again.

## Controls for Google Workspace administrators

* **Allow or block Profound as an app.** Manage Profound's access as described in Google's [Control which third-party and internal apps access Google Workspace data](https://knowledge.workspace.google.com/admin/apps/control-which-apps-access-google-workspace-data). This setup may require Profound's OAuth client ID. Your Profound account team can provide it.
* **Limited access is enough for the Google Drive integration.** Google's [Limited access](https://knowledge.workspace.google.com/admin/apps/control-which-apps-access-google-workspace-data#trustorlimit) setting allows apps to use [unrestricted scopes](https://developers.google.com/workspace/drive/api/guides/api-specific-auth#nonsensitive-scopes) only. `drive.file`, `openid`, and `userinfo.email` are all unrestricted, so the Google Drive integration works under **Limited**.
* **One client covers Profound's Google integrations.** Because Profound's Google Drive, Gmail, Google Search Console, and Google Ads integrations share one OAuth client (Google Analytics has its own), an allowlist or block applies to all of them.


# Using Google Drive in Workflows
Source: https://docs.tryprofound.com/integrations/google-drive/using-google-drive-in-workflows

Learn how to read and write Google Docs, Google Sheets, and Google Slides with Workflow nodes

After you [connect your Google account](/integrations/google-drive/connect-google-drive-to-profound), the following nodes become available in Profound Workflows. Find them in the Workflow Builder under **Google Suite**: the **Google Docs** and **Google Sheets** groups, and the **Google Slides** node.

Every node runs as the Google account you select under **Google Account**. It can only reach the files you added through Google's file picker and the files Profound created, so [add the files](/integrations/google-drive/connect-google-drive-to-profound#step-3) before you build with them.

## Available nodes

**Create Document**

### Create Document

Creates a Google Doc.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Document Title** — The title of the new document

#### Optional inputs

| Input | Default | Description |
| - | - | - |
| **Content format** | **Plain Text** | How to read **Initial Content**: **Plain Text**, **Markdown**, or **HTML**. Use the format of the content you paste or reference. |
| **Initial Content** | - | The content of the new document. Markdown and HTML are converted to formatted text by Google. HTML must contain tags, such as `<p>...</p>`. |
| **Destination folder** | **My Drive** | The Drive folder to create the document in. Select a folder you added through the picker, or select **Add a folder from Google Drive** to add one. The folder can be in a shared drive. Accepts a variable that holds a folder ID. |

#### Output

A JSON object with the new document's `documentId`, `document_url`, and `title`.

> [!NOTE]
>
> If Profound can't write the content into a document it just created in a folder, it moves the empty document to the trash so the folder doesn't collect blank files.

**Get Document**

### Get Document

Reads the content of a Google Doc.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Document** — Search the documents you added and select one, or paste the document's URL. Also accepts a document ID or a variable from a previous node's output

#### Optional inputs

* **Export format** — **Plain text** (default), **HTML**, **Markdown**, or **Raw (Docs API JSON)**. Raw returns Google's full document JSON, including every tab.

#### Output

The document content in the selected format, as text.

Documents whose export exceeds 5 MB aren't returned and the node reports that the document is too large.

**Create Spreadsheet**

### Create Spreadsheet

Creates a spreadsheet and fills its first sheet with rows, or adds a new sheet with rows to an existing spreadsheet.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Title** — The title of the new spreadsheet, or of the new sheet when you add to an existing spreadsheet
* **Rows** — The data to write, in the format selected under **Input format**
* **Input format** — **CSV** (selected by default when you add the node), **TSV (tab-separated)**, **Markdown table**, **HTML table**, or **JSON**. JSON accepts an array of objects, whose keys become the header row, an object of arrays keyed by column, or an array of arrays. A code fence around the content is removed before parsing

#### Optional inputs

| Input | Default | Description |
| - | - | - |
| **Add to Existing Spreadsheet** | - | Leave empty to create a new spreadsheet. Select an existing spreadsheet, or paste its URL, to add a new sheet tab named **Title** instead. |
| **Value input option** (under advanced settings) | **Formula / number (interpret as typed)** | How Google reads the written values. **Formula / number** interprets formulas and numbers as if you typed them into a cell, for example `=SUM(A1:B1)`. **Plain text (no parsing)** stores them as text. |

#### Output

For a new spreadsheet, a JSON object with `spreadsheetId`, `spreadsheet_url`, and `title`. For a new sheet, a JSON object with `spreadsheetId`, `spreadsheet_url`, `sheetId`, and `sheet_title`.

> [!TIP]
>
> To copy data between sheets, set **Input format** to the **Output format** you used in [**Get Sheet Values**](#get-sheet-values). CSV, TSV, HTML, Markdown, and the JSON shapes round-trip.

**Get Row**

### Get Row

Reads one row from a sheet.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Spreadsheet** — Search the spreadsheets you added and select one, or paste its URL
* **Row selection** — **Specific row**, **First row**, or **Last row**
* **Row number** — The row to read when **Row selection** is **Specific row**. Matches the row numbers in the spreadsheet

#### Optional inputs

* **Sheet name** — The sheet tab to read. If omitted, the first sheet is used

#### Output

A JSON object with `row_number`, `values` (the cell values in order), and `record`. `record` maps the header names in row 1 to the row's values, and is `null` when the row you read is row 1 or the sheet is empty.

**Get Sheet Values**

### Get Sheet Values

Reads a range of cells.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Spreadsheet** — Search the spreadsheets you added and select one, or paste its URL

#### Optional inputs

| Input | Default | Description |
| - | - | - |
| **Sheet name** | First sheet | The sheet tab to read. |
| **Range** | Whole sheet | An A1 range within the sheet, for example `A1:D10`. |
| **Value render option** (under advanced settings) | **Formatted value** | **Formatted value** returns values as they're displayed, **Unformatted value** returns the underlying values, and **Formula** returns formulas instead of their results. |
| **Output format** (under advanced settings) | **Rows (2D array)** | The shape of the output: **Rows (2D array)**, **Records (list of dicts)**, **Columns (dict of lists)**, **CSV (string)**, **TSV (tab-separated)**, **HTML table**, **Markdown table**, or **Raw (full API response)**. Records, Columns, CSV, TSV, HTML, and Markdown treat the first row of the range as the header. |

#### Output

A JSON object with the data under `values`, in the selected shape. **Raw** returns Google's full response instead, with `values` and `majorDimension`.

> [!NOTE]
>
> **Records** and **Columns** need unique header names. If two columns share a name, rename one in the spreadsheet or pick another format.

**Update Cell**

### Update Cell

Writes one value to one cell.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Spreadsheet** — Search the spreadsheets you added and select one, or paste its URL
* **Cell** — The cell in A1 notation, for example `A1`
* **Value** — The value to write, as a string, number, or boolean

#### Optional inputs

* **Sheet name** — The sheet tab to write to. If omitted, the first sheet is used
* **Value input option** (under advanced settings) — **Formula / number (interpret as typed)** (default) or **Plain text (no parsing)**

#### Output

Google's update response, including the updated range.

**Write Row**

### Write Row

Appends one row to a sheet, or overwrites one existing row.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Spreadsheet** — Search the spreadsheets you added and select one, or paste its URL
* **Values** — A JSON array of cell values that map to columns in order, for example `["Alice", "alice@example.com", 42, true]`
* **Write mode** — **Append to end** (default) adds the row after the last row with data. **Overwrite specific row** replaces the row you set under **Row number**
* **Row number** — The row to overwrite when **Write mode** is **Overwrite specific row**

#### Optional inputs

* **Sheet name** — The sheet tab to write to. If omitted, the first sheet is used
* **Value input option** (under advanced settings) — **Formula / number (interpret as typed)** (default) or **Plain text (no parsing)**
* **Start column** (under advanced settings) — The column letter to start the row at, for example `A` or `E`. If empty, Google detects where the table is when appending, and overwriting starts at column `A`

#### Output

A JSON object with `updated_range` and the `row_number` that was written.

**Write Rows**

### Write Rows

Appends several rows to a sheet, or updates several existing rows, in one run.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Spreadsheet** — Search the spreadsheets you added and select one, or paste its URL
* **Mode** — **Add rows** (default) appends after the last row with data. **Update rows** overwrites the rows you name
* **Row Data Format** — **Arrays** (default) or **JSON Objects**
* **Row Data** — A JSON array with one entry per row. Its shape depends on **Mode** and **Row Data Format**:

| Mode | Arrays | JSON Objects |
| - | - | - |
| **Add rows** | Each inner array is one row, mapped to columns in order: `[["Alice", "alice@example.com", 90], ["Bob", "bob@example.com", 85]]` | Keys are column letters: `[{"A": "Alice", "B": "alice@example.com", "C": 90}]` |
| **Update rows** | The first element of each row is the row number to update: `[[1, "Alice", "alice@example.com", 95], [3, "Charlie", "charlie@example.com", 88]]` | Each object includes a `row_number` key: `[{"row_number": 1, "A": "Alice", "B": "alice@example.com"}]` |

#### Optional inputs

* **Sheet name** — The sheet tab to write to. If omitted, the first sheet is used
* **Value input option** (under advanced settings) — **Formula / number (interpret as typed)** (default) or **Plain text (no parsing)**
* **Start column** (under advanced settings) — The column letter to start rows at, for example `A` or `E`. If empty, Google detects where the table is when adding rows, and updates start at column `A`

#### Output

A JSON object with `rows_created`, `rows_updated`, and `updated_range`.

**Google Slides**

### Google Slides

Reads the text content of a Google Slides presentation.

#### Required inputs

* **Google Account** — Select a connected account from the dropdown
* **Presentation** — Search the presentations you added and select one, or paste its URL

#### Output

The presentation's text as plain text.

Presentations whose export exceeds 5 MB aren't returned and the node reports that the presentation exceeds the maximum size.

***

## Key concepts

### Files Profound can reach

Profound holds Google's `drive.file` permission, which covers only the files you selected in Google's file picker and the files Profound created. Each file dropdown lists those files for the selected account, and the **Add files from Google Drive** action at the bottom of the dropdown opens the picker so you can add more.

Files that Profound creates with **Create Document** and **Create Spreadsheet** are available to later nodes without adding them.

### Selecting a file

The **Document**, **Spreadsheet**, and **Presentation** inputs accept:

* A file selected from the dropdown
* The file's URL, such as `https://docs.google.com/document/d/<FILE_ID>/edit` or `https://drive.google.com/open?id=<FILE_ID>`
* The file's ID
* A variable from a previous node's output, for example the `documentId` returned by **Create Document**

If the value is neither an ID nor a URL, Profound treats it as the file name and looks for an exact match, ignoring case, among the files it can reach. The node stops with an error when no file matches or when more than one does. Use the ID or URL in that case.

### Sheets, ranges, and values

* **Sheet name** is the name of a tab in the spreadsheet. Leave it empty to work with the first tab.
* **Range** and **Cell** use A1 notation, for example `A1:D10` or `B2`. Leave **Range** empty to read the whole sheet.
* **Value input option** decides how Google stores what you write. **Formula / number (interpret as typed)** treats `=SUM(A1:B1)` as a formula and `42` as a number, as if you typed them into the cell. **Plain text (no parsing)** stores them as text.

***

## Troubleshooting

> **The file dropdown is disabled or says no files are granted yet**
>
> * Select a **Google Account** first. The dropdown stays disabled until you do.
> * If the dropdown reads **No files granted yet**, Profound can't reach any file for this account yet. Select **Add files from Google Drive** and pick the files in Google's picker.
> * Changing the **Google Account** clears the selected file, because the files each account can reach differ.

> **A node reports that the document, spreadsheet, or presentation isn't found**
>
> Profound can only open files you added through the picker or files it created. Check that the file is one of them and still exists in Drive. If you pasted a URL or ID, confirm that it belongs to the same Google account you selected.

> **A node reports that no file, or more than one file, matches a name**
>
> The value you entered isn't an ID or URL, so Profound searched by name. Select the file from the dropdown, or paste its URL or ID.

> **A Sheets node reports that a worksheet isn't found**
>
> **Sheet name** must match the tab name in the spreadsheet exactly. Copy it from the spreadsheet, or leave the field empty to use the first tab.

> **A Sheets node reports that the range is outside the sheet**
>
> The error tells you how many rows and columns the sheet has. Use a range within those bounds, or add rows and columns to the sheet first.

> **Get Sheet Values reports duplicate header names**
>
> **Records** and **Columns** output formats use the first row as keys, so every header must be unique. Rename the duplicate columns in the spreadsheet, or switch **Output format** to **Rows**, **CSV**, or another format.

> **Create Spreadsheet reports that the content appears to be a different format**
>
> **Input format** must match the data under **Rows**. For example, JSON data with **Input format** set to **CSV** is rejected. Change **Input format** to the format of your data.

> **Create Document reports that the HTML content has no tags**
>
> With **Content format** set to **HTML**, **Initial Content** must contain HTML tags, such as `<p>...</p>`. Wrap the text in tags or switch **Content format** to **Plain Text** or **Markdown**.

> **Create Document can't find or access the destination folder**
>
> * If Profound can't find the folder, select it again through **Add a folder from Google Drive**.
> * If Profound doesn't have access to the folder, select a folder where the connected Google account can add files.
> * For a folder in a shared drive, the connected account must be a member of the shared drive, and your Google Workspace rules must allow it to add files there.
> * If the drive or folder is full, free up space or select a folder in another drive.

> **A node reports that the document or presentation is too large**
>
> Profound exports up to 5 MB per document or presentation. Split the file, or read a smaller document.

> **A node reports that you don't have permission to edit**
>
> The connected Google account has view-only access to the file. Ask the file's owner for Editor access, then run the node again.

> **A node reports that Google's rate limit or quota was exceeded**
>
> Google limits how many requests it accepts in a period of time. Wait, then run the Workflow again. If this keeps happening, ask your Google Workspace administrator to review the API quota.

> **A node reports that the integration needs to be re-authorized**
>
> Google no longer accepts Profound's token for the account. In **Settings** > **Integrations** > **Google Drive**, revoke the connection and [connect the account again](/integrations/google-drive/connect-google-drive-to-profound#step-2).


# Integrations vs connectors: what's the difference?
Source: https://docs.tryprofound.com/integrations/integrations-vs-connectors

What integrations and connectors are, and which one to use for Agents, Workflows and Aim

Integrations and connectors link Profound to the tools your team already uses: your CMS, communication tools, analytics platforms, and data sources. Instead of copying data in and out of Profound by hand, you connect a service once and let Profound read from it and write to it as part of your regular workflow.

> [!TIP]
>
> To let your AI tools access your Profound data, connect them to the [Profound Model Context Protocol (MCP) server](/mcp/overview).

## The difference between an integration and a connector

An **integration** is a native Profound tool that connects an external service to [Profound Workflows](/agents/workflows/about-workflows). Each integration adds a set of nodes to the Workflow Builder, and each node performs a specific action in that service, such as creating a CMS entry, reading spreadsheet rows, or sending a message.

A **connector** connects an external service to Profound AI Marketer and Agents through MCP. Once connected, AI Marketer can call that service directly in a conversation to pull in its data and use it alongside your Profound data.

In short: integrations power Workflows, and connectors power AI Marketer and Agents. Some services are available as both, but the two lists are not identical, and you set up each one separately.

You manage both in the **Integrations** section of the Profound organization settings. You can also manage integrations directly from a node's configuration panel in the Workflow Builder.

## Using integrations in Profound Workflows

Once a service is connected, its nodes become available in the Workflow Builder. You chain them together and use them like any other node.

For example, a weekly reporting Workflow can pull your Visibility data with a [Profound data node](/agents/workflows/nodes/data/visibility-score), summarize the biggest changes with a [Prompt LLM node](/agents/workflows/nodes/prompt-llm), and post the summary to your team channel with the [Slack node](/integrations/slack/connect-slack-to-profound). The report is delivered to where your team works, without anyone opening a dashboard.

Each integration has its own set of nodes with specific inputs and outputs. Find the node descriptions on the page of each integration in [Integration guides](./overview).

## Using connectors with AI Marketer

Connectors extend what AI Marketer can see. When you connect a service, AI Marketer can query it in real time and blend its data with your Profound insights, so your analysis is no longer limited to what lives inside Profound.

For example, with the G2 connector in place, you can ask "Which of our topics with rising buyer intent on G2 have the lowest Visibility Score?", and AI Marketer will pull the intent data from G2, combine it with your Visibility data, and provide an answer grounded in both.

To add a connector, open the service's page in **Integrations**, go to the **Connector** tab, and authorize the connection.

## FAQ

> **Is there an integration and a connector for every service?**
>
> No. Integrations and connectors are built and prioritized separately, so a service may be available as one, both, or neither. Integrations require Profound-built Workflow nodes for each action, while connectors depend on the service exposing an MCP server. That is why the two lists overlap but do not match exactly.

> **What if the tool I need isn't available?**
>
> If the tool has a public API, you can often bring its data into a Workflow with the call API node. If the service has an MCP server, you can create a custom connector in **Integrations** > **Add custom connector** and use with AI Marketer and Agents.

> **What data can Profound access through an integration?**
>
> Only what its Workflow nodes need to do their job. For example, an integration whose node sends messages does not read your message history. You can review the requested permissions during the connection flow and revoke access at any time from the Integration page.

## Get help

Not sure whether you need an integration or a connector, or running into trouble connecting one? [Contact support](mailto:support@tryprofound.com) and the team will point you in the right direction.


# Connect Looker to Profound
Source: https://docs.tryprofound.com/integrations/looker/connect-looker-to-profound



[Looker](https://cloud.google.com/looker) is Google Cloud's enterprise business intelligence platform. Teams connect it to a data warehouse (BigQuery, Snowflake, Redshift, and others), define metrics in **LookML**, and use Looker's UI to explore data, build reports, and share dashboards.

This is **not** [Looker Studio](https://lookerstudio.google.com/) (formerly Google Data Studio). It is also **not** the same as Profound's [Tableau connector](/rest-api/integrations/tableau/usage), which exports Profound analytics into Tableau. Profound's Looker integration works in the opposite direction: **Profound Workflows query data from your Looker instance**.

## What Looker terms mean

| Term | What it is |
| - | - |
| **LookML** | Looker's modeling language. Analytics teams use it to define dimensions, measures, joins, and business logic on top of raw warehouse tables. |
| **Model** | A LookML project exposed for querying, such as `marketing` or `core`. Workflow inline queries reference a model by name. |
| **Explore** | A starting point for ad hoc analysis within a model — a curated set of related tables and fields users can query. Inline queries need both a model and an explore name. |
| **Look** | A saved report or query in Looker. If your team already built the metric in Looker, Workflows can run it directly with **Run Look**. |
| **Dashboard** | A collection of Looks and visualizations. Profound does not embed dashboards; Workflows run individual Looks or inline queries and receive the underlying data. |

## What this integration does

Profound connects to **your** Looker instance using API credentials (`client_id`, `client_secret`, and instance URL). Once connected, Profound Workflows can:

* **Run a saved Look** — execute an existing Look and return the result as JSON for downstream Workflow steps
* **Run an inline query** — query a specific model and Explore with custom fields, filters, sorts, and advanced options

> [!NOTE]
>
> The integration is **read-only** and **org-level**. Profound runs queries against Looker and returns the results; it does not create, update, or delete Looker content. One Looker connection is shared across your organization and can be selected in any Workflow that needs it.

## Where you can use Looker in Profound

| Surface | What it does |
| - | - |
| **Profound Workflows** | Add **Run Look** or **Run Inline Query** steps to pull Looker data into a Workflow |
| **Sheets** | Import rows from a saved Look into a spreadsheet for analysis or batch Workflow runs |

Both surfaces use the same org-level Looker connection and the same read-only query capabilities.

## Prerequisites

Before connecting Looker to Profound, ensure:

* You have access to a Google Cloud Looker instance (for example, `https://your-company.cloud.looker.com`)
* You can create or use a Looker user with API credentials
* That user has permission to access the saved Looks, models, Explores, and fields your Workflows will query

> **How to create Looker API credentials**
>
> 1. In Looker, go to **Admin → Users** and select the user Profound should use for API access
> 2. In the user's **API credentials** section, create a new API key pair
> 3. Copy the generated **Client ID** and **Client Secret** before leaving the page, Looker may not show the secret again.
> 4. Confirm the user can access the Looks, models, Explores, dimensions, and measures your Workflows will need

## Connect Your Account

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **Looker**, select it, then select **+ Add account**.
4. In the **Connect Looker** dialog, enter your connection details:
   * **Client Secret**: the Looker API `client_secret`
   * **Looker Base URL**: your instance URL, for example `https://your-company.cloud.looker.com`
   * **Client ID**: the Looker API `client_id`
5. Select **Connect account**.

Profound validates the credentials by authenticating with Looker (Looker API 4.0) and confirming the API user. Once connected, Looker appears as an available integration in Workflows and Sheets.

## Troubleshooting

> **The connection fails to save**
>
> Confirm the base URL is the root Looker instance URL, such as `https://your-company.cloud.looker.com`. Profound also accepts URLs that include `/api` or `/api/4.0` and normalizes them to the root instance URL.

> **No Looks appear in Workflow nodes**
>
> Make sure the Looker API user has access to the folders and saved Looks you expect to use. If permissions were recently changed, retry after a short delay.

> **A query returns a Looker permission error**
>
> Check the Looker user's role, model set, and field-level permissions. Profound can only run queries that the connected Looker user is allowed to run.


# Using Looker in Workflows
Source: https://docs.tryprofound.com/integrations/looker/using-looker-in-workflows



After connecting your Looker account, the following actions become available as Workflow steps. Each step requires you to select a connected **Looker Account** from a dropdown.

Profound pulls **query results** from Looker — rows of metrics and dimensions your analytics team has already modeled — not Looker dashboards or visualizations. Results are returned as structured data (typically JSON) that later Workflow steps can summarize, compare, or combine with other integrations.

## Available actions

| Action | What it does |
| - | - |
| **Run Look** | Executes a saved Look by ID and returns the result |
| **Run Inline Query** | Builds and runs a query against a model, Explore, and field list you specify in the Workflow |

**Run Look**

#### **Run Look**

Execute a saved Look from your Looker instance and return the result as structured data.

**Required inputs**

* **Looker Account**
* **Look** — a saved Look from the selected account. The dropdown loads Looks from Looker and shows each Look's title with its ID.

**Optional inputs**

* **Result Format** — the format returned by Looker. Defaults to `json_bi`, which is recommended for downstream Workflow steps.
* **Row Limit** — maximum number of rows to return.

**Output**

* **Look Results** — a JSON value containing the returned Looker result.

**Tip:** Use **Run Look** when the query already exists as a saved Look in Looker. This keeps business logic governed in Looker and is the safest option for recurring Workflow automations.

**Run inline query**

#### **Run Inline Query**

Run a query against a specific Looker model and Explore with fields, filters, sorts, and advanced query options.

**Required inputs**

* **Looker Account**
* **Model** — the Looker model name, such as `marketing`
* **Explore** — the Explore name, such as `web_event`
* **Fields** — comma-separated Looker fields, such as `web_event.date, web_event.unique_users`

**Optional inputs**

* **Sorts** — comma-separated sorts, such as `web_event.date desc`
* **Row Limit** — maximum rows to return. Defaults to `500`.
* **Result Format** — defaults to `json_bi`

**Advanced inputs**

* **Filter Expression** — Looker filter expression using `${field}` syntax
* **Pivots** — comma-separated fields to pivot on
* **Subtotals** — comma-separated fields to subtotal on
* **Fill Fields** — comma-separated fields to fill missing rows for, usually date fields
* **Dynamic Fields** — JSON string for table calculations or custom dimensions
* **Column Limit** — maximum number of columns to return
* **Total** — whether Looker should include totals
* **Row Total** — row total type, such as `right`
* **Query Timezone** — optional IANA timezone, such as `America/Los_Angeles`
* **Server Table Calcs** — whether table calculations should run on the server
* **Apply Formatting** — whether Looker formatting should be applied to the result

**Output**

* **Query Results** — a JSON value containing the returned Looker result.

**Tip:** Use field names exactly as they appear in Looker, including the `view.field` prefix. For date filters, use Looker filter syntax from your LookML model (for example, `30 days` on a date field).

***

## What data is available

Query results depend on the Look or Explore you run. In general, Looker returns the fields defined in your LookML model:

| Data type | Description |
| - | - |
| **Dimensions** | Attributes you group or filter by, such as date, channel, region, or product |
| **Measures** | Aggregated metrics, such as revenue, session count, conversion rate, or unique users |
| **Filters & sorts** | Applied at query time — either baked into a saved Look or specified in an inline query |
| **Totals & subtotals** | Optional row or column totals when enabled on inline queries |

Profound returns the raw query result as structured data (typically JSON). Your Workflow or sheet decides what to do with it — summarize it, compare periods, or pass rows to another step.

## Supported result formats

Both actions support multiple Looker result formats. **`json_bi`** is the default and recommended format for Workflows because it returns structured JSON that downstream steps can parse reliably.

Other supported formats: `json`, `csv`, `txt`, `html`, `md`, `xlsx`, `sql`, `png`, and `jpg`. Use non-JSON formats only when a later step explicitly needs that output type.

## Limits

* **Run Inline Query** defaults to **500 rows** and supports up to **5,000 rows** per run.
* **Run Look** supports an optional row limit override up to **5,000 rows**.
* The connected Looker user's permissions still apply: Profound cannot return data the API user cannot access.

***

## Choosing an action

| Use **Run Look** when… | Use **Run Inline Query** when… |
| - | - |
| The query is already saved and approved in Looker | The Workflow needs dynamic fields, filters, or date windows |
| You want Looker to remain the source of truth for query logic | You are building parameterized reporting inside the Workflow |
| You need a simple, governed recurring report | Earlier Workflow steps determine which metrics to fetch |

***

## Common use cases

* Pull a saved traffic or revenue Look into a weekly summary Workflow
* Fetch governed KPIs from Looker and pass them to a content-generation step
* Run dynamic inline queries with date windows that change based on Workflow inputs
* Combine Looker metrics with data from other integrations in a single Workflow
* Import a saved Look into a Sheet to analyze metrics alongside Workflow-generated columns

***

## Best practices

* Prefer `json_bi` for downstream Workflow steps because it returns structured data
* Start with a saved Look when possible; use inline queries only when the Workflow needs dynamic logic
* Keep row limits focused so later steps receive only the data they need
* Make sure the connected Looker user has access to every model, Explore, field, and Look used by the Workflow

***

## Common errors

> **Select an integration first**
>
> The Look dropdown depends on the selected Looker account. Choose a **Looker Account** before selecting a saved Look.

> **No Looks found**
>
> The connected Looker user may not have access to any saved Looks, or the Looks may be in folders the user cannot access.

> **Invalid model, Explore, or field**
>
> Inline queries must use Looker model, Explore, and field names exactly as they exist in Looker.

> **Permission denied**
>
> Profound uses the connected Looker API user. Update that user's Looker role, model set, or folder permissions if the Workflow cannot access the requested data.


# Connect Microsoft Outlook to Profound
Source: https://docs.tryprofound.com/integrations/microsoft-365/connect-microsoft-outlook-to-profound

Set up the Outlook integration for Profound Workflows

This guide walks you through connecting your Outlook mailbox to Profound. Once connected, Profound [Workflows](/agents/workflows/about-workflows) can send emails from your mailbox.

> [!NOTE]
>
> Profound connects to Microsoft services at the user level: each team member connects their own Microsoft account, and the Outlook node sends emails as that account. Profound doesn't read your mailbox.

## Before you start

You'll need:

* Microsoft account with an [Outlook](https://www.microsoft.com/en-us/microsoft-365/outlook/email-and-calendar-software-microsoft-outlook) mailbox
* [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Connect Outlook**

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Outlook**, select it, then select **Add account**.
* Sign in with the Microsoft account that owns the mailbox your Workflows should send emails from, and accept the permissions that Profound requests.

If your account appears under **Connected accounts** with the **Personal** label, the connection was successful.

**2. Verify the connection**

To confirm that Workflows can send emails from your mailbox, test the node in the Workflow builder:

* In Profound, [create a new Workflow](/agents/workflows/create-workflow) and choose **Workflow builder**, or open an existing Workflow draft.
* Add the [**Outlook**](/integrations/microsoft-365/using-outlook-in-workflows#outlook) node, then select your account under **Outlook Account**.
* Enter your own address under **To**, and fill in **Subject** and **Body**.
* Select **Test node** (the play  icon). The output reads `Message sent successfully.` and the email arrives in your inbox.

> [!WARNING]
>
> Testing this node sends a real email. Confirm the recipients and the content before running it.

You can also connect Outlook directly in the Workflow builder. Add the Outlook node, open the **Outlook Account** dropdown in the node settings, and select **Add integration**.

## Revoking access

Revoke Profound's access to your mailbox anytime by selecting **Revoke** in the connected account's action (**...**) menu. Outlook nodes that use the connection stop working until you connect the account again.

Other Microsoft integrations with their own connections, such as Microsoft SharePoint, stay available to Workflows.


# Connect Microsoft SharePoint to Profound
Source: https://docs.tryprofound.com/integrations/microsoft-365/connect-microsoft-sharepoint-to-profound

Set up the Microsoft SharePoint integration for Profound Workflows

This guide walks you through connecting your OneDrive and SharePoint files to Profound. Once connected, Profound [Workflows](/agents/workflows/about-workflows) can read and create Word documents, read and write Excel workbooks, and read PowerPoint presentations.

> [!NOTE]
>
> Profound connects to Microsoft services at the user level: each team member connects their own Microsoft account, and every Workflow node runs with that account's permissions. A node reaches only the files that the connected account can open.

## Before you start

You'll need:

* [Microsoft 365](https://www.microsoft.com/en-us/microsoft-365) work or school account with access to the [OneDrive](https://www.microsoft.com/en-us/microsoft-365/onedrive/online-cloud-storage) and [SharePoint](https://www.microsoft.com/en-us/microsoft-365/sharepoint/collaboration) files you want Workflows to use
* [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Connect Microsoft SharePoint**

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Microsoft SharePoint**, select it, then select **Add account**.
* Sign in with your Microsoft account in the pop-up dialog and accept the permissions that Profound requests.

If your account appears under **Connected accounts** with the **Personal** label, the connection was successful.

**2. Verify the connection**

To confirm that Workflows can reach your files, test a node in the Workflow builder:

* In Profound, [create a new Workflow](/agents/workflows/create-workflow) and choose **Workflow builder**, or open an existing Workflow draft.
* Add the Microsoft Word [**Get Document**](/integrations/microsoft-365/using-word-in-workflows#get-document) node, then select your account under **Microsoft Account**.
* Under **Document**, search for a Word document by name, paste its SharePoint or OneDrive link, or select **Browse documents in OneDrive or SharePoint** to pick one.
* Select **Test node** (the play  icon) and check that the output contains the document text.

If you plan to use Excel or PowerPoint nodes, test them the same way.

You can also connect Microsoft SharePoint directly in the Workflow builder. Add a Microsoft Word, Excel, or PowerPoint node, open the **Microsoft Account** dropdown in the node settings, and select **Add integration**.

## Revoking access

Revoke Profound's access to your files anytime by selecting **Revoke** in the connected account's action (**...**) menu. Word, Excel, and PowerPoint nodes that use the connection stop working until you connect the account again.

Other Microsoft integrations with their own connections, such as Outlook, stay available to Workflows.


# Troubleshooting Microsoft 365 setup and Workflow errors
Source: https://docs.tryprofound.com/integrations/microsoft-365/microsoft-365-integrations-troubleshooting

Fixes for common Microsoft 365 integration errors

## Connection and files

> **A node reports that the Microsoft integration needs to be re-authorized**
>
> Microsoft no longer accepts the credentials Profound stored, for example because your password changed or an administrator revoked Profound's access.
>
> Connect your Microsoft account again as described in the [Microsoft SharePoint](/integrations/microsoft-365/connect-microsoft-sharepoint-to-profound) or [Outlook](/integrations/microsoft-365/connect-microsoft-outlook-to-profound) setup guide.

> **The file picker reports that your tenant hasn't granted Profound access to SharePoint**
>
> Your Microsoft 365 tenant hasn't consented to the SharePoint permissions that Profound's app requests. Connecting your account again doesn't fix this: ask a tenant administrator to grant them, then try again.

> **The file field is disabled, or the picker doesn't show a file**
>
> * Select a **Microsoft Account** first. The field stays disabled until you do.
> * Search reaches only the files Microsoft has indexed. Use **Browse** to reach a file you created a moment ago.
> * Confirm that your Microsoft account can open the file. Profound sees only what you can see.

> **A node reports that a file was not found**
>
> Check that the link points to the file itself and that the file hasn't been moved or renamed. A Word document must be a `.docx` or `.docm` file, a workbook a `.xlsx` or `.xlsm` file, and a presentation a `.pptx` file.

> **A node reports that Microsoft is rate limiting requests**
>
> Microsoft limits how many requests it accepts in a period of time. Wait, then run the Workflow again.

> **A node reports that the requested Microsoft item was not found**
>
> The node can't find the folder or the document you're trying to write to. Use **Browse folders in OneDrive or SharePoint** to select a valid folder or document.

## Outlook

> **Outlook refuses the sender address**
>
> The **From** address must be the connected mailbox's own address or one of its aliases. Shared mailboxes and other users' mailboxes aren't supported. Leave **From** empty to send as the connected account.

> **The send fails because of a recipient address**
>
> Every address under **To**, **Cc**, and **Bcc** must be valid, and a single email can have up to 500 recipients across the three fields combined. Remove or correct the address and run the node again.

## Word

> **A created Word document loses its tables or images**
>
> This is expected. [**Create Document**](/integrations/microsoft-365/using-word-in-workflows#create-document) keeps headings, lists, bold, italic, and links, and turns tables, code blocks, and images into plain text.

> **A document is too large to read**
>
> [**Get Document**](/integrations/microsoft-365/using-word-in-workflows#get-document) refuses a file past its size limit rather than returning part of it. Split the document, or copy the pages you need into a smaller file.

## Excel

> **Get Row reports that a row number is required**
>
> In [**Get Row**](/integrations/microsoft-365/using-excel-in-workflows#get-row), set **Row selection** to **Specific row** and enter the **Row number**, or change the selection to **First row** or **Last row**.

> **Two Workflow steps append to the same workbook at once**
>
> Excel can't safely add rows from two sources at the same time, so Profound waits for the first write to finish before it starts the second. If the wait takes too long, the node reports that another append to the workbook is in progress, and nothing is written. Run the step again, or combine the rows into a single [**Write Rows**](/integrations/microsoft-365/using-excel-in-workflows#write-rows) node instead of several.

> **A read is refused for being larger than one response serves**
>
> The range holds more data than [**Get Values**](/integrations/microsoft-365/using-excel-in-workflows#get-values) or **Get Row** returns in one call. Narrow the **Range**, or read the worksheet in parts.

> **A number or a formula is written to the worksheet as text**
>
> In the node settings, set **Value property** to **Values**, which interprets each value as typed. Use **Formulas** when you want the value written as a formula.

> **A workbook has no visible worksheet**
>
> The node uses the first visible worksheet when **Worksheet name** is empty. If every worksheet is hidden, enter the name of the worksheet to use.

> **A write node reports a value error**
>
> The data a node attempts to write isn't in the correct format. To see the expected format, clear the data field (usually **Rows** or **Row data**) and read the placeholder text.

## PowerPoint

> **A presentation is too large to read**
>
> [**Get Content**](/integrations/microsoft-365/using-powerpoint-in-workflows#get-content) refuses a file past its size limit rather than returning part of it. Split the presentation, or copy the slides you need into a smaller file.


# Using Excel in Workflows
Source: https://docs.tryprofound.com/integrations/microsoft-365/using-excel-in-workflows

Read and write Excel workbooks with Microsoft Excel Workflow nodes

After you [connect Microsoft SharePoint](/integrations/microsoft-365/connect-microsoft-sharepoint-to-profound), the following nodes become available in Profound Workflows under **Microsoft Suite** > **Microsoft Excel**. Use them to read values and rows from a workbook, write cells and rows back, and create new workbooks or worksheets.

Every node runs with the permissions of the account you select under **Microsoft Account**. A node reaches only the files that the connected account can open.

> [!NOTE]
>
> New to Workflows? See [Getting started with Workflows](https://help.tryprofound.com/articles/2212787792-create-an-agent) in the Profound Help Center.

## Shared settings

Every node that works on an existing workbook takes a **Workbook** and an optional **Worksheet name**. Leave the worksheet empty to use the first visible worksheet. [**Create Workbook**](#create-workbook) creates a new file instead, so it has no **Workbook** field.

### Choosing a workbook

In the node settings, select your **Microsoft Account**, then select a workbook in one of the following ways:

* **Search by name** in the **Workbook** field. Search reaches only the files Microsoft has indexed, so a file created a moment ago may not appear yet.
* **Paste the SharePoint or OneDrive link** of the workbook into the same field.
* **Browse** by selecting **Browse workbooks in OneDrive or SharePoint** at the bottom of the list. This opens the Microsoft file picker, which also finds files search can't show yet.
* **Use a variable** from an earlier node's output, for example the `workbook_ref` that [**Create Workbook**](#create-workbook) returns.

The nodes work with `.xlsx` and `.xlsm` files.

### Value property

Several nodes take a **Value property** under **Advanced settings**. It controls how cell values are handled:

* **Values** — Interpret the value as typed, or as returned by the read nodes, so `10` becomes the number 10 and `=A1+1` becomes a formula.
* **Formulas** — Treat every value in that field as a formula.
* **Text** — Return the cell as it's displayed. Available only in the read nodes, such as [Get Values](#get-values).

A single write can carry up to 100,000 cells, and a cell can hold up to 32,767 characters. Numbers longer than 15 digits are written as text, so Excel doesn't round them.

## Available nodes

**Get Values**

### Get Values

Reads the values in a range of cells.

#### Required inputs

* **Microsoft Account** — Select a connected account from the dropdown
* **Workbook** — Search for a workbook by name, paste its SharePoint or OneDrive link, or browse for it

#### Optional inputs

| Input | Default | Description |
| - | - | - |
| **Worksheet name** | First visible worksheet | The worksheet to read values from. |
| **Range** | Used range (the range of cells that have values) | A range of cells in A1 notation, such as `A1:D10`. Leave it empty to read all values in the worksheet. |
| **Value property** (under **Advanced settings**) | **Values** | Which cell property to return. Learn more in [Value property](#value-property) |
| **Output format** (under **Advanced settings**) | **Rows (2D array)** | The shape of the output: **Rows (2D array)**, **Records (list of dicts)**, **Columns (dict of lists)**, **CSV (string)**, **TSV (tab-separated)**, **HTML table**, **Markdown table**, or **Raw (full API response)**. **Records** and **Columns** use the first row of the range as column names. |

#### Output

A JSON object with the values in the selected format, the output format, the workbook reference ID, the name of the worksheet, and the range of cells that was read.

> [!TIP]
>
> To write the values back with [**Create Workbook**](#create-workbook), select the same output format in this node as the input format in that node.

**Get Row**

### Get Row

Reads one row in a worksheet.

#### Required inputs

* **Microsoft Account** — Select a connected account from the dropdown
* **Workbook** — Search for a workbook by name, paste its SharePoint or OneDrive link, or browse for it
* **Row selection** — **Specific row** (default), **First row**, or **Last row**
* **Row number** — Required when **Row selection** is set to **Specific row**. Rows are numbered from `1` up to `1,048,576`

#### Optional inputs

* **Worksheet name** — The worksheet to read the row from. Defaults to the first visible worksheet
* **Range** — Limits which columns the row is read from. For example, `A1:D10` returns only columns A to D. Leave it empty to read every column that has data

#### Output

A JSON object with the row as an array of values, the row number, and a `record` field that pairs each row value with the column name from the first row.

When the row read is the first row, the `record` field is empty.

**Update Cell**

### Update Cell

Writes a value into a single cell.

#### Required inputs

* **Microsoft Account** — Select a connected account from the dropdown
* **Workbook** — Search for a workbook by name, paste its SharePoint or OneDrive link, or browse for it
* **Cell** — The cell to write to, such as `B2`
* **Value** — The value to write: text, a number, a boolean, or a formula such as `=SUM(A1:B1)`. To write a formula, select **Formulas** under **Value property**

#### Optional inputs

* **Worksheet name** — The worksheet to write to. Defaults to the first visible worksheet
* **Value property** (under **Advanced settings**) — **Values** (default) or **Formulas**. Learn more in [Value property](#value-property)

#### Output

A JSON object with the workbook reference ID and URL, the name of the worksheet, the range of cells that was updated, and the number of updated cells.

**Write Rows**

### Write Rows

Adds or updates rows in a worksheet.

#### Required inputs

| Input | Description | Notes |
| - | - | - |
| **Microsoft Account** | Select a connected account from the dropdown | - |
| **Workbook** | Search for a workbook by name, paste its SharePoint or OneDrive link, or browse for it | - |
| **Mode** | **Add rows** (default) appends the rows after the last non-empty row. **Update rows** overwrites existing rows | - |
| **Row data format** | **Arrays** (default) or **JSON Objects** | Learn more in [How to shape Row data](#how-to-shape-row-data) |
| **Row data** | The rows to write | See [How to shape Row data](#how-to-shape-row-data) |

#### How to shape **Row data**

| Row data format | Mode | Shape | Example |
| - | - | - | - |
| **Arrays** | **Add rows** | Each inner array is one worksheet row. Its values are written to columns in order | `[["Alice", "alice@example.com", 90], ["Bob", "bob@example.com", 85]]` |
| | **Update rows** | The first value of each inner array is the number of the row to update | `[[1, "Alice", "alice@example.com", 95], [3, "Charlie", "charlie@example.com", 88]]` |
| **JSON Objects** | **Add rows** | Each object's keys are column letters | `[{"A": "Alice", "B": "alice@example.com", "C": 90}]` |
| | **Update rows** | Each object also carries a `row_number` key with the number of the row to update | `[{"row_number": 1, "A": "Alice", "B": "alice@example.com"}]` |

#### Optional inputs

* **Worksheet name** — The worksheet to write to. Defaults to the first visible worksheet
* **Value property** (under **Advanced settings**) — **Values** (default) or **Formulas**. Learn more in [Value property](#value-property)
* **Start column** (under **Advanced settings**) — The column to start writing at, for example `A` or `E`. Leave empty to start from the first column (`A`)

#### Output

A JSON object with the workbook reference ID and URL, the name of the worksheet, the range of cells that was updated, and the number of rows that were created and updated.

> [!NOTE]
>
> In **Update rows** mode, the rows are written one at a time. If Excel reports a conflict partway through, the rows written before it stay written. Run the node again to rewrite the same cells.

**Create Workbook**

### Create Workbook

Creates a workbook, or adds a worksheet to a workbook that already exists.

#### Required inputs

| Input | Description | Notes |
| - | - | - |
| **Microsoft Account** | Select a connected account from the dropdown | - |
| **Title** | The name of the new workbook, or of the new worksheet when **Add to Existing Workbook** is set | Enter the name without the file extension. The node appends `.xlsx` automatically |
| **Rows** | The data to write into the new workbook or a worksheet | Select the data format under **Input format** |
| **Input format** | The format of the data to write: **JSON** (default), **Markdown table**, **CSV**, **TSV (tab-separated)**, or **HTML table** | JSON accepts an array of objects, an array of arrays, or an object that maps each column name to a list of values |

#### Optional inputs

| Input | Default | Description |
| - | - | - |
| **Add to Existing Workbook** | Empty | Leave it empty to create a new workbook. Select a workbook to add a worksheet to it instead. **Destination folder** and **Worksheet name** don't apply to an existing workbook, so they disappear when you select one. |
| **Destination folder** | OneDrive root | The folder that receives the new workbook. Search for a folder by name, paste its link, or browse for it. |
| **Worksheet name** (under **Advanced settings**) | `Sheet1` | The name of the first worksheet in the new workbook. |
| **Value property** (under **Advanced settings**) | **Values** | **Values** or **Formulas**. Learn more in [Value property](#value-property) |

#### Output

A JSON object with the workbook reference ID, URL, and title, and the name of the worksheet the data was written to.

> [!TIP]
>
> To store the output of [**Get Values**](#get-values) in a new workbook, select the same format under **Output format** in that node and under **Input format** in this node.

## Troubleshooting

See [Troubleshooting Microsoft 365 setup and Workflow errors](/integrations/microsoft-365/microsoft-365-integrations-troubleshooting#excel) for errors the Excel nodes report.


# Using Outlook in Workflows
Source: https://docs.tryprofound.com/integrations/microsoft-365/using-outlook-in-workflows

Send emails from your mailbox with the Outlook Workflow node

After you [connect your Outlook mailbox](/integrations/microsoft-365/connect-microsoft-outlook-to-profound), the following nodes become available in Profound Workflows under **Microsoft Suite**. Use them to send emails from the connected mailbox.

Every node runs against the account you select under **Outlook Account** and sends emails as that account.

## Available nodes

**Outlook**

### Outlook

Sends an email from the connected mailbox. Microsoft saves the sent email in the mailbox's **Sent Items**.

#### Required inputs

* **Outlook Account** — Select a connected Outlook account from the dropdown
* **To** — One or more recipient addresses
* **Subject**
* **Body** — The message text. Enter it as plain text. If the text looks like Markdown (headings, lists, links, bold text), Profound converts it to HTML before sending. Otherwise, it's sent as plain text. The node can't send attachments

#### Optional inputs (Advanced settings)

| Input | Description |
| - | - |
| **From** | A different sender address. It must be the connected mailbox's own address or one of its aliases. Sending from a shared mailbox or another user's mailbox isn't supported. |
| **Cc** and **Bcc** | Additional recipients. An invalid address makes the send fail. |
| **Reply-To** | The addresses that replies go to. |

A single email can have up to 500 recipients across **To**, **Cc**, and **Bcc** combined.

#### Output

A text confirmation that the message was sent, followed by the body that was sent.

> [!WARNING]
>
> Selecting **Test node** sends a real email. Confirm the recipients and the content before running the node.

## Troubleshooting

See [Troubleshooting Microsoft 365 setup and Workflow errors](/integrations/microsoft-365/microsoft-365-integrations-troubleshooting#outlook) for errors the Outlook node reports.


# Using PowerPoint in Workflows
Source: https://docs.tryprofound.com/integrations/microsoft-365/using-powerpoint-in-workflows

Read PowerPoint presentations with the Microsoft PowerPoint Workflow node

After you [connect Microsoft SharePoint](/integrations/microsoft-365/connect-microsoft-sharepoint-to-profound), the following nodes become available in Profound Workflows under **Microsoft Suite** > **Microsoft PowerPoint**. Use them to read the text of a presentation, slide by slide.

Every node runs with the permissions of the account you select under **Microsoft Account**. A node reaches only the files that the connected account can open.

## Available nodes

**Get Content**

### Get Content

Reads the text of a presentation.

#### Required inputs

* **Microsoft Account** — Select a connected account from the dropdown.
* **Presentation** — A `.pptx` file. Provide it in one of the following ways:
  * **Search by name**: search reaches only the files Microsoft has indexed, so a file created a moment ago may not appear yet.
  * **Paste the SharePoint or OneDrive link** of the presentation.
  * **Browse** by selecting **Browse presentations in OneDrive or SharePoint** at the bottom of the presentations list. This opens the Microsoft file picker, which also finds files search can't show yet.
  * **Use a variable** from an earlier node's output.

#### Optional inputs

* **Include speaker notes** — Adds each slide's speaker notes after its text, under a `Speaker notes:` line. Off by default.

#### Output

The presentation text as one string in Markdown format, in slide order. Each slide starts with a `## Slide N` heading, and a slide with no text keeps its heading. Bulleted paragraphs become list items, and nested bullets are indented. Table cells become plain paragraphs. Images and charts aren't included.

> [!NOTE]
>
> The node reads up to 5,000,000 characters of text from a file of up to 25 MB. Reading a larger presentation fails with an error and doesn't return any results.

## Troubleshooting

See [Troubleshooting Microsoft 365 setup and Workflow errors](/integrations/microsoft-365/microsoft-365-integrations-troubleshooting) for errors the PowerPoint nodes report.


# Using Word in Workflows
Source: https://docs.tryprofound.com/integrations/microsoft-365/using-word-in-workflows

Read and create Word documents with Microsoft Word Workflow nodes

After you [connect Microsoft SharePoint](/integrations/microsoft-365/connect-microsoft-sharepoint-to-profound), the following nodes become available in Profound Workflows under **Microsoft Suite** > **Microsoft Word**. Use them to read the text of a Word document and to create new documents in OneDrive or SharePoint.

Every node runs with the permissions of the account you select under **Microsoft Account**. A node reaches only the files that the connected account can open.

## Available nodes

**Create Document**

### Create Document

Creates a Word document from plain text, Markdown, or HTML.

#### Required inputs

* **Microsoft Account** — Select a connected account from the dropdown
* **Title** — The document title without the file extension (the node appends `.docx` automatically). If a file with the same name already exists in the selected folder, Microsoft 365 gives the new file a numbered name, such as `Document 1`

#### Optional inputs

| Input | Default | Description |
| - | - | - |
| **Content format** | **Plain text** | How to read the data in the **Initial content** field: **Plain text**, **Markdown**, or **HTML**. |
| **Initial content** | Empty | The document body, up to 1,000,000 characters. The writer keeps headings, lists, bold, italic, and links, and turns tables, code blocks, and images into plain text. |
| **Destination folder** | OneDrive root | The folder that the new document is saved into. Search for a folder by name, paste its link, or browse for it. |

#### Output

A JSON object with the saved document's reference ID, URL, title, and name. The name and title values are the ones Microsoft 365 saved. If they differ from the **Title** you entered, a file with that title already existed.

> [!TIP]
>
> Pass `document_ref` from the output to the [**Get Document**](#get-document) node as the **Document** to read the contents of the created document.

**Get Document**

### Get Document

Reads the text of a Word document.

#### Required inputs

* **Microsoft Account** — Select a connected account from the dropdown
* **Document** — A `.docx` or `.docm` file. Provide it in one of the following ways:
  * **Search by name**: search reaches only the files Microsoft has indexed, so a file created a moment ago may not appear yet.
  * **Paste the SharePoint or OneDrive link** of the document.
  * **Browse** by selecting **Browse documents in OneDrive or SharePoint** at the bottom of the documents list. This opens the Microsoft file picker, which also finds files search can't show yet.
  * **Use a variable** from an earlier node's output, for example the `document_ref` that [**Create Document**](#create-document) returns.

#### Optional inputs

* **Export format** — **Markdown** (default) or **Plain text**. Markdown keeps headings, bullet and numbered lists, bold, italic, tables, and links. Plain text separates paragraphs with a newline and writes each table row as one line with a tab between cells.

#### Output

The document body as one string in the selected format. Tracked changes are applied as accepted. The output leaves out headers, footers, footnotes, endnotes, comments, and images.

> [!NOTE]
>
> The node reads up to 5,000,000 characters of text from a file of up to 25 MB. A larger document fails with an error instead of returning a partial result.

## Troubleshooting

See [Troubleshooting Microsoft 365 setup and Workflow errors](/integrations/microsoft-365/microsoft-365-integrations-troubleshooting) for errors the Word nodes report.


# Get started with integrations
Source: https://docs.tryprofound.com/integrations/overview

Connect your tools to make them available to Profound Workflows, Agents and Aim

> [!NOTE]
>
> **Available on**: all plans
>
> **Find it in**: Settings in the organization name menu at the top left of the screen > Account > Integrations
>
> **Required role**: Admin to connect and manage integrations. All roles can view them.

Integrations connect third-party tools to Profound. After you connect an account, its actions and data become available to Profound Workflows, Agents and AI Marketer (Aim). Learn more in [Integrations vs connectors: what's the difference?](/integrations/integrations-vs-connectors).

> [!NOTE]
>
> This section covers tool integrations and connectors for Profound Workflows, Agents and Aim. Agent Analytics log integrations are documented in the [Agent Analytics section](/agent-analytics/aws_firehose_cloudfront). To connect AI tools to the Profound MCP server, see the [MCP documentation](/mcp/overview).

## Connecting your tools to Profound

Open the Integrations page in either of these ways:

1. Select your organization name in the top-left corner of the platform, select **Settings**, then select **Integrations** under **Account** in the left sidebar.
2. Go to **Context > Integrations** in the left sidebar.

Both paths open the Integrations page, which shows a card for each available tool. Select a card to connect your account or check the status of an existing connection.

## Integration guides

> **[Adobe Experience Manager](/integrations/adobe-experience-manager/setup/aem-setup-guide)**
>
> Find, create, and update Content Fragments in AEM as a Cloud Service

> **[Contentful](/integrations/contentful/connect-contentful-to-profound)**
>
> Create and update entries in a Contentful space

> **[Drupal](/integrations/drupal/connect-drupal-to-profound)**
>
> Work with content on your Drupal site

> **[Framer](/integrations/framer/connect-framer-to-profound)**
>
> Browse collections and create items in Framer CMS

> **[G2](/integrations/g2/connect-g2-to-profound)**
>
> Bring G2 reviews, ratings, and buyer intent into Workflows, Agents and Aim

> **[Gamma](/integrations/gamma/connect-gamma-to-profound)**
>
> Generate content in Gamma

> **[Gong](/integrations/gong/connect-gong-to-profound)**
>
> Get call details and transcripts from Gong

> **[Google Drive](/integrations/google-drive/google-drive-permissions-and-security)**
>
> Use your Drive files in Knowledge Bases and Workflows

> **[Looker](/integrations/looker/connect-looker-to-profound)**
>
> Query data from your Looker instance

> **[Microsoft Outlook](/integrations/microsoft-365/connect-microsoft-outlook-to-profound)**
>
> Send emails from your mailbox with Profound Workflows

> **[Microsoft SharePoint](/integrations/microsoft-365/connect-microsoft-sharepoint-to-profound)**
>
> Work with Word, Excel, and PowerPoint files in OneDrive and SharePoint

> **[Salsify](/integrations/salsify/connect-salsify-to-profound)**
>
> Work with your Salsify Product Lists

> **[Sanity](/integrations/sanity/sanity)**
>
> Create, update, and publish content in your Sanity project

> **[Slack](/integrations/slack/connect-slack-to-profound)**
>
> Bring Aim into your Slack workspace and send Agent results to channels

> **[WordPress](/integrations/wordpress/connect-wordpress-to-profound)**
>
> Work with posts on your WordPress site


# Connect Salsify to Profound
Source: https://docs.tryprofound.com/integrations/salsify/connect-salsify-to-profound

A guide on how to set up the Salsify integration and connect it to Profound Workflows

This guide walks you through connecting your Salsify organization to Profound. Once connected, Profound [Workflows](/agents/workflows/about-workflows) can work with your Salsify Product Lists.

> [!NOTE]
>
> Profound connects to Salsify at the organization level. One connected account gives every Workflow in your Profound organization access to one Salsify organization, with the permissions of the Salsify user that owns the API key.

## Before you start

You'll need:

* [Salsify](https://www.salsify.com/) account with permissions to create users, API keys, and Product Lists
* [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Create a Salsify user for Profound**

Profound acts in Salsify as the user that owns the API key, so Salsify applies that user's permissions to every request.

* In Salsify, [create a dedicated user](https://help.salsify.com/ProductXM/s/article/manage-users-and-roles) for the integration, for example `profound-integration@yourcompany.com`.
* Give the user access only to the products and Product Lists that you want to use with Profound.

> [!NOTE]
>
> Creating a dedicated user for Profound is a good practice because it gives you one switch for the integration: disabling that user in Salsify disables the entire integration.

**2. Copy the API key**

* Sign in to Salsify as the user from [Step 1](#step-1).
* Open your user profile menu and select **API Access**.
* Select **Show API key**, copy the key, and store it securely in your secret manager.

**3. Copy the organization ID**

The organization ID is the segment of any Salsify URL that follows `/orgs/`. For example:

```text wrap theme={null}
https://app.salsify.com/app/orgs/<ORGANIZATION_ID>/products
```

Copy your organization ID and keep it handy to use in later steps.

**4. Copy the Product List IDs**

Profound reads products from and writes membership changes to Product Lists. Salsify's API doesn't let Profound create lists for you, so create them in Salsify first as instructed in Salsify's [Creating Product Lists](https://help.salsify.com/ProductXM/s/article/creating-new-products-lists) documentation.

Once you have your list ready, open it and copy the ID from the URL segment that follows `product_lists/`:

```text wrap theme={null}
https://app.salsify.com/app/orgs/s-1a2b3c4d5e6f7890/product_lists/<PRODUCT_LIST_ID>
```

Depending on your organization, a Product List ID is either a number or a UUID with an `s-` prefix (for example, `s-123a-456b-789c-...`). Profound accepts both.

Copy the ID and keep it handy to use in later steps.

**5. Connect Salsify to Profound**

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Salsify**, select it, then select **+ Add account**.
* In the **Connect Salsify** dialog, enter the **Salsify API key** from [Step 2](#step-2) and the **Organization ID** from [Step 3](#step-3).
* Select **Connect account**.

You can also connect Salsify directly in the Workflow Builder. Add a Salsify node, open the **Salsify Organization** dropdown in the node settings, and select **+ Add integration**.

If your Salsify organization appears in the list of connected accounts, the connection was successful.

**6. Verify the connection**

To confirm that the Salsify user can reach the products your Workflows will use, test the nodes in the Workflow Builder:

* In Profound, [create a new Workflow](/agents/workflows/create-workflow), or open an existing Workflow draft.
* Add the Salsify [**List Products**](/integrations/salsify/using-salsify-in-workflows#list-products) node, then select the connected Salsify organization.
* Enter the source list ID from [Step 4](#step-4) under **Product Lists**.
* Select **Test node** (the play  icon) and check that the output contains the products you expect.
* If you intend to update products or lists with Profound Workflows, test the [**Update Products**](/integrations/salsify/using-salsify-in-workflows#update-products), [**Add Products to List**](/integrations/salsify/using-salsify-in-workflows#add-products-to-list), and [**Remove Products from List**](/integrations/salsify/using-salsify-in-workflows#remove-products-from-list) nodes the same way.

## Updating or revoking access

* To replace the API key or the organization ID, go back to **Integrations** > **Salsify** in Profound settings, select **Edit credentials** in the connected account's action **...** menu, enter the new values, and select **Update**.
* To revoke Profound's access, select **Revoke** in the same menu. Salsify Workflow nodes stop running until you connect Salsify again.

## Troubleshooting

> **API Access tab is missing from your Salsify profile menu**
>
> Your Salsify user doesn't have permission to use the API. Ask your Salsify administrator to grant it or to create the Profound integration user for you.

> **Profound can't save the credentials**
>
> * Check that you copied the whole API key.
> * Confirm that nobody has [disabled the Salsify user](https://help.salsify.com/ProductXM/s/article/disable-user) that owns the key. Disabling a user removes their access to Salsify, and their API key stops working with it.
> * Confirm that the **Organization ID** is the URL segment after `/orgs/` (not the organization name), and that it belongs to the same organization as the API key.

> **The connection saves, but a node reports that access is denied**
>
> This can happen if your Salsify user doesn't have permissions to work with the products and lists the Workflow uses. In Salsify, [review the permissions](https://help.salsify.com/ProductXM/s/article/granular-user-group-level-permissions) of Profound's user.

> **A node reports that the Salsify rate limit is reached**
>
> Salsify limits how many requests it accepts in a period of time. The error tells you how many seconds remain until the limit resets. Wait, then run the Workflow again.
>
> To use fewer requests, read more products per page and run fewer Salsify nodes at the same time.

> **Requests to Salsify time out or can't connect**
>
> A timeout is often temporary, so running the node again should fix it. If your Salsify organization only accepts requests from allowlisted IP addresses, ask your Profound account team to route your organization's Salsify requests through a static IP address.


# Using Salsify in Workflows
Source: https://docs.tryprofound.com/integrations/salsify/using-salsify-in-workflows

Learn how to manage products and list memberships with Salsify Workflow nodes

After you [connect your Salsify organization](/integrations/salsify/connect-salsify-to-profound), the following nodes become available in Profound Workflows. Use them to read product content from a Salsify Product List, improve it with other Workflow nodes, write it back, and record which products are done.

Every node runs against the organization you select under **Salsify Organization** and uses the permissions of the Salsify user that owns the API key.

## Available nodes

**List Products**

### List Products

Reads a page of products from one or more Salsify Product Lists. A Product List scope is always required: Profound never reads the whole Salsify organization.

#### Required inputs

* **Salsify Organization** — Select a connected organization from the dropdown
* **Product Lists** — The IDs of the Product Lists to read from

#### Optional inputs

| Input | Default | Description | Limits |
| - | - | - | - |
| **Match** | **OR** | How the Product Lists combine. **OR** reads products in any of the lists, **AND** reads only products in every list. | - |
| **Excluded Lists** | - | The IDs of Product Lists whose members are left out of the result. | - |
| **Property Equality** | - | One or more property and value pairs that a product must match. Select **Add clause**, then enter the case-sensitive property ID and the exact value. | An alternative to **Advanced Filter**, the two can't be used together. A property ID or value can't contain `,` `=` `^` `:` `'` `{` `}` or `*`. To match those characters, use **Advanced Filter** instead. |
| **Raw Filter** (under **Query options**) | - | Edit the whole filter as JSON, including keys the fields above don't show. | - |
| **Advanced Filter** (under **Query options**) | - | A raw [Salsify filter expression](https://developers.salsify.com/reference/salsify-filtering-language-syntax) that narrows the lists. | An alternative to **Property Equality**, the two can't be used together. Can't contain an equals sign (`=`). |
| **Page** (under **Query options**) | `1` | Which page to read. | Page multiplied by results per page can't exceed `10,000`. |
| **Results per Page** (under **Query options**) | `100` | Products per page. | `100` is the maximum. |

#### Output

A JSON object with a `data` array of products and a metadata object with the pagination details and the total number of matching products.

> [!TIP]
>
> Combine **Product Lists** with **Excluded Lists** to [build a work queue](#build-a-work-queue). Read the source list and exclude the destination list, so each run returns only the products that aren't done yet.

**Get Product**

### Get Product

Reads every property of one product.

#### Required inputs

* **Salsify Organization** — Select a connected organization from the dropdown
* **Product ID** — Your organization's product ID, often the SKU. Case-sensitive. Enter it as text or use a variable from a previous node's output

#### Output

A JSON object with the product details under `data`. Localizable properties come back as objects keyed by locale, for example `"Entity Name": { "en-US": "Java Monster Energy Drink, 12-Pack, 15 Oz Cans", "fr-FR": ... }`.

> [!TIP]
>
> Run this node once during setup. The output shows the property IDs your organization uses and which of them are localizable.

**Create Products**

### Create Products

Creates up to 100 products in one run. Only creates the product records and doesn't add them to Product Lists.

#### Required inputs

* **Salsify Organization** — Select a connected organization from the dropdown
* **Properties** — The properties to include in the new product record. Enter a **Sample product ID** to use an existing product as a template: list the properties that product carries, then pick the ones you need for the new product.
* **Data format**:
  * **Structured** — Shows one card per product with a field for each property.
  * **Raw JSON** — Takes an array of objects that map property IDs to values. Accepts a variable from an earlier node's output.
* **Products** — Up to 100 products. Every product must include your organization's product ID property, often the SKU, with a unique value.

#### Output

When the products are created, the node returns a JSON object with the count of created products. The node doesn't return the created products' data, use [**Get Product**](#get-product) to read it.

**Update Products**

### Update Products

Updates up to 100 existing products in one run. Only the properties you supply change, and every other property stays as is.

#### Required inputs

* **Salsify Organization** — Select a connected organization from the dropdown
* **Properties** — Product properties to update. Enter a **Sample product ID** to use an existing product as a template: list the properties that product carries, then pick the ones you need to add to the product being updated.
* **Data format**:
  * **Structured** — Shows one card per product with a field for each property.
  * **Raw JSON** — Takes an array of objects that map property IDs to values. Can accept variables from earlier nodes' outputs.
* **Products** — Up to 100 products. Each product ID may appear once and is case-sensitive. Leave the property empty to remove the property in Salsify.

#### Output

The node returns a JSON object with the count and the IDs of the products it attempted to update. The updated products are listed under `product_ids`, and the products Salsify has no record of are listed under `not_found`.

> [!NOTE]
>
> This node can change every property except the product's ID.

**Add Products to List**

### Add Products to List

Adds up to 100 existing products to a Salsify Product List.

#### Required inputs

* **Salsify Organization** — Select a connected organization from the dropdown
* **Product List ID** — The ID of the destination list, from the Salsify URL after `product_lists/` and before any query parameters (`?`)
* **Product IDs** — A list of product IDs or a variable that holds the list

#### Output

When the products are added, the node returns a JSON object with the count of added products, their IDs, and the destination Product List ID.

> [!NOTE]
>
> Salsify applies the changes asynchronously, so a [**List Products**](#list-products) node that runs immediately after this node may not show the change.

**Remove Products from List**

### Remove Products from List

Removes up to 100 products from a Salsify Product List. The change applies to list membership only, the product records stay in Salsify.

#### Required inputs

* **Salsify Organization** — Select a connected organization from the dropdown
* **Product List ID** — The ID of the list, from the Salsify URL after `product_lists/` and before any query parameters (`?`)
* **Product IDs** — A list of product IDs or a variable that holds the list

#### Output

When the products are removed, the node returns a JSON object with the count of removed products, their IDs, and the Product List ID.

> [!NOTE]
>
> Salsify applies the changes asynchronously, so a [**List Products**](#list-products) node that runs immediately after this node may not show the change.

***

## Build a work queue

The Salsify nodes are designed to run as a loop: each run takes a batch of products from a source list, improves the content, writes it back to Salsify, and moves the finished products to a destination list. This work queue lets one Workflow improve a whole product catalog one batch at a time.

Here's an example of a work queue.

**1. Create a source list and a destination list**

In Salsify, create two Product Lists, for example `Profound: to optimize` and `Profound: optimized`. Add the products you want to work on to the `to optimize` list and leave the `optimized` list empty.

**2. Read the source list with List Products**

Add a [**List Products**](#list-products) node. Enter the `to optimize` list ID under **Product Lists** and the `optimized` list ID under **Excluded Lists**, so a product that's already finished is never read again.

**3. Improve the content**

Pass the products to your content nodes (such as a [Prompt LLM](https://help.tryprofound.com/articles/1457177722-prompt-llm)) to rewrite the marketing copy on each product record.

**4. Write the content back with Update Products**

Add an [**Update Products**](#update-products) node. For each product, set **Product ID** and the properties to update with the new copy.

**5. Record the finished products with Add Products to List**

Add an [**Add Products to List**](#add-products-to-list) node with the `optimized` list ID and the product IDs from this run. This Product List in Salsify is now your progress report.

**6. Clear the products from the queue with Remove Products from List**

Add a [**Remove Products from List**](#remove-products-from-list) node with the `to optimize` list ID and the same product IDs. The products leave the queue but stay in Salsify.

## Troubleshooting

> **The property picker is empty or disabled**
>
> * Select a **Salsify Organization** first. The picker stays disabled until you do.
> * Enter a **Sample product ID**. The picker lists the properties of that one product, so use a product that carries the properties you want to write.
> * If the picker reads **Available when the payload below is a list of products**, the **Products** payload is text or a variable. Switch **Data format** to **Raw JSON** to edit it.

> **List Products fails with a Salsify error**
>
> * Confirm that the Product List ID is the URL segment after `product_lists/` and before any query parameters (`?`).
> * Confirm that the list still exists in Salsify.
> * If you use **Advanced Filter**, check that it's a valid [Salsify filter expression](https://developers.salsify.com/reference/salsify-filtering-language-syntax).

> **List Products returns no products from a list that has products**
>
> * Confirm the Product List ID. Two lists can have similar names.
> * Check that a **Property Equality** clause or an **Advanced Filter** doesn't exclude every product.
> * If you added the products to the list a moment ago, wait and run the node again. Salsify applies changes asynchronously, and your [**List Products**](#list-products) node may not have picked up the change yet.

> **A write node reports that properties don't exist**
>
> Property IDs are case-sensitive. Copy the exact ID from the property picker or from a [**Get Product**](#get-product) node output. If the property is new, create it in Salsify first.

> **Create Products reports that a product must name the product ID property**
>
> A product in the batch is missing the property that holds your organization's product ID. Add it, with a unique value, to every product in the batch.

> **Update Products reports products as not found**
>
> Salsify has no record of those product IDs. Read the IDs with a [**List Products**](#list-products) node and confirm they're correct.

> **A read is refused for going past the 10,000-result limit**
>
> Reduce **Page** or **Results per Page**, or scope the read to a smaller Product List. Alternatively, [build a work queue](#build-a-work-queue), which always reads the first page only.

> **A change made in Salsify doesn't appear in the next read**
>
> Salsify updates list membership asynchronously. Wait, then run the node again.

> **A node reports that the Salsify rate limit is reached**
>
> The error tells you how many seconds remain until the limit resets. Wait, then run the Workflow again. To avoid reaching rate limits, read more products per page and run fewer Salsify nodes at the same time.


# Connect Sanity to Profound
Source: https://docs.tryprofound.com/integrations/sanity/sanity



## Prerequisites

Before connecting Sanity to Profound, ensure:

* You have a [Sanity.io](https://www.sanity.io) account with an active project
* You have **Editor** or **Administrator** permissions on the project
* You have generated an API token with write access

## Generate a Sanity API Token

1. Go to your [Sanity project management console](https://www.sanity.io/manage).
2. Select your project.
3. Navigate to **API → Tokens**.
4. Click **Add API token**.
5. Give the token a descriptive name (for example, "Profound Integration").
6. Set the permissions to **Editor** (minimum required).
7. Click **Save** and copy the token.

**Important:** Copy the token immediately, it won't be shown again. Store it securely.

## Find Your Project ID

Your Project ID is displayed at the top of your project's management console, or in the URL: `https://www.sanity.io/manage/project/<project-id>`.

## Connect Your Account

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **Sanity**, select it, then select **+ Add account**.
4. In the **Connect Sanity** dialog, enter your **API Token** and **Project ID**.
5. Select **Connect account**.

Once connected, Sanity will appear as an available integration and can be selected inside Profound Workflows.

> [!NOTE]
>
> Profound requires a token with **Editor** permissions at minimum. Tokens with read-only (Viewer) access will not work for creating or updating documents.


# Using Sanity in Workflows
Source: https://docs.tryprofound.com/integrations/sanity/using-sanity-in-workflows



After connecting your Sanity account, the following actions become available as Workflow steps. Each step requires you to select a connected **Sanity Project** from a dropdown.

**Create resource**

#### **Create Resource**

Create a new document in your Sanity dataset.

**Required inputs**

* **Sanity Project** — Select the Sanity project you connected in Profound
* **Dataset** — Select the dataset (e.g., `production`)
* **Document Details** — A JSON object with your document fields (must include `_type`)

**Optional inputs**

* **Rich Text Content** — A JSON object mapping field names to HTML strings. Profound automatically converts HTML to Sanity's Portable Text format
* **Publish Resource** — Check this to publish immediately, or leave unchecked to create a draft

By default, new documents are created as **drafts**. Profound automatically generates a draft ID prefixed with `drafts.` for you. Enable **Publish Resource** to make the document live immediately.

**Tip:** Leave **Publish Resource** unchecked to create drafts for review in Sanity Studio before publishing.

**Update resource**

#### **Update Resource**

Update an existing document in your Sanity dataset.

**Required inputs**

* **Sanity Project** — Select the Sanity project you connected in Profound
* **Dataset** — Select the dataset
* **Resource ID** — The Sanity document ID to update
* **Update Details** — A JSON object with the fields you want to change

**Optional inputs**

* **Rich Text Content** — A JSON object mapping field names to HTML strings, auto-converted to Portable Text

**Important:** Only the fields you provide will be modified; all other fields remain unchanged. The document's `_type` cannot be changed after creation.

**Get resource**

#### **Get Resource**

Retrieve a single document by ID.

**Required inputs**

* **Sanity Project** — Select the Sanity project you connected in Profound
* **Dataset** — Select the dataset
* **Resource ID** — The Sanity document ID to fetch

The output includes the document's full content with all fields and metadata. Both published and draft documents can be retrieved (drafts use the `drafts.` prefix).

**List resources**

#### **List Resources**

Retrieve a list of documents from your Sanity dataset.

**Required inputs**

* **Sanity Project** — Select the Sanity project you connected in Profound
* **Dataset** — Select the dataset

**Optional inputs**

* **Document Type** — Filter by a specific Sanity document type, or leave as "All Types"
* **Include drafts?** — Uses the raw perspective to include both published and draft documents
* **Page** — Page number to retrieve (defaults to 1)
* **Per Page** — Number of items per page (defaults to 100)

The output is a structured list of documents ordered by creation date (newest first) that can be used in downstream Workflow steps.

**Publish resource**

#### **Publish Resource**

Publish a draft document to make it live.

**Required inputs**

* **Sanity Project** — Select the Sanity project you connected in Profound
* **Dataset** — Select the dataset
* **Draft ID** — The draft document ID to publish (must start with `drafts.`)

Publishing replaces the published version with the draft content and removes the draft document. This mirrors the publish behavior in Sanity Studio.

***

## Key Concepts

### Documents and Types

Every document in Sanity has a `_type` field that determines its schema. Sanity types are the equivalent of "content models" or "templates": they define the structure and fields a document can have.

When creating a document with the **Create Resource** step, the `_type` field in your data determines which schema the document follows. The type must already exist in your Sanity project's schema.

Common examples of document types:

| Type | Example Use |
| - | - |
| `post` | Blog articles |
| `page` | Static pages |
| `product` | E-commerce products |
| `author` | Content creators |
| `category` | Taxonomy/classification |

**Tip:** To see which types exist in your project, use the **List Resources** step without a type filter: the returned documents will show the available `_type` values.

### Datasets

Sanity projects can have multiple datasets (for example, `production`, `staging`, `development`). Each dataset is an independent content store. Always specify the correct dataset when configuring your Workflow steps.

### Drafts vs. Published

Sanity uses a dual-ID system for content versioning:

* **Published documents** have a plain ID (for example, `abc123`)
* **Draft documents** have the same ID prefixed with `drafts.` (for example, `drafts.abc123`)

When you create a resource without enabling **Publish Resource**, Profound automatically creates it as a draft. You can later publish it using the **Publish Resource** step, or publish directly from Sanity Studio.

***

## Working with Portable Text (Rich Text)

Sanity uses [Portable Text](https://www.sanity.io/docs/block-content) as its native rich text format. Portable Text is a JSON-based specification that represents structured content as an array of blocks.

### Providing Rich Text Content

You have two options for providing rich text content to Sanity fields:

**HTML input (recommended)**

Use the **Rich Text Content** field to pass HTML strings. Profound automatically converts them to Portable Text blocks.

Map field names to HTML strings:

```json theme={null}
{
  "body": "<h1>Welcome</h1><p>This is a <strong>blog post</strong> with <em>formatting</em>.</p><ul><li>First item</li><li>Second item</li></ul>"
}
```

**Supported HTML elements:**

| Element | Portable Text equivalent |
| - | - |
| `<h1>` through `<h6>` | Heading styles |
| `<p>` | Normal paragraph |
| `<strong>`, `<b>` | Bold mark |
| `<em>`, `<i>` | Italic mark |
| `<u>` | Underline mark |
| `<code>` | Code mark |
| `<a href="...">` | Link annotation |
| `<ul>`, `<ol>`, `<li>` | Bullet and numbered lists |
| `<blockquote>` | Blockquote style |
| `<br>` | Line break |

**Tip:** This is the easiest way to get rich text into Sanity. Generate HTML from an LLM step and pass it directly to Rich Text Content. Profound handles the conversion.

**Native Portable Text JSON**

If you already have Portable Text JSON (for example, from a **Get Resource** step), you can pass it directly in the **Document Details** field:

```json theme={null}
{
  "_type": "post",
  "title": "My Post",
  "body": [
    {
      "_type": "block",
      "_key": "abc123",
      "style": "normal",
      "markDefs": [],
      "children": [
        {
          "_type": "span",
          "_key": "def456",
          "marks": [],
          "text": "Hello world"
        }
      ]
    }
  ]
}
```

Each block requires:

* `_type`: Always `"block"` for text blocks
* `_key`: A unique string identifier
* `style`: Block style (`"normal"`, `"h1"` through `"h6"`, `"blockquote"`)
* `markDefs`: Array of complex mark definitions (e.g., links with `href`)
* `children`: Array of spans with `_type`, `_key`, `text`, and `marks`

Formatting marks include: `"strong"`, `"em"`, `"underline"`, `"code"`.

For list items, add:

* `listItem`: `"bullet"` or `"number"`
* `level`: Nesting level (integer)

***

## Mapping Sanity Schemas to Workflow Steps

If you have an existing Sanity schema (document type), here's how to map it to a **Create Resource** step in your Workflow.

### Example: Blog Post Schema

Suppose your Sanity Studio has a `post` type with these fields:

```js theme={null}
// In your Sanity Studio schema
defineType({
  name: 'post',
  title: 'Post',
  type: 'document',
  fields: [
    defineField({ name: 'title', type: 'string' }),
    defineField({ name: 'slug', type: 'slug' }),
    defineField({ name: 'author', type: 'reference', to: [{ type: 'author' }] }),
    defineField({ name: 'body', type: 'blockContent' }),
    defineField({ name: 'publishedAt', type: 'datetime' }),
    defineField({ name: 'categories', type: 'array', of: [{ type: 'reference', to: [{ type: 'category' }] }] }),
  ],
})
```

### Step-by-Step Setup

1. **Add a Create Resource step** to your Workflow and select your connected Sanity project from the **Sanity Project** dropdown.

2. **Select the Dataset** from the dropdown (for example, `production`).

3. **Fill in Document Details** with your document fields as JSON. This is where you provide simple values like titles, slugs, references, and categories:

   ```json theme={null}
   {
     "_type": "post",
     "title": "My Post Title",
     "slug": {
       "_type": "slug",
       "current": "my-post-title"
     },
     "author": {
       "_type": "reference",
       "_ref": "author-id-123"
     },
     "publishedAt": "2025-01-15T10:30:00Z",
     "categories": [
       {
         "_type": "reference",
         "_ref": "category-id-456"
       }
     ]
   }
   ```

   To use dynamic values from upstream steps, type `/` inside the field to open the variable menu and select a variable. It will appear as a pill badge inline with your JSON. For example, you could replace `"My Post Title"` with a variable from a previous LLM step.

4. **Fill in Rich Text Content** for any Portable Text fields (like a blog body). Map the field name to an HTML string. Profound converts it to Sanity's format automatically:

   ```json theme={null}
   {
     "body": "<h1>Welcome</h1><p>This is my post content.</p>"
   }
   ```

   You can also insert a variable here to pass in HTML generated by a previous step. Type `/` to open the variable menu and select the output from an LLM or other upstream step as the value.

5. **Check or uncheck Publish Resource** — check it to publish immediately, or leave it unchecked to create a draft for review in Sanity Studio.

**Tip:** Any text field in the node settings supports **variables**: type `/` to open the variable menu and insert outputs from upstream steps. Variables appear as interactive pill badges and are resolved at runtime.

### Field Type Reference

| Sanity Field Type | JSON Format in Data |
| - | - |
| `string` | `"value"` |
| `text` | `"multi-line value"` |
| `number` | `42` |
| `boolean` | `true` / `false` |
| `datetime` | `"2025-01-15T10:30:00Z"` |
| `date` | `"2025-01-15"` |
| `url` | `"https://example.com"` |
| `slug` | `{ "_type": "slug", "current": "my-slug" }` |
| `reference` | `{ "_type": "reference", "_ref": "document-id" }` |
| `image` | `{ "_type": "image", "asset": { "_type": "reference", "_ref": "image-asset-id" } }` |
| `blockContent` | Use **Rich Text Content** field for automatic conversion, or pass Portable Text JSON directly |
| `array` of references | `[{ "_type": "reference", "_ref": "id" }]` |
| `object` | `{ "fieldA": "value", "fieldB": 42 }` |

***

## Common Workflows

### Generate and Publish a Blog Post

1. **LLM Step** — Generate a title, slug, and HTML body from a prompt
2. **Create Resource** — Create a Sanity `post` document using the generated content
3. **Publish Resource** — Publish the draft (or set `publish_resource: true` in step 2)

### Update Existing Content

1. **Get Resource** — Fetch the current document by ID
2. **LLM Step** — Generate updated content based on the existing document
3. **Update Resource** — Patch only the fields that changed

### Content Migration

1. **List Resources** — Fetch documents from one dataset
2. **Loop** — Iterate over the list
3. **Create Resource** — Create each document in a different dataset

### Draft Review Pipeline

1. **Create Resource** — Create content as a draft (default behavior)
2. **Human Review** — Manual review step in Sanity Studio
3. **Publish Resource** — Publish approved drafts via the Workflow

***

## Permissions and Roles

Profound validates that your API token has sufficient permissions on connection. The following Sanity roles are supported:

| Role | Can Create | Can Update | Can Publish | Can Read |
| - | - | - | - | - |
| **Administrator** | Yes | Yes | Yes | Yes |
| **Editor** | Yes | Yes | Yes | Yes |
| **Developer** | Yes | Yes | Yes | Yes |
| **Viewer** | No | No | No | Yes |

**Minimum required role:** Editor. Tokens with Viewer-only access will fail when attempting to create, update, or publish documents.

***

## Troubleshooting

> **Error: Invalid Sanity API key or token has expired**
>
> Your API token is invalid or has expired. Generate a new token in your [Sanity management console](https://www.sanity.io/manage) under **API → Tokens** and reconnect.

> **Error: Insufficient permissions**
>
> Your API token doesn't have Editor-level access. Create a new token with **Editor** permissions and reconnect.

> **Error: Sanity project not found**
>
> The Project ID in your integration settings is incorrect. Verify the Project ID in your [Sanity management console](https://www.sanity.io/manage). It's displayed at the top of your project page and in the URL.

> **Error: Document type (\_type) must be provided**
>
> When using **Create Resource**, your Data field must include a `_type` property that matches a document type defined in your Sanity schema. For example: `{ "_type": "post", "title": "Hello" }`.

> **Error: Resource ID is not a draft**
>
> The **Publish Resource** step requires a draft document ID that starts with `drafts.`. If the document is already published, it doesn't need to be published again. Check that you're passing the draft ID, not the published ID.

> **HTML not converting to rich text correctly**
>
> Make sure you're using the **Rich Text Content** field (not the **Document Details** field) for HTML-to-Portable-Text conversion. The Rich Text Content field expects a JSON object mapping field names to HTML strings: `{ "body": "<p>Hello</p>" }`.

***

## Additional Resources

Learn more about Sanity's content model and API:

* [Sanity Content Lake](https://www.sanity.io/docs/content-lake) — Understanding datasets and documents
* [Portable Text](https://www.sanity.io/docs/block-content) — Sanity's rich text format specification
* [GROQ Query Language](https://www.sanity.io/docs/groq) — Querying content in Sanity
* [HTTP API Reference](https://www.sanity.io/docs/http-api) — Full Sanity API documentation
* [Mutations API](https://www.sanity.io/docs/http-mutations) — Creating, updating, and deleting documents


# Connect Slack to Profound
Source: https://docs.tryprofound.com/integrations/slack/connect-slack-to-profound

Install the Profound app in your Slack workspace, choose what it can do, and start using Aim in Slack

The Profound app for Slack brings Aim into your workspace. Ask about your brand's AI visibility, citations, and competitors in a channel or a direct message, and get answers, files, and links to Profound documents right in the thread. The same connection can also send Profound Workflows' results to Slack channels, start automations from Slack messages and reactions, and deliver Profound notifications privately.

## How Profound works in Slack

* **Ask in a channel.** Mention **@Profound** in a public or private channel it's a member of. Profound replies in a thread, and replies in that thread continue the conversation without another mention.
* **Ask privately.** Send Profound a direct message, or open the Profound app's **Messages** tab and pick a suggested prompt.
* **Get results where you work.** Profound reacts with 👀 while it works and ✅ when it's done. It uploads files it generates to the thread, links documents it creates in Profound, and asks before it takes an action that needs your approval.
* **Deliver and trigger.** Workflows can post their results to channels you choose, and messages or reactions in channels you choose can start automations.
* **Notify.** Profound can send private notifications to people whose Slack account matches their Profound account.

## Before you start

You'll need:

* A [Profound](https://www.tryprofound.com/) account with the admin role. Only admins can connect accounts and manage channels.
* Permission to install apps in your Slack workspace. If your workspace [requires approval for apps](https://slack.com/help/articles/222386767-Manage-app-approval-for-your-workspace), a Slack admin needs to approve Profound.

## Setup guide

To locate the Slack integration in the Profound platform:

* In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
* In the left sidebar, select **Integrations** under **Account**.
* Search for **Slack** and select it.

**1. Start the connection**

In the Slack integration view, navigate to the **Integration** tab and select **Add account**.

**2. Choose what Profound can do**

In the **Choose Slack permissions** dialog, select one or more purposes. The list under the purposes updates to show every Slack permission they need.

| Purpose | What it sets up |
| - | - |
| **Workspace bot** | Conversations with Aim in public channels, private channels, and direct messages, plus suggested prompts, file analysis, generated file uploads, the organization and category commands, status reactions, Profound document previews, channel management, private notifications, and your workspace icon in Profound |
| **Workflow messages** | Workflows can send messages to public and private channels you choose |
| **Automation triggers** | Messages and reactions in channels you choose can start automations |
| **Notifications** | Private Profound notifications for people whose Slack account matches their Profound account |

[Permissions and security](/integrations/slack/slack-permissions-and-security) lists each permission and why Profound needs it. When you're ready, select **Continue**.

> [!NOTE]
>
> When you connect Slack from a Workflow's Slack node, the purpose that node needs is already selected.

**3. Approve Profound in Slack**

Slack opens its authorization page. Confirm that it shows the workspace you want to connect, review the permissions, and select **Allow**.

**4. Verify the connection**

Slack returns you to Profound. If the workspace appears in the list of connected accounts, Profound is installed and ready to use.

If the account shows **Connected—setup incomplete**, Slack didn't grant every permission a selected purpose needs. Select **Retry setup** to authorize again with the same purposes.

> [!NOTE]
>
> A Slack workspace connects to Profound once, and its purposes can't be changed after setup finishes. To use different purposes, [revoke the connection](#revoking-access-and-uninstalling) and connect the workspace again.

## After you connect

**1. Add Profound to channels**

On the **Integration** tab, under **Slack channels**, select **Add AI Marketer to channel** and search for a public channel. To add Aim to a private channel, tag **@Profound** in the channel.

**2. Link your Slack account**

Profound matches each person to their own Profound account, so answers only use data that person can access in Profound.

If the email on your Slack profile matches your Profound account, Profound recognizes you automatically. Otherwise, the first time you message Profound, it sends you a private message with a **Link account** button. Select it, then select **Verify with Slack** on the **Link your Slack identity** page. The link expires after 15 minutes.

To review or revoke your linked identity, open the Slack integration in Profound and navigate to the **User identity** tab.

**3. Start a conversation**

Mention **@Profound** in a channel or send it a direct message. The **Messages** tab suggests prompts to start with:

* Summarize my brand's AI visibility
* Which sources get cited most?
* Compare me to a competitor

**4. Choose an organization and category**

If you can use more than one Profound organization or category, Profound asks which ones to use in this workspace. Pick them in the message Profound sends. To check or reset your choice later, use these commands in Slack:

* `/profound-org status` or `/profound-org reset`
* `/profound-category status` or `/profound-category reset`

You can also set a **Default category** on the **User identity** tab of the Slack integration in Profound.

## Using Profound in Slack

* **Threads:** Anyone who can see a thread with Profound can reply to it, and Profound answers every reply with the Profound access of the person who started the thread. Start conversations about restricted data in a direct message or a private channel.
* **Files you share:** Attach up to 3 files to a message, 10 MB each and 20 MB in total. Profound reads CSV, TXT, Markdown, PNG, JPG, WebP, PDF, XLSX, DOCX, and PPTX files, including files posted earlier in the thread.
* **Files Profound creates:** Profound uploads generated files to the thread, up to 10 files of 10 MB each per reply. Documents and apps it creates in Profound are listed under **Created in Profound** with a link.
* **Stopping and approvals:** Select **Stop** to end a response. When an action needs your approval, Profound posts **Approve** and **Deny** buttons.
* **Document previews:** Links to Profound documents on `platform.tryprofound.com` expand into a preview. Documents shared only with specific people show as **Private Profound Doc**.

Note that Aim may make mistakes. Please double-check responses.

## Revoking access and uninstalling

To disconnect a workspace, open the Slack integration in Profound, open the **...** action menu on the connected account, and select **Revoke**. Profound deletes its stored Slack token and stops responding in the workspace. When no other Profound connection uses the workspace, Profound also asks Slack to revoke the token.

You can also [remove the Profound app](https://slack.com/help/articles/360003125231-Remove-apps-and-custom-integrations-from-your-workspace) from your workspace in Slack. If you do, also revoke the connection in Profound so Profound deletes the stored token.

To learn what Profound keeps after you disconnect and how to request deletion, see [What Profound stores](/integrations/slack/slack-permissions-and-security#what-profound-stores).

## Troubleshooting

> **The account shows Connected—setup incomplete**
>
> Slack didn't grant every permission a selected purpose needs. The dialog lists the purposes that are missing permissions. Select **Retry setup** to authorize again with the same purposes, and approve every permission Slack shows.

> **Profound says the workspace is already connected**
>
> Each Slack workspace connects to Profound once, and the existing connection keeps its token and purposes. To use different purposes, revoke the existing connection and connect the workspace again.

> **Profound doesn't reply**
>
> * Check that the connection includes the **Workspace bot** purpose. Connections with only Workflow messages, Automation triggers, or Notifications don't answer conversations.
> * In a channel, check that Profound is a member and that you mentioned **@Profound**.
> * If Profound sent you a **Link account** message, finish linking your account first.

> **The account link expired**
>
> Links expire after 15 minutes. Request a new one from the private message Profound sent you, or send **link account** to Profound in its **Messages** tab.

> **A Workflow's Slack node can't use the workspace**
>
> A Workflow that sends messages needs a connection with the **Workflow messages** purpose, and a Workflow started by messages or reactions needs **Automation triggers**. If the workspace was connected without the purpose the node needs, revoke the connection and connect again with every purpose you use.

## Support and privacy

* For help, email [support@tryprofound.com](mailto:support@tryprofound.com).
* Read Profound's [Privacy Policy](https://www.tryprofound.com/legal/privacy-policy) and visit the [Trust Center](https://trust.tryprofound.com/).


# Permissions and security
Source: https://docs.tryprofound.com/integrations/slack/slack-permissions-and-security

Slack permissions Profound requests, why it needs each one, and how it stores and deletes your Slack data

The Profound app for Slack only requests the permissions the purposes you select need. This page explains how the connection works, each Slack permission Profound can request and why, and how Profound stores and deletes Slack data.

## Connection model

* **One installation per workspace.** A Profound admin connects the workspace through Slack's standard [OAuth installation flow](https://docs.slack.dev/authentication/installing-with-oauth/). Profound acts in Slack as the **Profound** app, never as the person who installed it.
* **Bot permissions only.** Profound requests bot scopes. It doesn't request user scopes, so it can't read or post as any person in your workspace.
* **Each person uses their own Profound access.** Profound matches the person messaging it to their Profound account, either because their Slack email matches their Profound account or because they verified their Slack identity with [Sign in with Slack](https://docs.slack.dev/authentication/sign-in-with-slack/). Answers only use data that account can access in Profound.
* **Threads run as the person who started them.** Anyone who can see a thread with Profound can reply, and Profound answers every reply with the access of the person who started the thread.
* **Profound only acts on messages meant for it.** In channels where Profound is a member, Slack sends Profound new messages. Profound responds to messages that mention it and to replies in threads it's part of. It passes other channel messages on only when the connection includes the **Automation triggers** purpose, to start the automations you set up.

## Purposes and features

When you connect Slack, you select one or more purposes in the **Choose Slack permissions** dialog. Each purpose turns on a set of features, and each feature needs specific Slack permissions.

| Purpose | Features |
| - | - |
| **Workspace bot** | Public channel conversations, private channel conversations, direct conversations, assistant suggestions, analyze shared attachments, receive generated files, organization and category commands, processing status reactions, document previews, channel management, private notifications, workspace icon |
| **Workflow messages** | Public channel destinations, private channel destinations, workflow message delivery, channel management |
| **Automation triggers** | Public channel destinations, private channel destinations, message triggers, reaction triggers, channel management |
| **Notifications** | Private notifications |

If Slack doesn't grant a permission a purpose's main features need, the connection shows **Connected—setup incomplete** and those features stay off until you finish setup.

## Slack permissions

| Scope | Features that use it | How Profound uses it |
| - | - | - |
| `app_mentions:read` | Public and private channel conversations | Receives messages that mention **@Profound** so it can reply in a thread |
| `channels:history` | Public channel conversations, message triggers | Reads replies in public-channel threads Profound is part of, so a conversation continues without another mention. With message triggers, receives public-channel messages that start automations |
| `groups:history` | Private channel conversations, message triggers | The same as `channels:history`, for private channels Profound is a member of |
| `im:history` | Direct conversations | Reads the messages you send Profound in a direct message or its **Messages** tab |
| `channels:read` | Public channel conversations and destinations | Lists public channels so you can choose destinations and so Profound can route conversations |
| `groups:read` | Private channel conversations and destinations | Lists the private channels Profound is a member of |
| `chat:write` | Conversations, workflow message delivery, private notifications | Posts Profound's replies, Workflow results, and notifications |
| `im:write` | Conversations, commands, private notifications | Opens a direct message to send your account-link button and private notifications |
| `users:read` | Conversations, commands, private notifications | Looks up the Slack profile of the person messaging Profound, and the people a notification is for |
| `users:read.email` | Conversations, commands, private notifications | Matches Slack members to their Profound accounts by email |
| `assistant:write` | Assistant suggestions | Shows suggested prompts in the Profound app's **Messages** tab |
| `files:read` | Analyze shared attachments | Downloads files shared in a conversation with Profound so it can read them |
| `files:write` | Receive generated files | Uploads files Profound generates to the conversation |
| `commands` | Organization and category commands | Adds the `/profound-org` and `/profound-category` commands |
| `reactions:write` | Processing status reactions | Adds 👀 while Profound works on a message and ✅ when it's done |
| `reactions:read` | Reaction triggers | Receives reactions in chosen channels that start automations |
| `links:read` | Document previews | Detects links to Profound documents on `platform.tryprofound.com` |
| `links:write` | Document previews | Shows a preview of those Profound documents |
| `channels:join` | Channel management | Joins a public channel when an admin adds AI Marketer (Aim) to it in Profound |
| `channels:manage` | Channel management | Leaves a public channel when an admin removes Aim from it in Profound |
| `groups:write` | Channel management | Leaves a private channel when an admin removes Aim from it in Profound |
| `team:read` | Workspace icon | Shows your workspace's name and icon next to the connection in Profound |

Profound doesn't request `im:read`, any group direct message scope (`mpim:*`), or any user scope.

> [!NOTE]
>
> Slack adds new permissions to a workspace's existing Profound installation instead of replacing them. If your workspace connected Profound before, Slack may still hold permissions granted then. Learn more in Slack's [OAuth installation documentation](https://docs.slack.dev/authentication/installing-with-oauth/).

## What Profound stores

Profound stores:

* the workspace's bot token, encrypted as described in [Token storage](#token-storage)
* the connection's workspace ID, name, and icon, the purposes and features you selected, and the permissions Slack granted
* for each conversation thread, its workspace, channel, and thread IDs, the Profound organization and categories it uses, and the Profound user who started it. These records don't contain message text
* account links between Slack members and Profound accounts, and each person's default organization and category
* the messages you send Profound, the files you share with it, and its replies, as Profound conversation history in your organization
* records of the messages Profound posted, used to avoid posting a reply twice. Profound deletes these after 30 days

Revoking the connection deletes the bot token. It doesn't delete conversation history, thread records, or account links. You can revoke your own account link on the **User identity** tab of the Slack integration in Profound. To request access to or deletion of other Slack data, email [support@tryprofound.com](mailto:support@tryprofound.com). Profound's [Privacy Policy](https://www.tryprofound.com/legal/privacy-policy) explains how long Profound keeps personal data and how to make a request.

## Token storage

Profound envelope-encrypts the bot token with an [AWS Key Management Service (KMS)](https://docs.aws.amazon.com/kms/latest/developerguide/overview.html) key reserved for customer data and stores the ciphertext in Amazon DynamoDB. The encryption context binds each record to one Profound organization and provider, so a record can't be decrypted under another organization's identity. Profound's main database holds only a reference to the encrypted record, never the token.

## Revoking access

Admins revoke the connection in Profound under **Settings** > **Integrations** > **Slack**, from the **...** action menu on the connected account, by selecting **Revoke**. When you revoke:

* Profound deletes the encrypted token and marks the connection as revoked. The deletion isn't reversible.
* If no other Profound connection uses the same workspace, Profound calls Slack's [`auth.revoke`](https://docs.slack.dev/reference/methods/auth.revoke/) to revoke the token.
* Profound stops responding in the workspace, and Workflows and automations that use the connection stop working until you connect again.

> [!NOTE]
>
> Slack workspace admins can also [remove the Profound app](https://slack.com/help/articles/360003125231-Remove-apps-and-custom-integrations-from-your-workspace) in Slack. Slack then stops accepting Profound's token. Revoke the connection in Profound too, so Profound deletes the stored token.

## Controls for Slack admins

* **Approve or block Profound.** Use [app approval](https://slack.com/help/articles/222386767-Manage-app-approval-for-your-workspace) to decide whether members can install Profound in your workspace.
* **Choose where Profound can read.** Profound only reads channels it's a member of. Remove it from a channel in Slack or in Profound's Slack settings at any time.
* **Remove the app.** Removing Profound from your workspace ends its access immediately.


# Connect Wordpress to Profound
Source: https://docs.tryprofound.com/integrations/wordpress/connect-wordpress-to-profound



1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings**.
2. In the left sidebar, select **Integrations** under **Account**.
3. Search for **WordPress**, select it, then select **+ Add account**.
4. In the **Connect WordPress** dialog, enter:
   * **Username** (your WordPress login username)
   * **Application Password** (the application password you generated)
   * **Site URL** (for example, [https://example.com](https://example.com))
5. Select **Connect account**.

Once connected, the WordPress site will appear as an available integration and can be selected inside Profound Workflows. Multiple WordPress sites can be connected to the same Profound organization.


# Create an Application Password in WordPress
Source: https://docs.tryprofound.com/integrations/wordpress/create-an-application-password-in-word-press



1. Log in to your WordPress admin panel.
2. Navigate to **Users → Profile** (or **Users → Edit** for the relevant admin user).
3. Scroll to **Application Passwords**.
4. Create a new application password (for example, “Profound Integration”).
5. Copy the generated password immediately. It will not be shown again.

**Important Notes**

* Use your actual WordPress **username**, not the application password label.
* Application Passwords require WordPress 5.6 or later.
* Treat the application password like a standard credential: it grants REST API access to your site.


# WordPress
Source: https://docs.tryprofound.com/integrations/wordpress/overview

Let Workflows create, update, retrieve, and list posts on a connected WordPress site

The WordPress integration allows Profound Workflows to programmatically create, update, retrieve, and list posts on a connected WordPress site. This enables fully automated content pipelines, such as generating, transforming, and publishing articles, directly from Profound Workflows.

## Setup & Authentication

### Step 1: Create an application password in WordPress

1. Log in to your WordPress admin panel.
2. Navigate to **Users > Profile** (or **Users > Edit** for the relevant admin user).
3. Scroll to **Application Passwords**.
4. Create a new application password (for example, "Profound Integration").
5. Copy the generated password immediately. It won't be shown again.

**Important notes**

* Use your **actual WordPress username**, not the application password label.
* Application Passwords require **WordPress 5.6 or later**.
* Treat the application password like a **standard credential**. It grants REST API access to your site.

### Step 2: Connect WordPress to Profound

1. In Profound, select your organization name in the top-left corner of the platform, then select **Settings > Integrations > WordPress**.
2. Select **+ Add account**.
3. Enter your **Username** (your WordPress login username), **Password** (the application password you generated), and **Site URL** (for example, [https://example.com](https://example.com)).
4. Click **Connect account** to finish.

Once connected, the WordPress site will appear as an available integration and can be selected inside Workflows. Multiple WordPress sites can be connected to the same Profound organization.

## Using WordPress in Workflows

After connecting a site, WordPress actions become available as Workflow steps. Each step requires you to select a connected WordPress site from a dropdown.

### 1. Create post

Create a new post on a connected WordPress site.

**Required inputs:**

* WordPress site
* Title
* Content

**Optional inputs:**

* Status (for example, Draft or Publish)
* Excerpt
* Slug

Use this step after content generation or transformation steps to draft or publish new articles.

#### 2. Update post

Update an existing post.

**Required inputs:**

* WordPress site
* Post ID

**Optional inputs:**

* Status
* Title
* Content
* Excerpt
* Slug

This step is useful for publishing drafts, revising existing content, or programmatically updating posts created earlier in the Workflow.

> [!WARNING]
>
> Only the fields you provide will be modified. All other fields remain unchanged.

#### 3. List posts

Retrieve a list of posts from a connected WordPress site.

**Required inputs:**

* WordPress site

**Optional inputs:**

* Page (pagination)
* Per page (number of posts to return)
* Post status (for example, **Publish**)

The output is a structured list of posts that can be used in downstream Workflow steps, for example to select a post to update.

#### 4. Get post

Retrieve a single post by ID.

Required inputs:

* WordPress site
* Post ID

The output includes the post's data and can be referenced by later steps in the Workflow.

## Tips and troubleshooting

1. If you don't see **Application Passwords** in WordPress, confirm your WordPress version and check whether a security plugin is disabling the feature.
2. If you lose the application password, generate a new one in WordPress and reconnect the integration in Profound.
3. Ensure the Site URL exactly matches how your WordPress site is accessed, including https and any subdirectory if applicable.


# Using WordPress in Workflows
Source: https://docs.tryprofound.com/integrations/wordpress/setup-and-authentication



After connecting a site, WordPress actions become available as Workflow steps. Each step requires you to select a connected **WordPress Site** from a dropdown.

**Create post**

#### **Create Post**

Create a new post on a connected WordPress site.

**Required inputs**

* **WordPress Site**
* **Title**
* **Content**

**Optional inputs**

* **Status** (e.g., Draft or Publish)
* **Excerpt**
* **Slug**

This step is commonly used after content generation or transformation steps to draft or publish new articles.

**Update post**

#### **Update Post**

Update an existing post.

**Required inputs**

* **WordPress Site**
* **Post ID**

**Optional inputs**

* **Status**
* **Title**
* **Content**
* **Excerpt**
* **Slug**

This is useful for publishing drafts, revising existing content, or programmatically updating posts created earlier in the Workflow.

**Important Notes:** Only the fields you provide will be modified; all other fields remain unchanged.

**List posts**

#### **List Posts**

Retrieve a list of posts from a connected WordPress site.

**Required inputs**

* **WordPress Site**

**Optional inputs**

* **Page** (pagination)
* **Per Page** (number of posts to return)
* **Post Status** (e.g., Publish)

The output is a structured list of posts that can be used in downstream Workflow steps (for example, selecting a post to update).

**Get post**

#### **Get Post**

Retrieve a single post by ID.

**Required inputs**

* **WordPress Site**
* **Post ID**

The output includes the post's data and can be referenced by later steps in the Workflow.


# Tips & Troubleshooting
Source: https://docs.tryprofound.com/integrations/wordpress/tips-and-troublshooting



* If you do not see **Application Passwords** in WordPress, confirm your WordPress version and check whether a security plugin is disabling the feature.
* If you lose the application password, generate a new one in WordPress and reconnect the integration in Profound.
* Ensure the Site URL exactly matches how your WordPress site is accessed (including https and any subdirectory, if applicable).


# About Activity Logs
Source: https://docs.tryprofound.com/platform-config/activity-logs/about

Review the audit trail of events in your Profound organization

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen
>
> **Required role**: Admin or custom

Activity Logs give admins a single record of the important events in their Profound organization. Review each event with the responsible user and a timestamp, then filter, count, and export the results without leaving the platform.

## When organization activity goes untracked

IT and security teams are accountable for every tool their organization runs on. Without an audit trail, questions like "who invited this user?" or "when was that role changed?" have no reliable answer. Compliance reviews stall, troubleshooting turns into guesswork, and unexplained changes erode trust.

## A single audit trail for your organization

Activity Logs record high-impact events across your organization and attribute each one to the person who performed it. Instead of reconstructing events from memory or support tickets, admins filter the logs to the events that matter, confirm what happened, and export the evidence for compliance or internal review.

## No setup required

Activity Logs live in **Settings**, under the **Organization** section, and are available to users with an Admin role. No setup is required: open the page and your organization's event history is already there.

> [!TIP]
>
> Learn more about roles and access in [Roles and permissions documentation](/platform-config/people/roles-and-permissions).

## Key features

### Coverage of high-impact events

Activity Logs capture organization, login, and membership changes, such as role updates. Each entry records the action, the user who performed it, the IP address the change came from, the category the action belongs to, and the time it happened.

### Logical change records

Activity Logs record events as logical changes, so a bulk edit appears as one entry rather than hundreds of individual rows.

### Filters with live counts

Narrow the logs by the events you care about and watch result counts update as you filter. Live counts show how often an event occurred without scrolling through the full history.

### Export

Export the filtered logs for compliance reviews, security audits, or record-keeping. The export matches your active filters, so you share what you see.

## Retention

Profound retains the full history of Activity Logs. There is no retention limit.


# Activity Logs FAQ
Source: https://docs.tryprofound.com/platform-config/activity-logs/faq

Frequently asked questions about Activity Logs

> **Who can access Activity Logs?**
>
> The logs contain sensitive information about your organization, so access is limited to Admin users.

> **Which events do Activity Logs capture?**
>
> Activity Logs record high-impact, user-triggered events. View the full list in the [Recorded actions reference](/platform-config/activity-logs/recorded-actions).

> **Do Activity Logs include API or MCP activity?**
>
> No. Activity Logs record events from the Profound platform. Requests made through the API, the MCP server, or elsewhere outside the platform don't appear in the logs.

> **Why do some entries name Profound as the actor?**
>
> Some actions are performed by Profound on your organization's behalf, for example, during support work. The log attributes these to Profound so you can distinguish them from actions taken by your team.

> **Why am I not seeing any events prior to July 24, 2026?**
>
> Activity Logs started collecting events on July 24, 2026, so earlier events aren't available.

> **Can I connect logged changes to visibility results?**
>
> Yes. Activity Logs record each change with a timestamp. Line up the changes (such as Agent runs) against your visibility trends in Answer Engine Insights to understand the connection.


# Get started with Activity Logs
Source: https://docs.tryprofound.com/platform-config/activity-logs/getting-started

Review, filter, and export your organization's Activity Logs for the first time

Activity Logs are available to users with an Admin role. No setup is required: once you open the page, your organization's event history is already there.

> [!TIP]
>
> View or revise members' roles on the **People** page under the **Organization** section. See [Add and manage organization members](/platform-config/people/add-and-manage-org-members).

Here's how to review and export your organization's activity for the first time.

## Step 1: Open Activity Logs

In Profound, select your organization name at the top left of the platform, then select **Settings**.

![Profound platform landing page screenshot with the organization name menu expanded and the Settings option highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/activity-logs/getting-started-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=e67d2aa51f37d23d0b03ac60f398dcaf)

Select **Activity Logs** under **Organization** in the left navigation sidebar.

![Settings sidebar with Activity Logs highlighted next to the Activity Logs table and filters](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/activity-logs/getting-started-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=c8676e28e78ef5a0331f2ee4b94a5314)

## Step 2: Filter the logs

Use the date, action, and other filters above the logs table to view the events you care about.

![Activity Logs filter bar highlighted with the date range picker open showing preset ranges and a calendar](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/activity-logs/getting-started-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=ead8c7f10e155d1f3f6161f20426ac87)

## Step 3: Export the results

Export the logs to share them with your security team or keep it for compliance records:

* Select **Export** on the top right corner of the logs table.

![Activity Logs page with the Export button highlighted above the filtered log table](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/activity-logs/getting-started-4.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=37eb137c526822781888e56affef453a)

* Select the columns you want to export, toggle **Apply current filters** to export filtered records, and select **Export**.

![Export Activity Logs dialog with column checkboxes, Apply current filters toggle, and Export CSV highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/activity-logs/getting-started-5.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=5a4802f715cf80aafdc5846571da7b98)


# Recorded actions reference
Source: https://docs.tryprofound.com/platform-config/activity-logs/recorded-actions

The events Activity Logs record in your Profound organization

| Category | Action |
| - | - |
| Login | Member login |
| Membership | Member joined the organization |
| | Member added to the organization |
| | Member removed from the organization |
| | Member role updated |
| | Invite created |
| | Invite deleted |
| Organization | SSO connection added |
| | SSO connection disabled |
| | API key created |
| | API key revoked |
| | OAuth 2.0 client created |
| | OAuth 2.0 client revoked |
| | OAuth 2.0 client secret rotated |
| | Category deleted |
| | Category disabled |
| | Category enabled |
| | Pixel created Beta |
| | Pixel activated or deactivated Beta |
| | Pixel allowed domains updated Beta |
| | Pixel ad account updated Beta |
| Organization - Agency | Workspace created |
| | Pitch workspace extended |
| | Workspace moved to another plan |
| | Pitch workspace promoted to client workspace |
| | Client workspace reactivated |
| Agents | Agent created |
| | Agent published |
| | Agent run |
| | Agent deleted |
| | Agent duplicated |
| Prompts | Prompts added |
| | Prompts edited |
| | Prompts enabled |
| | Prompts disabled |
| | Prompts deleted |
| | Prompts restored |
| | Topics added |
| | Topics edited |
| | Topics enabled |
| | Topics disabled |
| | Topics deleted |
| | Topics restored |
| | Personas added |
| | Personas edited |
| | Personas deleted |


# Analytics Setup
Source: https://docs.tryprofound.com/platform-config/analytics-setup

Manage the categories, prompts, assets, regions, languages, and platforms your organization tracks

> [!NOTE]
>
> **Available on:** Enterprise plan (non-agency)
>
> **Find it in:** Settings in the organization name menu at the top left of the screen
>
> **Required role:** Admin or custom

Analytics Setup is where Admins view and manage how your organization tracks AI visibility: the categories you monitor, the owned and competitor assets in each category, how many prompts each category has, and the regions and languages you monitor in.

Analytics Setup applies to your entire organization. For category-level changes such as adding a competitor or renaming a category, use the [Answer Engine Insights settings](/answer-engine-insights/settings).

## Categories

The Categories table lists every category with its owned domain, competitors, prompts allocated, regions, and languages. Search the table to find a category, or use the row menu to **Open**, **Duplicate**, or **Delete** it.

![Categories table with the Search categories field and a row's three-dot menu highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/analytics-setup-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=40f7bbed0c2ea01db7fbc888d81cb53d)

### Create a category

Select **+ Add category** and follow the dialog to set up:

* **Category details**: name, prompts to allocate, regions, and languages. [Regions and languages](/platform-config/analytics-setup#regions-and-languages) come from your organization's lists.
* **Owned assets**: your brand or product, with its name and website. Add more than one if needed.
* **Competitors**: the brands you want to compare against, each with a name and website.

> [!WARNING]
>
> You can't rename an asset after saving it, so check the spelling before you select **Save**.

### Edit a category

Open a category and select **Edit category** to change its name, prompts, regions, or languages.

To add assets, select **+ Add asset**. To edit an asset, select the edit icon in its table row.

![AI Visibility category page with Edit Category, Add Asset, and the asset edit icon highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/analytics-setup-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=9da71253fd4064204925140f9dbd954b)

### Shared categories

Categories shared with your organization appear in the table as read-only. Only the category owner can edit them.

## Prompts

**Prompts Allocated** is the number of prompts a category can use out of your organization's total. When you create or edit a category, the form shows how many prompts remain to allocate. Moving prompts between categories doesn't affect your total prompt limits.

![AI Visibility category page with the Prompts Allocated and Prompts Used cards highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/analytics-setup-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=c8a9dac8f62bd516796461699a6a0bdf)

**Prompts Used** is the number of prompts the category has used out of its allocated prompts. To add or edit the prompts themselves, select Modify prompts. This opens the [Prompt Designer](/answer-engine-insights/prompt-designer) for that category.

## Assets

Assets are the brands or products you track in a category. Each asset is either **Owned** (your brand or product) or a **Competitor**.

Add and edit assets on the [category page](/platform-config/analytics-setup#edit-a-category). Each category must keep at least one owned asset enabled.

## Regions and languages

Regions are the geographic areas you monitor your categories in. Languages are the languages your prompts run in.

Both work in two ways:

* **Organization lists**: the **Regions** and **Languages** tables on the Analytics Setup page list every region and language available to your organization, when each was added, and how many categories use it. The counter shows how many regions you use against your limit.
* **Per category**: when you create or edit a category, choose its regions and languages from the organization list. Pick the ones relevant to your audience for the best results.

To add a region to your organization's lists, select **+ Add region**, pick a **Region type** (**Country**, **State / Province**, or **City**), and search for the region. To add a language, select **+ Add language**, then pick a language and a country.

![Regions and Languages tables with the Add region and Add language buttons highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/analytics-setup-4.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=08419685fed7b2458e650aaff4e7b58e)

> [!WARNING]
>
> Deleting a region or a language from the organization list removes it from every category in your organization.

## Platforms

**Platforms Purchased** section on the Analytics Setup page lists the answer engines your contract covers. On a category page, the Platforms section shows which of those engines the category runs on.

Platforms are view-only. To change them for a category, contact your Profound team.

## Limits

Your contract with Profound sets your category, prompt, region, and language limits. When you reach a limit, Analytics Setup lets you know. Deleting a category, region, or language frees up capacity.


# Analytics Setup FAQ
Source: https://docs.tryprofound.com/platform-config/analytics-setup-faq

Frequently asked questions about Analytics Setup

> **Why can't I edit a category?**
>
> It's likely owned by another organization. Shared categories are read-only, and only the owner can edit them.

> **Can I change which platforms a category runs on?**
>
> Not in Analytics Setup. Contact your Profound team to add or remove platforms.

> **How do I get more categories, prompts, regions, or languages?**
>
> Your limits come from your contract with Profound. Contact your engagement manager to change them.

> **What is the difference between Analytics Setup and the settings in Answer Engine Insights?**
>
> Analytics Setup is for Admins and covers your whole organization: categories, prompt allocation, assets, regions, languages, and platforms. The settings in Answer Engine Insights cover one category and let you add or remove competitors and rename the category. Learn more in [Answer Engine Insights settings](/answer-engine-insights/settings).


# Configure SSO for a Profound organization
Source: https://docs.tryprofound.com/platform-config/authentication/configure-sso

Set up Single Sign-On with Profound over SAML or OIDC

Set up Single Sign-On (SSO) with Profound over Security Assertion Markup Language (SAML) or OpenID Connect (OIDC). Profound supports various identity providers (IdPs), including Microsoft Azure AD, Google Workspace, Okta, and any custom SAML-compliant or OIDC-compliant IdP.

> [!NOTE]
>
> **QUICK FACTS**
>
> * SSO setup is **self-serve**. Anyone can complete it and pass the values to their IT team.
> * You set up SSO in the configuration portal, which **takes you outside Profound**.
> * **Exit the setup and come back anytime**. The values you entered or generated stay in place, and your team keeps signing in as usual until you enable SSO in the last step of the setup guide.
> * **Profound doesn't publish fixed SSO connection values** such as the ACS URL, Entity ID, or Redirect URI, and you don't need to request them. The portal generates them for your organization as you follow the [setup steps](#setup-steps).

## Before you start

Before you follow the setup steps, read the [SSO FAQ](/platform-config/authentication/sso-faq) to make sure you select the right setup options.

When you're ready to start the setup, check that you have:

* The [Admin role](/platform-config/people/roles-and-permissions) in your Profound organization
* Administrative access to your IdP
* The domain you want to enable for SSO
* An account in your IdP on the SSO domain, for running the test sign-in
* Access to your DNS records to verify domain ownership

## Setup steps

**1. Open the SSO configuration portal**

1. In Profound, select your organization name at the top left of the screen, then select **Settings**.
2. Select **Enterprise Setup** under **Organization** in the left navigation sidebar. The **Single Sign-On** card shows your current SSO status and verified domains.
3. Select **Configure SSO** to open the configuration portal.

   ![Enterprise Setup page screenshot showing the Single Sign-On card with the Configure SSO button highlighted](https://mintcdn.com/profound-37face47/KA2VQvbNybfiIidf/images/enterprise-connections/sso-configure.png?fit=max\&auto=format\&n=KA2VQvbNybfiIidf\&q=85\&s=7ffcd1116e918e85a10ca092d912ffc3)

**2. Verify your domain**

The portal takes you to the domain verification step. Follow it as described in [Domain verification guide](/platform-config/authentication/sso-domain-verification). The remaining steps become available after the domain is verified.

> [!NOTE]
>
> This step is for adding a DNS record, not for connecting your IdP. You select your IdP, such as Okta or Microsoft Entra ID, in the next step.

Profound checks the email domain of each user who signs in through your IdP against your verified domains, and rejects the sign-in when it doesn't match or when your organization has no verified domains.

After your domain is verified, the **Single Sign-On** card shows the status **In progress** until you enable SSO.

**3. Select your identity provider**

Select your IdP from the list of supported options. After you select it, the portal shows setup instructions specific to your provider.

If your provider isn't on the list, use the custom SAML or OIDC connection option. If you run your own authorization server, follow [Configure SSO with a custom OIDC connection](/platform-config/authentication/sso-custom-oidc).

> [!NOTE]
>
> The SSO configuration portal's Okta option connects to the default Okta authorization server. If you use a custom authorization server in Okta, follow [Configure SSO with a custom OIDC connection](/platform-config/authentication/sso-custom-oidc). If you already started the Okta flow, select **Reset Connection** in the portal to start over.

**4. Follow the IdP-specific instructions**

The portal walks you through connecting your IdP to Profound. It shows the values to copy from Profound into your IdP, and the values to retrieve from your IdP and enter into the portal. Copy the Profound values from the portal. They're generated per SSO connection and don't appear anywhere else in Profound.

The values you exchange depend on your SSO protocol:

**OIDC**

* Profound → IdP: the Authorized Redirect URI
* IdP → Profound: the Discovery Endpoint, Client ID, and Client Secret

**SAML**

* Profound → IdP: the ACS URL, Service Provider Entity ID, and Metadata URL
* IdP → Profound: the SSO URL, Entity ID, and X.509 Certificate

Follow the instructions shown in the portal for your specific IdP and protocol.

**5. Test the connection**

The final step in the portal validates your connection. Select **Test sign-in** to run a real sign-in through your IdP. The portal redirects you to your IdP, you sign in with an account on the domain you verified, and your IdP returns you to the portal with the result:

* **Test successful**: your IdP returned a valid response and the connection is configured correctly.
* **Test failed**: the portal shows the validation error and debugging steps. Repeat [Step 4](#step-4) to correct the connection values, then run the test again.

Run the test as many times as you need. The portal records each attempt in the **Sessions** list on its SSO screen. Select a session to see the request sent to your IdP and the response it returned. These details help when you troubleshoot with your IdP administrator or Profound's [customer support](mailto:support@tryprofound.com).

**Notes**:

* Testing happens in your own Profound organization, a separate test organization isn't needed.
* Running a test sign-in only affects the account you test with.
* Your team keeps signing in as usual until you enable SSO in the next step.

To start over with a different IdP or protocol, select **Reset Connection** in the portal and follow the prompts.

**6. Enable SSO**

After the test passes, enable SSO for your domain directly from the portal. Back on the **Enterprise Setup** page, the status reads **SSO Enabled** and your domain is marked **Verified**. Select **Check status** to refresh.

![Single Sign-On card on the Enterprise Setup page showing the SSO Enabled status, a Verified domain, and the redacted OIDC initiate login URI, with the status, the Verified badge, and the Check status button highlighted](https://mintcdn.com/profound-37face47/CAyR0u4TprZVEFzS/images/enterprise-connections/sso-success.png?fit=max\&auto=format\&n=CAyR0u4TprZVEFzS\&q=85\&s=780bbe5478919330184ec087df729186)

After you enable SSO, Profound automatically directs users whose email addresses match your configured domains to your IdP for authentication.

## Log in directly to your organization from your IdP

For OIDC connections, the **Single Sign-On** card on the **Enterprise Setup** page shows an **OIDC initiate login URI** that is specific to your organization. This helps when your company operates multiple Profound organizations that each have an SSO connection. Your IT admin can add each organization's URI to your IdP, for example as an Okta tile or an equivalent app link, so users log in directly to that organization. Sign-in takes them straight into the platform.

![Single Sign-On card on the Enterprise Setup page with the OIDC initiate login URI highlighted. The URI shows the Profound platform domain, and the rest of the address is redacted](https://mintcdn.com/profound-37face47/CAyR0u4TprZVEFzS/images/enterprise-connections/sso-initiate-login-uri.png?fit=max\&auto=format\&n=CAyR0u4TprZVEFzS\&q=85\&s=0828e305eac43532a2b916a12773c21a)

The URI is currently not available for SAML connections.

## Notes

* Each Profound organization needs its own SSO connection.
* Verifying the root domain doesn't cover its subdomains: verify each subdomain your users sign in with.
* Keep each domain's verification TXT record in place for as long as SSO is enabled for that domain.
* For SAML connections, renew your IdP signing certificate before it expires. The **Single Sign-On** card shows a warning as the expiry date approaches.
* Contact [customer support](mailto:support@tryprofound.com) if you need help during setup.


# Sign-in security and password requirements
Source: https://docs.tryprofound.com/platform-config/authentication/sign-in-security

Password rules for email sign-in, what happens at sign-in when SSO is enabled, and which security controls your identity provider handles

Users sign in to Profound with an email address and password, or through your identity provider (IdP) with [Enterprise Single Sign-On (SSO)](/platform-config/authentication/sso-overview). For password accounts, Profound applies its own rules. For SSO users, multi-factor authentication (MFA), device requirements, and password rules live in the IdP.

## Password requirements

Profound checks the password when a user creates an account, accepts an invitation, resets their password, or changes it in their account settings. A password must:

* Be at least 12 characters long.
* Be [secure](#what-makes-a-password-secure).
* Not match a password the user has used in Profound before.

The same rules apply to every organization. Profound doesn't offer per-organization password policies.

> [!NOTE]
>
> MFA isn't currently available for password sign-in. To require a second factor, [set up Enterprise SSO](/platform-config/authentication/configure-sso) and enforce MFA in your IdP.

### Change or reset a password

To change a password, open **Settings** from the organization name menu at the top left of the screen, then select **Account** > **Profile** and update **Password**.

Profound asks for the current password, and the new password must differ from it. The **Password** row appears only when your current session started with a password. If you signed in through SSO, the row is hidden even if your account has a password.

To reset a forgotten password, select **Forgot password?** on the sign-in page, and Profound emails you a reset link.

### What makes a password secure

A secure password is long and hard to predict. Profound treats the following as easy to guess, even when they meet the length requirement:

* Dictionary words, names, and well-known passwords such as `password1234`.
* Personal details such as a birthday, a username, or part of an email address.
* Keyboard patterns such as `qwertyuiop`, sequences such as `123456789012`, and repeated characters such as `aaaaaaaaaaaa`.
* Simple variations of any of these, such as capitalizing the first letter, swapping letters for similar-looking symbols, or adding a number at the end.

> [!TIP]
>
> Length and unpredictability count for more than mixing character types: A passphrase made of several unrelated words, such as `harbor-violin-cactus-thread`, is stronger than a short password with symbols in it, such as `P@ssw0rd!`.

## Sign-in when SSO is enabled

When a user enters their email address on the sign-in page, Profound checks the sign-in policy for that address before asking for a password:

* **SSO required**: users whose email domain is one of your organization's verified domains go straight to your IdP.
* **SSO optional**: if your organization has an SSO connection but doesn't require it, users choose **Continue with SSO** or **Continue with password**.
* **Other domains**: users whose email domain isn't one of your verified domains, such as agency or contractor accounts, keep signing in with an email address and password. Learn how to provision these users alongside your directory in [Directory Sync](/platform-config/people/directory-sync).

[Enabling SSO](/platform-config/authentication/configure-sso) turns off password sign-in for users on your verified domains, so they go straight to your IdP. To allow password sign-in alongside SSO (for example, during a rollout), contact [customer support](mailto:support@tryprofound.com).

Users who have only ever signed in through SSO have no Profound password. If password sign-in is allowed for them later, they select **Forgot password?** on the sign-in page to set the password.

## Controls your identity provider handles

With Enterprise SSO, your IdP authenticates the user and Profound accepts the result. Profound doesn't configure, enforce, or verify any of the following, so set them up in your IdP:

* MFA.
* Managed-device requirements: use your IdP's conditional access or device trust policies.
* Network or location restrictions: use your IdP's conditional access policies.
* Password rules for SSO users: length, history, expiry, and lockout policies.


# Configure SSO with a custom OIDC connection
Source: https://docs.tryprofound.com/platform-config/authentication/sso-custom-oidc

Connect your own OpenID Connect authorization server to Profound SSO

Set up Single Sign-On (SSO) with Profound through a custom OpenID Connect (OIDC) connection. Use this guide when your identity provider (IdP) isn't in the SSO configuration portal's list, or when your users authenticate through an authorization server you run yourself rather than your IdP's default one.

> [!NOTE]
>
> **QUICK FACTS**
>
> * SSO setup is **self-serve**. Anyone can complete it and pass the values to their IT team.
> * You set up SSO in the configuration portal, which **takes you outside Profound**.
> * **Exit the setup and come back anytime**. The values you entered or generated stay in place, and your team keeps signing in as usual until you enable SSO in the last step of this guide.
> * **Profound doesn't publish fixed SSO connection values** such as the redirect URI, and you don't need to request them. The portal generates them for your organization as you follow the [setup steps](#setup-steps).

## Before you start

Before you follow the setup steps, read the [SSO FAQ](/platform-config/authentication/sso-faq) to make sure you select the right setup options.

When you're ready to start the setup, check that you have:

* The [Admin role](/platform-config/people/roles-and-permissions) in your Profound organization
* Administrative access to your IdP and to the OIDC authorization server your users sign in through
* The domain you want to enable for SSO
* Access to your DNS records to verify domain ownership
* An account in your IdP on the SSO domain, for running the test sign-in. You assign this account to the Profound application in [Step 4](#step-4)

## Setup steps

**1. Open the SSO configuration portal**

1. In Profound, select your organization name at the top left of the screen, then select **Settings**.
2. Select **Enterprise Setup** under **Organization** in the left navigation sidebar. The **Single Sign-On** card shows your current SSO status and verified domains.
3. Select **Configure SSO** to open the configuration portal.

   ![Enterprise Setup page screenshot showing the Single Sign-On card with the Configure SSO button highlighted](https://mintcdn.com/profound-37face47/KA2VQvbNybfiIidf/images/enterprise-connections/sso-configure.png?fit=max\&auto=format\&n=KA2VQvbNybfiIidf\&q=85\&s=7ffcd1116e918e85a10ca092d912ffc3)

**2. Verify your domain**

The portal takes you to the domain verification step. The [Domain verification guide](/platform-config/authentication/sso-domain-verification) walks through this step and its troubleshooting in detail.

After your domain is verified, the **Single Sign-On** card shows the status **In progress** until you enable SSO.

**3. Select the custom OIDC option**

On the identity provider step, select the custom OIDC option instead of a named provider.

The portal switches to the **Custom OIDC** flow, which has its own steps:

1. Provide an Identity Provider Name
2. Create an Application
3. Add Claims
4. Provide your OIDC Configuration
5. Test Single Sign-On

Enter a name for your identity provider and continue.

**4. Create an application in your authorization server**

The portal shows the **Login redirect URI** that your authorization server takes users to after they sign in. It's generated for your connection and doesn't appear anywhere else in Profound. Copy it from the portal.

![Step 2 of the Custom OIDC flow in the configuration portal, Create an Application, listing the requirements for the application and highlighting the Login redirect URI field with its Copy button](https://mintcdn.com/profound-37face47/aTf2hmI2OEek9so_/images/enterprise-connections/sso-custom-oidc-redirect-uri.png?fit=max\&auto=format\&n=aTf2hmI2OEek9so_\&q=85\&s=30505cc437644bef37aed712d227a0fe)

In your authorization server, create a web application for Profound that:

* Uses the OIDC sign-in method.
* Uses the authorization code grant.
* Authenticates with a client ID and client secret.
* Has Profound's **Login redirect URI** as its sign-in redirect URI. Replace any placeholder your IdP pre-fills, such as a `localhost` address.
* Is assigned to the users or groups who sign in to Profound, including the account you plan to test with.

Keep the application's client ID and client secret for [Step 6](#step-6). If you choose to configure your application to use private key JWT instead of a client secret, register your key as the portal instructs and select that authentication method in [Step 6](#step-6).

Go back to the configuration portal and select **Continue**.

> [!NOTE]
>
> If your custom authorization server has its own access policies, make sure one of them allows the application you created for Profound to request tokens.

**5. Add claims to the ID token**

The portal lists the claims that Profound reads from the ID token your authorization server issues. For many providers the required claims are included by default. For others, you need to add them to the ID token in your authorization server's settings.

![Step 3 of the Custom OIDC flow in the configuration portal, Add Claims, listing the required claims to add to the ID token: sub, email, given\_name, and family\_name](https://mintcdn.com/profound-37face47/aTf2hmI2OEek9so_/images/enterprise-connections/sso-custom-oidc-claims.png?fit=max\&auto=format\&n=aTf2hmI2OEek9so_\&q=85\&s=0c741c5de992a575ece1cac38a83fbd5)

| Claim | Required | Description |
| - | - | - |
| `sub` | Yes | The user's stable and unique ID in your IdP |
| `email` | Yes | The user's email address, which Profound checks against your verified domains |
| `given_name` | Yes | The user's first name |
| `family_name` | Yes | The user's last name |

Add the claims, then select **Continue**.

If your authorization server returns these claims only from its [userinfo endpoint](https://openid.net/specs/openid-connect-core-1_0.html#UserInfo) and not in the ID token, continue without adding them to the ID token and turn on **Use userinfo endpoint** in [Step 6](#step-6).

**6. Provide your OIDC configuration**

Enter the values from your authorization server:

* **Discovery endpoint**: the URL of your authorization server's OIDC discovery document. For example, `your-company.okta.com/.../.well-known/openid-configuration`.
* **Client ID** and **Client secret**: the credentials of the application you created in [Step 4](#step-4).

  ![Step 4 of the Custom OIDC flow in the configuration portal, Provide your OIDC Configuration, highlighting the Discovery endpoint, Client ID, and Client secret fields and the collapsed Advanced settings section](https://mintcdn.com/profound-37face47/aTf2hmI2OEek9so_/images/enterprise-connections/sso-custom-oidc-configuration.png?fit=max\&auto=format\&n=aTf2hmI2OEek9so_\&q=85\&s=7aa1227c4e3fba4b698af31792eb1eef)

> [!TIP]
>
> To check that you have a correct discovery endpoint URL before you enter it in the portal, open it in a browser: the correct endpoint returns a JSON document whose `issuer` value matches your authorization server.

Expand **Advanced settings** and check that the values match your application. The defaults work for most authorization servers:

| Setting | Default | When to change it |
| - | - | - |
| **Authentication method** | Client secret basic | Select **Client secret POST** or **Private key JWT** if your application is registered with that method instead. With **Private key JWT**, Profound generates the signing key pair. Register the public key with your authorization server so it can verify Profound's sign-in requests |
| **ID token signing algorithm** | RS256 | Select the algorithm your authorization server signs ID tokens with, if it isn't RS256 |
| **Use userinfo endpoint** | Off | Turn on if your authorization server doesn't include the required claims in the ID token. The portal then reads the user profile from the userinfo endpoint listed in the discovery document |
| **Require PKCE** | On | Keep it on unless your authorization server can't complete the sign-in with Proof Key for Code Exchange (PKCE) |

Select **Continue**. The portal checks that the discovery endpoint is reachable before it lets you continue.

**7. Test single sign-on**

When you're ready to test the connection, select **Continue to sign-in**.

![Step 5 of the Custom OIDC flow in the configuration portal, Test Single Sign-On, with the Continue to sign-in button highlighted next to the I am ready for a test sign-in checkbox](https://mintcdn.com/profound-37face47/aTf2hmI2OEek9so_/images/enterprise-connections/sso-custom-oidc-test.png?fit=max\&auto=format\&n=aTf2hmI2OEek9so_\&q=85\&s=8e67e3bfe8e97717d98730f942801487)

The portal redirects you to your authorization server. Sign in with an account on your verified domain that's assigned to the Profound application.

* If the test succeeds, continue to the next step.
* If the test fails, the portal shows the reason. Select **Show session details** for the session ID, which helps when you troubleshoot with your IdP administrator or Profound's [customer support](mailto:support@tryprofound.com). Correct the configuration, then select **Retry Single Sign-On**.

> [!TIP]
>
> Make sure the account you're using to test the connection is assigned to the Profound application in your IdP.

Run the test as many times as you need. Testing happens in your own Profound organization, and your team keeps signing in as usual until you enable SSO in the next step.

**8. Enable SSO**

After the test passes, enable SSO for your domain directly from the portal.

Back on the **Enterprise Setup** page, the **Single Sign-On** card status reads **SSO Enabled**, shows the **OIDC initiate login URI**, and your domain is marked **Verified**. Select **Check status** to refresh.

![Enterprise Setup page screenshot showing SSO Enabled status, a Verified domain, and the Check status button highlighted](https://mintcdn.com/profound-37face47/CAyR0u4TprZVEFzS/images/enterprise-connections/sso-success.png?fit=max\&auto=format\&n=CAyR0u4TprZVEFzS\&q=85\&s=780bbe5478919330184ec087df729186)

After you enable SSO, Profound automatically directs users whose email addresses match your configured domains to your authorization server for authentication.

## Troubleshooting

> **Discovery endpoint is unreachable.**
>
> Check that the URL you entered is your authorization server's OIDC discovery document. It must:
>
> * Start with your authorization server's issuer URL
> * End with `/.well-known/openid-configuration`
>
> Some users can confuse the discovery endpoint with your authorization server's metadata URL. It looks similar, but it doesn't end with `openid-configuration`.

> **The test sign-in fails with The identity provider denied access.**
>
> Your IdP refused to sign the test user in to the Profound application. Make sure that the account you're using to test the connection is assigned to the application in your IdP. If you use an Okta custom authorization server, also confirm that an access policy on that server applies to the Profound application.

## Notes

* Each Profound organization needs its own SSO connection.
* Verifying the root domain doesn't cover its subdomains: verify each subdomain your users sign in with.
* Keep each domain's verification TXT record in place for as long as SSO is enabled for that domain.
* Contact [customer support](mailto:support@tryprofound.com) if you need help during setup.


# Domain verification for SSO
Source: https://docs.tryprofound.com/platform-config/authentication/sso-domain-verification

Prove that your organization owns a domain by adding a DNS TXT record

Domain verification is a part of the [Single Sign-On (SSO) configuration](/platform-config/authentication/configure-sso) process. It lets your organization claim ownership of a domain, such as `example.com`. After a domain is verified, Profound can trust it for features that need a higher level of security, starting with [SSO](/platform-config/authentication/sso-overview).

Verification is self-serve. You add a single DNS TXT record containing a token that Profound generates for your organization. Profound checks your DNS for the record and marks the domain as **Verified** when it finds it. You don't need to change your website, mail setup, or other DNS records.

## How it works

1. You enter the domain you want to verify in the [SSO configuration portal](/platform-config/authentication/configure-sso#step-1).
2. Profound generates a unique verification token for that domain and shows you the TXT record to add. When it recognizes your DNS provider, it also shows provider-specific instructions.
3. You add the TXT record in your DNS provider's console, alongside any records already on the host.
4. Profound looks up the domain's TXT records. When it finds the token, the domain becomes **Verified** and you can continue with the rest of your SSO configuration.

The token proves ownership because only someone with control over the domain's DNS can publish it. Profound compares the email domain of each user who signs in with SSO against your verified domains, so verification also confirms which users belong to your organization.

## Before you start

Before you follow the steps below, read the [SSO FAQ](/platform-config/authentication/sso-faq) to make sure you select the right setup options.

When you're ready to verify your domain, check that you have:

* The [Admin role](/platform-config/people/roles-and-permissions) in your Profound organization
* Access to the DNS records for the domain you want to verify (or an IT contact who has it)
* The domain that appears in your users' email addresses. For example, if your team signs in with `name@example.com`, verify `example.com`, even if your identity provider runs at `example.myidp.com`

## Verify a domain

**1. Open the SSO configuration portal**

> [!TIP]
>
> You can exit the configuration portal at any time and pick up where you left off later.

1. In Profound, select your organization name at the top left of the screen, then select **Settings**.
2. Select **Enterprise Setup** under **Organization** in the left navigation sidebar. The **Single Sign-On** card lists your domains and their verification status.
3. Select **Add domain**, and the SSO configuration portal opens on the domain verification step.

   ![Enterprise Setup page screenshot showing the Single Sign-On card with the Add domain button highlighted](https://mintcdn.com/profound-37face47/JrExWei_LMQcmC7i/images/enterprise-connections/domain-verification-add-domain.jpg?fit=max\&auto=format\&n=JrExWei_LMQcmC7i\&q=85\&s=76e04c2ad728dcafec7d814922bcad22)

**2. Enter your domain**

Enter the domain you want to verify and select **Continue**.

![Verify your organization domain screen in the portal with the Enter your domain field and Continue button](https://mintcdn.com/profound-37face47/JrExWei_LMQcmC7i/images/enterprise-connections/domain-verification-enter-domain.png?fit=max\&auto=format\&n=JrExWei_LMQcmC7i\&q=85\&s=8360623e76848ee7d2a68cd39629d648)

The portal detects your DNS provider and takes you to the setup instructions specific to it.

![Add DNS records screen in the portal showing provider-specific instructions for Cloudflare, starting with signing in to the Cloudflare dashboard](https://mintcdn.com/profound-37face47/JrExWei_LMQcmC7i/images/enterprise-connections/domain-verification-provider-instructions.png?fit=max\&auto=format\&n=JrExWei_LMQcmC7i\&q=85\&s=14e07351110e1cd38dbd8d3a3128f2e9)

**3. Get the TXT record from the portal**

The portal generates a verification token and shows the TXT record to add: the **Host** (or **Name**) and the **Value** that contains the token.

![TXT record details in the portal: type TXT, host @, and a value beginning with profound-, each with a copy button](https://mintcdn.com/profound-37face47/JrExWei_LMQcmC7i/images/enterprise-connections/domain-verification-txt-record.png?fit=max\&auto=format\&n=JrExWei_LMQcmC7i\&q=85\&s=d89d6c57cf01b24d10e017d39f3249fc)

**4. Add the TXT record in your DNS provider**

Sign in to your DNS provider (for example, Cloudflare, GoDaddy, Route 53, or Squarespace Domains) and create a new TXT record with the host and value shown in the portal. Leave the time to live (TTL) at your provider's default.

Keep any existing TXT records on the same host, such as the Sender Policy Framework (SPF) record, or other verification records from other services. Add the new record next to them instead of replacing them.

If someone else manages your DNS, copy the host and value from the portal and send them to that person. They don't need a Profound account to add the record.

**5. Wait for Profound to detect the record**

Return to the SSO configuration portal and select the option to check the record, or select **Check status** on the **Enterprise Setup** page. When Profound finds the token, the domain status changes to **Verified**.

![Enterprise Setup page screenshot showing the domain marked Verified and the Check status button highlighted](https://mintcdn.com/profound-37face47/JrExWei_LMQcmC7i/images/enterprise-connections/domain-verification-verified.jpg?fit=max\&auto=format\&n=JrExWei_LMQcmC7i\&q=85\&s=3e430530675aa8bd0eceded30d0273a7)

DNS changes usually appear within 15 to 30 minutes and can take up to 48 hours to take effect. If the domain still appears as **Pending** on the Single Sign-On card in Profound, leave the record in place and check back later.

You don't need to keep the configuration portal open while you wait. When you return to the **Enterprise Setup** page and select **Configure SSO**, the portal opens on the step you need next.

After the domain is verified, continue with the [Configure SSO](/platform-config/authentication/configure-sso#step-3) guide to connect your identity provider.

## Verify multiple domains

Each domain gets its own verification token, so repeat the [domain verification steps](#verify-a-domain) for every domain your users sign in with.

Add each TXT record to the domain it was issued for. For example, a record issued for `www.example.com` belongs on `www.example.com`, and `example.com` needs the separate record issued for `example.com`.

Each subdomain needs its own verification. Verifying `example.com` doesn't verify `mail.example.com`. Once verified, a subdomain is covered by your SSO connection [like any other verified domain](/platform-config/authentication/sso-overview#domain-matching-and-subdomain-support).

## Keep the record in place

Keep the verification TXT record in your DNS for as long as you use SSO with that domain. Profound rechecks the record periodically. If the record is removed or changed, the domain returns to **Pending** and sign-in through your identity provider stops working until the record is restored.

## Troubleshooting

> **The portal shows instructions for Cloudflare or another provider that isn't my identity provider.**
>
> This is expected. The portal detects the DNS provider that hosts your domain's records and shows the instructions for adding the TXT record there. This step verifies that your organization owns the domain. It's separate from the identity provider you connect later. Follow the instructions to add the record, and select your identity provider after the domain is verified.

> **The domain stays Pending after adding the record.**
>
> Confirm you added the record as a TXT record on the correct host and that the value matches the token in the portal, including any prefix. Then allow up to 48 hours for DNS propagation and select **Check status** again. If after 48 hours your domain verification is still pending, contact [customer support](mailto:support@tryprofound.com).

> **The TXT record ended up on `example.com.example.com`.**
>
> in many DNS consoles, the Host or Name field is relative to your domain, so entering `example.com` there creates a record for `example.com.example.com`. When the portal asks for a host of `example.com`, enter @ or leave the field blank instead. Both mean the domain itself.

> **You need to verify a domain again.**
>
> If the domain verification token has changed, remove the old TXT record from your DNS provider, and add the new one as instructed in the [domain verification guide](#verify-a-domain) above.


# SSO FAQ
Source: https://docs.tryprofound.com/platform-config/authentication/sso-faq

Frequently asked questions about Single Sign-On (SSO) setup, domains, organizations, and user provisioning

> **We have multiple Profound organizations. How do we set up SSO?**
>
> SSO is configured per organization. Each organization needs its own SSO connection, which means its own verified domain and its own Profound application in your IdP. Users who belong to several organizations typically see one Profound tile per organization in their IdP, and sign in to each organization through its own connection.
>
> You can use the same domain for the SSO connections of several organizations. Before you set up SSO for a second organization on a domain that's already verified for another one, contact [customer support](mailto:support@tryprofound.com): Profound needs to adjust settings on its side first. You can also use different protocols for each connection, even with the same identity provider.
>
> ![Diagram showing that each Profound organization has its own SSO connection: identity provider A connects to organization 1 over OIDC and to organization 2 over SAML, while identity providers B and C each connect to a further organization](https://mintcdn.com/profound-37face47/0oMqQAjfTDxpRjFF/images/enterprise-connections/sso-multiple-organizations.webp?fit=max\&auto=format\&n=0oMqQAjfTDxpRjFF\&q=85\&s=45dd57597fb68fd02c8413a34c0e479c)
>
> To set up each organization's SSO connection, follow the steps in [Configure SSO](/platform-config/authentication/configure-sso) or, if you're using a custom authorization server, see [Configure SSO with a custom OIDC connection](/platform-config/authentication/sso-custom-oidc).
>
> To add people to the correct organization through the correct SSO connection, invite them from the [People page](/platform-config/people/add-and-manage-team-members) in that organization's settings.

> **Where do I get the service provider details, such as the Entity ID, ACS URL, metadata, or redirect URI?**
>
> The SSO configuration portal generates them for your organization. Profound doesn't publish fixed connection values, and you don't need to request them from support.
>
> To get them, open **Settings** > **Organization** > **Enterprise Setup** and select **Configure SSO**. After you verify your domain and select your identity provider (IdP), the portal shows the values to copy into your IdP:
>
> * **Security Assertion Markup Language (SAML)**: the ACS URL, Service Provider Entity ID, and Metadata URL
> * **OpenID Connect (OIDC)**: the Authorized Redirect URI
>
> The values are generated per SSO connection and don't appear anywhere else in Profound. See [Step 4 of Configure SSO](/platform-config/authentication/configure-sso#step-4).

> **Why am I asked to add a DNS TXT record?**
>
> Domain verification and IdP configuration are separate steps. The DNS TXT record proves that your organization owns the email domain your users sign in with. You connect your IdP in the next step, after the domain is verified.
>
> The DNS instructions in the portal name the provider that hosts your domain's DNS records, such as Cloudflare or AWS Route 53. That provider isn't your IdP, and seeing it there doesn't mean your IdP selection is wrong. See [Domain verification for SSO](/platform-config/authentication/sso-domain-verification).

> **Can I verify a subdomain or a separate domain instead of the root domain?**
>
> Verify whichever domain appears after the @ in your users' email addresses, whether that's a root domain or a subdomain. If your team signs in with `name@example.com`, verify `example.com`. If they sign in with `name@mail.example.com`, verify `mail.example.com`. A different domain, such as a separate challenge domain set up only for the verification, can't stand in for it.
>
> Verifying a root domain doesn't cover its subdomains, and verifying a subdomain doesn't cover the root domain, so verify each domain your users sign in with. Once verified, your SSO connection covers it like any other verified domain. See [Verify multiple domains](/platform-config/authentication/sso-domain-verification#verify-multiple-domains).

> **Can we attach multiple email domains to one SSO connection?**
>
> Yes. Domains are verified for your organization, and your organization's SSO connection covers every verified domain. Verify each domain your users sign in with, and Profound directs users on any of them to your IdP.
>
> To add a domain, open **Settings** > **Organization** > **Enterprise Setup** and select **Add domain** on the **Single Sign-On** card. The portal opens on the domain verification step. Each domain gets its own TXT record, so add every record to the domain it was issued for. See [Verify multiple domains](/platform-config/authentication/sso-domain-verification#verify-multiple-domains).

> **Once SSO is on, can users outside our IdP, such as agencies or contractors, still sign in with email and password?**
>
> It depends on their email domain.
>
> * Users whose email domain is one of your verified domains go straight to your IdP. Enabling SSO turns off password sign-in for them. To allow password sign-in alongside SSO for these users, for example during a rollout, contact [customer support](mailto:support@tryprofound.com).
> * Users on other domains, such as agency or contractor accounts, keep signing in with an email address and password.
>
> See [Sign-in when SSO is enabled](/platform-config/authentication/sign-in-security#sign-in-when-sso-is-enabled). If you use Directory Sync, learn how to [keep adding external users](/platform-config/people/directory-sync#membership-source) alongside your directory.

> **What happens if our IdP is down? How does our admin get in?**
>
> Users on your verified domains sign in only through your IdP, so they can't sign in to Profound until your IdP is back. There's no separate emergency password for admins.
>
> If the outage lasts, contact [customer support](mailto:support@tryprofound.com). Profound can turn off SSO enforcement for your organization so your users sign in with email and password, and turn it back on when your IdP recovers. Users who have only ever signed in through SSO have no password yet and set one with **Forgot password?** on the sign-in page.

> **Can our IdP groups map to Profound roles or category access?**
>
> Not currently. Roles and category access are managed in Profound. Members provisioned through [Directory Sync](/platform-config/people/directory-sync) start with the Member role, and Admins [change roles](/platform-config/people/roles-and-permissions) and category access in Profound afterwards.

> **Does SSO create user accounts automatically, or do we still invite everyone?**
>
> SSO handles sign-in only. It doesn't add users to your organization on its own, so a user needs to be a member or have a pending invitation before they can sign in through your IdP.
>
> To stop inviting users manually, set up [Directory Sync (SCIM)](/platform-config/people/directory-sync). It provisions everyone your IdP assigns to Profound and removes access when your IdP unassigns or deactivates them. SSO and Directory Sync are independent: enable either one or both. With both enabled, new members sign in directly through your IdP and no invitation email is sent.

> **Is there a test or sandbox organization where our IT team can trial the setup before touching production?**
>
> No. Testing happens in your own Profound organization, and no separate test organization is needed.
>
> The setup is safe to trial in production:
>
> * Your team keeps signing in as usual until you enable SSO in the last step of the setup.
> * The portal's test sign-in only affects the account you test with, and you can run it as many times as you need.
> * You can exit the portal and come back anytime, and you can select **Reset Connection** to start over.

> **SAML or OIDC: which should we use, and can we switch after starting the setup?**
>
> Profound supports both. Use whichever protocol your IdP and IT team prefer: the portal's instructions cover both, and the setup steps are otherwise the same.
>
> To switch protocols or IdPs, select **Reset Connection** in the SSO configuration portal and follow the prompts, then start the IdP step again with the new choice. Do this before you enable SSO, so your team's sign-in isn't affected.
>
> If your users authenticate through an authorization server you run yourself, such as an Okta custom authorization server, use the custom OIDC option. See [Configure SSO with a custom OIDC connection](/platform-config/authentication/sso-custom-oidc).

> **Can we enforce MFA, password policies, or device and network restrictions through SSO?**
>
> Yes, in your IdP. With SSO, your IdP authenticates the user and Profound accepts the result. Profound doesn't configure, enforce, or verify multi-factor authentication (MFA), managed-device requirements, network or location restrictions, or password rules for SSO users, so set them up in your IdP.
>
> MFA isn't available for password sign-in in Profound. See [Controls your identity provider handles](/platform-config/authentication/sign-in-security#controls-your-identity-provider-handles).


# Enterprise Single Sign-On (SSO) overview
Source: https://docs.tryprofound.com/platform-config/authentication/sso-overview

Supported protocols, scope, and how to get started with SSO configuration

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen > Organization > Enterprise Setup
>
> **Required role**: Admin or custom

Enterprise Single Sign-On (SSO) lets users sign in with their corporate identity provider (IdP) credentials, such as Microsoft Entra ID, Okta, or Google Workspace. Profound keeps their user data synchronized with the IdP.

> [!NOTE]
>
> **QUICK FACTS**
>
> * **One Profound organization = one SSO connection**. If your company has multiple Profound organizations, follow the SSO setup guides in this section to configure a separate SSO connection for each of them.
> * **You can use the same domain for multiple SSO connections.** One email domain can back the SSO connections of several organizations.
> * **Read the [SSO FAQ](/platform-config/authentication/sso-faq)** for answers to the most common setup questions, from where to find the connection values to how SSO works with multiple organizations.

## Supported protocols

Profound supports multiple protocols for enterprise SSO, including Security Assertion Markup Language (SAML) and OpenID Connect (OIDC). Setup is tailored to your IdP, and custom SAML and OIDC connections are available if your provider isn't among the commonly supported options or you run your own [OIDC authorization server](/platform-config/authentication/sso-custom-oidc).

## Domain matching and subdomain support

Authenticating with SSO requires the user's email domain to match the exact domain the connection is configured with. For example, a user with the email address `john-smith@tryprofound.com` can use SSO only with the `tryprofound.com` domain.

Verifying a root domain (like `tryprofound.com`) doesn't cover its subdomains (like `dev.tryprofound.com`). Verify each subdomain your users sign in with, and your SSO connection covers it like any other verified domain.

Profound uses your verified domains to confirm that a user signing in belongs to your organization. You verify a domain by adding a DNS TXT record with a token generated by Profound. Learn more in [Domain verification](/platform-config/authentication/sso-domain-verification).

## SSO and Directory Sync

SSO lets your IdP handle how users sign in. To also let it control who has access, with users provisioned and deprovisioned automatically, set up [Directory Sync](/platform-config/people/directory-sync). The features are independent: enable either one or both.

## Security controls

Your IdP decides how users prove who they are. You configure and enforce requirements such as multi-factor authentication (MFA), managed devices, network or location restrictions, and password rules in your IdP, and Profound accepts the result of that sign-in. See [Sign-in security and password requirements](/platform-config/authentication/sign-in-security) for what Profound enforces for password accounts and how sign-in works once SSO is enabled.

## Getting started

Follow the [Configure SSO guide](/platform-config/authentication/configure-sso) to set up SSO for your organization. If you need help, contact [customer support](mailto:support@tryprofound.com).


# About Command Center
Source: https://docs.tryprofound.com/platform-config/command-center/about

A single view of how your team uses Profound, benchmarked against peers

> [!NOTE]
>
> **Available on**: the Enterprise plan, for customers with at least 10 Agent runs
>
> **Find it in**: Settings in the organization name menu at the top left of the screen
>
> **Required role**: Admin or custom

Command Center gives marketing leaders a single view of how their team uses Profound. Open the dashboard to see active users, agent runs, total hours saved, and other metrics benchmarked against peer enterprises in your industry.

## The ROI question every marketing leader gets

Marketing leaders running a team on Profound face a question: **what's the platform actually returning?** The signals are there in the work, but pulling ROI and adoption numbers out of raw usage data by hand is a project of its own.

Without a dedicated view, it's difficult to know who on the team has adopted the platform and how much time it saves. And even when those numbers are present, they're hard to judge in isolation: without a point of reference, you can't tell whether they meet the mark or fall short.

## One dashboard for adoption, ROI, and context

Command Center addresses those issues in one place. Adoption shows up as active users, agent runs, and a leaderboard of your most active users. ROI shows up as hours saved, calculated with manual time inputs you control. Context comes from benchmarking against peers by enterprise size or industry.

Instead of assembling a report from raw data, you read it off the dashboard or export it and bring it straight to leadership.

## How it works

Command Center lives in **Settings**, under the **Organization** section, and is available to users with an [Admin role](/platform-config/people/roles-and-permissions).

The **Overall Performance** section shows the metric cards, including Active Users, Agent Runs, Total Hours Saved, and Projects Tracked. The **Platform Usage** section shows your Power Users leaderboard and a breakdown of Agent Usage by Category.

By default, every metric shows cumulative, all-time data. Narrow to a specific time period when you need to report on a month or a quarter. You can also add a peer comparison to any metric, benchmarking against companies of similar enterprise size or in your industry, or view your numbers on their own.

## Key features

### Hours saved with a transparent breakdown

Total Hours Saved translates your team's Agent runs into time returned to the team. Select **Expand** on the card to open the **Total Hours Saved Breakdown**, which shows time saved by marketing category along with the active Agents, Agent runs, average process time, and the resulting time saved.

### Customizable process times

Each marketing category has a default average process time: the hours the task takes to complete by hand. The default values are in place on initial setup to help you get started. Adjust them anytime to match your team's pace and get the **Time Saved** figure you can trust and defend in front of leadership.

### Power Users leaderboard

The Power Users table ranks people on your team by hours saved and agent runs. Use it to recognize champions, spread their working patterns across the team, and spot where adoption needs a push. Select **View all users** to see the full list, or download the leaderboard to share it.

### Peer benchmarking

Every metric can be benchmarked against peer enterprises of similar size or in your industry, so your numbers are put into context. Benchmarking data is shared on an opt-in basis: you're opted in by default and can opt out of sharing your usage data anytime.

### Export

Select **Export**  to download an image of the dashboard that you can take to leadership or procurement or share in presentations.


# Command Center FAQ
Source: https://docs.tryprofound.com/platform-config/command-center/faq

Frequently asked questions about Command Center

> **Who can see Command Center?**
>
> Any user with an [Admin role](/platform-config/people/roles-and-permissions) can access the Command Center dashboard.

> **How is the Hours Saved metric calculated?**
>
> Time saved is estimated by multiplying the number of agent runs in each marketing category by the configured manual time for that category. Profound uses average manual completion time as the default, and you can edit the values per category. Open the **Total Hours Saved Breakdown** to see the exact numbers behind the total.

> **What happens to the Power Users leaderboard if someone leaves the team?**
>
> Users who have left your organization but have enough agent usage to rank still appear on the leaderboard, tagged as **Former Member**.

> **Where does the benchmarking data come from?**
>
> Benchmarking data comes from an opt-in, anonymized, aggregated pool of Profound users. Sharing benchmarking data is enabled by default.

> **Can I opt out of sharing my platform usage data?**
>
> Yes. To opt out, go to **Settings** in the top right of the **Command Center** view and toggle off the **Enable peer benchmarking** setting. Once you opt out of data sharing, you lose access to the benchmarking feature.
>
> ![Command Center dashboard Settings view screenshot with Enable peer benchmarking toggle highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/command-center/faq-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=265376e3a6860b19252cf7066870fd05)


# Get started with Command Center
Source: https://docs.tryprofound.com/platform-config/command-center/getting-started

Open Command Center and read its adoption and usage metrics

## Step 1: Open Command Center

Select your organization name at the top left of the platform, then select **Settings**.

![Profound platform landing page screenshot with the organization name menu expanded and the Settings option highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/command-center/getting-started-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=7850c97d7e7de313df41ddd9dcf2a389)

Select **Command Center** under **Organization** in the left navigation sidebar.

## Step 2: Review your team’s usage

* Scan your **Overall Performance** graphs: **Active User**s, **Agent Runs**, **Total Hours Saved**, and **Projects Tracked**.
* Then check the **Platform Usage** metrics: **Power Users** leaderboard and **Agent Usage** **by Category.**
* Use the **View** filter to switch between **Total** (metrics for all time) and **Time Frame** (metrics for the set time frame).

## Step 3: Customize your average process time values

* Select **Expand** on the **Total Hours Saved** card.

![Total hours saved card screenshot with the Expand button highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/command-center/getting-started-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=848d35a3c8571be8ff648c735a1cbae3)

* select **Edit Average Process Time** in the **Total Hours Saved Breakdown.**

![Total Hours Saved Breakdown view screenshot with Edit Average Process Time button highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/command-center/getting-started-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=498fd3c0debffa2a46240ee90b33e057)

* Adjust the hours per category so the hours saved figure reflects how long these tasks take your team.

## Step 4: Set your benchmark

Use the **Benchmark** filter to compare your metrics by **Enterprise Size** or **Industry**. Select **Hide** if you want to view your numbers without the comparison.

![Command Center view screenshot with the Benchmark filter expanded and highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/platform-config/command-center/getting-started-4.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=2826004357d7c60bff4d5037987b2d48)

## Step 5: Export and share

Select **Export** to download the dashboard, or download the Power Users ranking (**Power Users** > **View all users**) to share adoption highlights with your team.


# Add and manage organization members
Source: https://docs.tryprofound.com/platform-config/people/add-and-manage-org-members

Invite members to your Profound organization and manage their roles and category access

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen > People
>
> **Required role**: Admin or custom

Admins can add members to their Profound organization from the **Members** tab of the **People** page. Each member gets a role that controls what they can see and do, and category access that controls which categories they can work in. To give a group of members the same categories and shared resources at once, add them to a [team](/platform-config/people/teams/overview).

> [!NOTE]
>
> If your organization uses [Directory Sync](/platform-config/people/directory-sync), you add and remove the synced members through your identity provider instead. [Roles and permissions](/platform-config/people/roles-and-permissions) are always managed in Profound.

## Add a member

1. Select your organization name at the top left of the platform, then select **Settings**.
2. Select **People** under **Organization** in the left navigation sidebar. The **People & Permissions** page opens on the **Members** tab.
3. Select **Add User** or **Add External User** if your organization uses [Directory Sync](/platform-config/people/directory-sync).
4. Enter the new member's email address.
5. Under **Category**, choose the categories the member can work in. Keep **Select all** to grant access to every category, or select individual categories.
6. Select a role. For a breakdown of what each role can do, see [Roles and permissions](/platform-config/people/roles-and-permissions). Invitations offer preset roles only. To give the new member a [custom role](/platform-config/people/custom-roles), change their role after they join.
7. To invite several people at once, select **Add another user** and fill in a row for each person, or [import them from a CSV file](#import-members-from-a-csv-file).
8. Select **Send Invitation**.

The invitation appears in the **Pending** list until the new member signs in. To see it, set the **Invite Status** filter to **Pending**.

### Import members from a CSV file

In the invitation dialog, select **Import from CSV** and choose a file with one person per row: email address, then first name and last name (optional). A header row that contains an **email** column is skipped. Each import adds up to 50 rows. If the file has more, only the first 50 are imported.

Imported people appear as rows in the dialog. Check each row's **Category** and **Role** before you select **Send Invitation**.

### How the invitation is delivered

The new member receives an email from `no-reply@auth.mail.tryprofound.com` with a link to sign up. The link takes them to Profound, where they activate their account.

## Find members

The **Members** tab lists every active member of your organization with their email, role, join date, and last sign-in, 25 members per page. To find a member:

* Search by name, email, or role in the **Search members** field.
* Filter by **Role**. The list includes both pre-defined roles and your organization's custom roles.
* Filter by **Last sign-in**: **Signed in last 7 days**, **Signed in last 30 days**, **Inactive 30+ days**, **Inactive 90+ days**, **Inactive 180+ days**, or **Never signed in**.

To download the list of active members as a CSV file with the same columns, select **Export**.

## Manage existing members

* **Change a role**: Select the role next to the member's name and select a new one from the list. The list includes your organization's [custom roles](/platform-config/people/custom-roles).
* **Change category access**: Open the **...** action menu in the member's table row and select **Configure category access**. Select the categories the member can access, or select **Select all**. Hover over a category and select **only** to keep that category alone. A member needs at least one category. Select **Save**.
* **Remove a member**: Open the **...** action menu in the member's table row and select **Remove from organization**, then confirm with **Remove**. They lose access to your organization, and you can invite them again later.

You can manage only members whose permissions are equal to or lower than yours. You can't change your own role or remove yourself.

## Manage pending invitations

Set the **Invite Status** filter to **Pending** to list invitations that haven't been accepted yet, with the assigned role and the invitation date. To cancel an invitation, open its **...** action menu and select **Revoke**, then confirm with **Revoke Invitation**. You can send a new invitation at any time.


# Custom roles
Source: https://docs.tryprofound.com/platform-config/people/custom-roles

Create custom roles and permission sets and assign them to members in Profound

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen
>
> **Required role**: Admin or custom

Profound comes with a set of [preset roles](/platform-config/people/roles-and-permissions), such as Admin, Editor, and Member. When none of them matches how your team works, create a custom role: pick the permissions, [assign it to members](/platform-config/people/add-and-manage-org-members), and change it later without reassigning anyone.

A member with a custom role gets only the permissions that role grants. The custom role replaces their preset role rather than adding to it.

## Who can manage roles

Members with the **Manage user setup and permissions, including inviting users** permission see the **Roles** tab and can create, edit, and delete custom roles. Among the preset roles, only Admins have it. You can also grant it to a custom role.

## The Roles tab

1. Select your organization name at the top left of the platform, then select **Settings**.
2. Select **People** under **Organization** in the left navigation sidebar.
3. Select the **Roles** tab.

The table lists every role in your organization, newest first. Custom roles carry a **Custom** badge. Use the **Type** filter at the top of the screen to show only **Preset** or **Custom** roles.

Select a role to open it and see which permissions it grants. Preset roles are read-only, but you can duplicate one as the starting point for a custom role.

## Create a custom role

**1. Open the role editor**

On the **Roles** tab, select **Create Role**.

To start from an existing role instead, open its **...** action menu and select **Duplicate role**. The editor opens with the source role's permissions selected.

**2. Name and describe the role**

Enter a name in the **Custom role name** field and a **Role Description**. Both are required.

The name has to be unique in your organization. The description appears next to the role in the Roles table and helps other Admins understand what the role is for.

**3. Select permissions**

Under **Permissions**, expand a group and select the actions a member with this role can perform. Use **Select All** or **Clear All** to toggle a whole group. The counter next to each group shows how many of its permissions are selected. Select at least one permission to save the role.

**4. Create the role**

After you select **Create**, the role appears in the **Roles** table and is ready to use. Assign it from the **Members** tab, [the same way as a preset role](/platform-config/people/add-and-manage-org-members#manage-existing-members).

> [!NOTE]
>
> Invitations and [Directory Sync](/platform-config/people/directory-sync) assign preset roles only. Switch the member to a custom role after they join your organization.

## Custom role actions

Each row in the **Roles** table has a **...** action menu:

* **Open**: View the role and its permissions.
* **Edit role**: Change a custom role's name, description, or permissions. The change applies to every member with the role. Not available for preset roles.
* **Duplicate role**: Start a new custom role with this role's permissions preselected.
* **Delete**: Remove a custom role. This can't be undone. Not available for preset roles.


# Directory Sync (SCIM)
Source: https://docs.tryprofound.com/platform-config/people/directory-sync

Automatically add and remove organization members from your identity provider's directory

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen > Organization > Enterprise Setup
>
> **Required role**: Admin or custom

Directory Sync keeps your Profound organization's member list in sync with your identity provider (IdP) using System for Cross-domain Identity Management (SCIM) provisioning.

## How Directory Sync works

Once you connect a directory, it provisions everyone it lists into your Profound organization:

* **Assignments grant access.** When you connect a directory, everyone currently assigned to it is synced into your organization. After that, anyone your IdP assigns to Profound is provisioned automatically.
* **Unassignments remove access.** When your IdP unassigns or deactivates someone, their organization membership is removed and any pending invitation is revoked. Their Profound account itself isn't deleted: if they belong to other Profound organizations, those are unaffected.
* **New members start with the Member role.** People provisioned through Directory Sync get the [Member role](/platform-config/people/roles-and-permissions), and organization Admins can [change roles in Profound](/platform-config/people/add-and-manage-org-members) afterwards. If someone leaves the directory and is later reassigned, they return with the default Member role.
* **Members outside the directory can lose access.** When your IdP directory is the only membership source, any members you added manually on the [People page](/platform-config/people/add-and-manage-org-members) who aren't assigned in your directory lose access to your organization. To keep adding members from both sources, change the [membership source](#membership-source) setting.

### How new members sign in

How a new member signs in for the first time depends on your [Single Sign-On (SSO) setup](/platform-config/authentication/sso-overview):

* **With SSO enabled**, their account is created and they sign in directly through your IdP. No invitation email is sent.
* **Without SSO**, they receive an email invitation to create a password. The invitation stays valid for 30 days.

> [!NOTE]
>
> You don't need SSO to use Directory Sync. They are independent features: enable either one, or both.

## Configure Directory Sync

To configure Directory Sync, you need the [Admin role](/platform-config/people/roles-and-permissions) in your Profound organization and administrative access to your IdP.

**1. Locate the Directory Sync in ettings**

1. In Profound, select your organization name at the top left of the screen, then select **Settings**.
2. Select **Enterprise Setup** under **Organization** in the left navigation sidebar. The **Directory Sync** card shows your current sync status.

**2. Configure your membership source**

Review **Membership source** on the **Directory Sync** card:

* Select the gear  icon next to **Membership source**.
* To keep existing members who aren't in your directory and continue inviting external users, choose **Directory & external users**.
* To allow only members from your IdP directory, select **Directory only**.

> [!WARNING]
>
> With **Directory only**, members who aren't assigned in your directory will lose access as soon as the sync begins. Learn more in [Membership source](#membership-source).

Select **Save** and wait for the setting to save before continuing.

**3. Open the Directory Sync configuration portal and follow the provider-specific instructions**

On the **Directory Sync** card, select **Configure Directory Sync** to open the configuration portal.

The portal tailors the setup to your directory provider, walking you through creating a SCIM integration in your IdP, connecting it to Profound, and choosing which people and groups to sync.

**4. Test and verify**

Use the portal's test step to verify the directory connection. Once syncing is live, the **Directory Sync** card on the **Enterprise Setup** page shows **Directory Sync Enabled** and your current [membership source](#membership-source). Select **Check status** to refresh, or **Manage Directory Sync** to reopen the configuration portal.

## Membership source

The **Membership source** setting governs your Profound organization membership once the directory is connected:

* **Directory only** (the default): your IdP manages all members. Anyone who isn't assigned in your directory loses access to your organization, including members you added manually before connecting the directory. Inviting people in Profound is disabled.
* **Directory & external users**: your IdP still provisions and removes the people it lists, and Admins can also invite people outside your directory in Profound. Use this option when people who aren't in your internal employee directory, such as agency partners or contractors, need a seat in your organization.

In both cases, your directory remains the source of truth for the people it lists. Members you add in Profound are external users: the directory never provisions or removes them, and you manage them on the [People page](/platform-config/people/add-and-manage-org-members).

### Change the membership source

[Admins](/platform-config/people/roles-and-permissions) can change the membership source at any time as instructed in [Step 2](#step-2) of the setup guide.

When a directory is connected, switching from **Directory & external users** to **Directory only** checks for external users. If any are found, the dialog shows how many may external users will lose access once the setting is changed.

> [!NOTE]
>
> Before a directory is connected, the membership source setup dialog doesn't show how many users are affected by the setting change. This does not mean that all existing members will retain access after you connect the directory. With **Directory only** any members outside the directory lose access.

## Manage members with Directory Sync

Manage the members your directory lists through your IdP:

* The **People** page shows a banner explaining that people are managed by your identity provider.
* With the **Directory only** membership source, adding members, removing members, and revoking invitations are disabled in Profound. Make these changes in your IdP's directory instead.
* With the **Directory & external users** membership source, select **Add External User** on the **People** page to invite a user from outside your IdP directory. To invite several people at once, select **Add another user** or **Import from CSV**. Manage external users [from the People page](/platform-config/people/add-and-manage-org-members) in Profound.

Roles and category access are managed in Profound for everyone, regardless of the membership source. See [Add and manage organization members](/platform-config/people/add-and-manage-org-members).

## Admin failsafe

Directory Sync always keeps at least one Admin in your organization. If your directory unassigns the organization's only Admin, that person keeps their access and role. This prevents a misconfiguration in your IdP, such as an incorrect group assignment, from locking everyone out of administering the organization.

To remove your organization's only Admin, first grant the Admin role to someone else. Once another Admin exists, the directory's unassignment takes effect normally.

## Notes

* Directory Sync is configured per Profound organization: each organization needs its own Directory Sync connection and membership source.
* Contact [customer support](mailto:support@tryprofound.com) if you need help during setup.


# Roles and permissions
Source: https://docs.tryprofound.com/platform-config/people/roles-and-permissions

Learn about roles and permissions for organization members in Profound

Profound uses role-based access control. Every member of your organization has a role that determines which features they can see and what actions they can take.

This page describes the preset roles that every organization has. If none of them fits, Admins can [create a custom role](/platform-config/people/custom-roles) with its own set of permissions.

> [!NOTE]
>
> To assign or change a role, see [Add and manage organization members](/platform-config/people/add-and-manage-org-members).

## Roles at a glance

* **Admin**: Full access to every feature, plus organization management, including users and invitations, billing, and integrations. The only role that can change organization settings.
* **Editor**: Full access to product features, including managing prompts, assets, domains, and Knowledge Bases. No access to organization administration, except viewing integrations.
* **Member**: The standard role for day-to-day work. Views all analytics, works with Agents, Sheets, Profound Docs, and Skills, and uses AI Marketer (Aim), but can't change the Answer Engine Insights (AEI) configuration or manage Knowledge Bases.
* **Viewer - AEI Only**: Views AEI data, Shopping, Prompt Volumes, Dashboards, Context Manager, and integrations. Creates, views, and edits Profound Docs, but can't share, export, or delete them. No access to Agents, Sheets, Pages, or Projects.
* **Viewer - AEI + AI Marketer**: Everything the Viewer - AEI Only role has, plus Aim for analytics, Aim in Profound Docs, and the ability to create and edit Skills.
* **Agent Operator**: Runs existing Agents, creates Agents and Sheets from templates, and works with Sheets, Profound Docs, Prompt Volumes, Skills, and Aim for analytics. Views AEI data, Shopping, Agent Analytics, Pages, Projects, Suggestions, and integrations. Creates Dashboards, edits Projects, and runs an Agent in a Project. Can't create new Agents from scratch, edit Agents, or manage Sheets.

## Permissions by role

The tables below list each permission and the roles that have it.

### Organization and platform settings

Organization-level settings are reserved for Admins.

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| Manage user setup and permissions, including inviting users | | - | - | - | - | - |
| Manage billing and usage data | | - | - | - | - | - |
| Manage Agency Workspaces | | - | - | - | - | - |
| View integrations | | | | | | |
| Manage integrations (includes viewing them) | | - | - | - | - | - |
| View Activity Logs | | - | - | - | - | - |
| View Command Center | | - | - | - | - | - |

### Answer Engine Insights

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| View data | | | | | | |
| Manage prompts (includes viewing AEI data) | | | - | - | - | - |
| Manage assets | | | - | - | - | - |
| Manage domains | | | - | - | - | - |
| View chart annotations | | | | | | |
| Create, edit, and delete custom chart annotations (includes viewing them) | | | - | - | - | - |

Every role that can view chart annotations sees prompt-change markers on AEI charts. Roles with **View Activity Logs** access also see who made the change and a link to the Activity Logs.

### Analytics and reporting

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| View Shopping | | | | | | |
| View Dashboards | | | | | | |
| Create Dashboards | | | | - | - | |
| View Pages | | | | - | - | |
| View Agent Analytics | | | | - | - | |
| Manage Agent Analytics (includes View Agent Analytics) | | | - | - | - | - |
| View Prompt Volumes data | | | | | | |
| Manage Prompt Volumes watchlists (includes View Prompt Volumes data) | | | | - | - | |

### Agents and Sheets

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| View Agents | | | | - | - | |
| Edit Agents | | | | - | - | - |
| Create new Agents | | | | - | - | - |
| Create Agents from templates | | | | - | - | |
| Run Agents | | | | - | - | |
| View Sheets | | | | - | - | |
| Run Agents in Sheets | | | | - | - | |
| Create Sheets from templates | | | | - | - | |
| Add Agent columns to Sheets | | | | - | - | - |
| Manage and edit Sheets | | | | - | - | - |

### Profound Docs

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| Create Docs | | | | | | |
| View Docs | | | | | | |
| Edit Docs | | | | | | |
| Share Docs with individuals or the organization | | | | - | - | |
| Export Docs | | | | - | - | |
| Delete Docs | | | | - | - | |
| Use Aim in Docs | | | | - | | |

### Knowledge Bases and Skills

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| View Context Manager and Knowledge Bases | | | | | | |
| Edit and delete Knowledge Bases | | | - | - | - | - |
| Import from Notion or Google Drive into a Knowledge Base | | | - | - | - | - |
| Create and edit Skills | | | | - | | |

### Aim and Projects

| Permission | Admin | Editor | Member | Viewer - AEI Only | Viewer - AEI + AI Marketer | Agent Operator |
| - | :-: | :-: | :-: | :-: | :-: | :-: |
| Chat with Aim | | | | - | | |
| Aim (build Agents) | | | | - | - | - |
| View Projects | | | | - | - | |
| Edit Projects | | | | - | - | |
| Run an Agent in a Project | | | | - | - | |
| View Suggestions from Aim | | | | - | - | |


# Create and manage teams
Source: https://docs.tryprofound.com/platform-config/people/teams/create-and-manage-teams

Create teams in Profound, add their categories and members, request a team wallet, and share resources with a team

Teams group members so they share the same categories, resources, and credits. For what teams control and who can manage them, see [How teams work in Profound](/platform-config/people/teams/overview).

## Create a team

1. Select your organization name at the top left of the platform, then select **Settings**.
2. Select **People** under **Organization** in the left navigation sidebar, then select the **Teams** tab.
3. Select **Create team**.
4. Enter a **Team name**, for example "EMEA Marketing". Team names must be unique within your organization.
5. Enter an optional **Team description**.
6. Select **Create team**.

The team's page opens. Add the team's categories and members there.

## Manage teams

The **Teams** tab shows every team in your organization with its members. Switch between grid and list view, search by team name, or filter by **Category**. To open a team, select it, or open its **...** action menu and select **Open**.

On the team's page, select the name or description to edit it in place, or select the team icon to change its color. Everyone in your organization sees the new name.

### Manage a team's categories

* **Add categories**: In the **Categories** section, select **Add category**. Select the categories the team should have access to, or select the checkbox in the header row to select them all, then select **Add category**. The table shows each category's allocated prompts, regions, and languages.
* **Remove a category**: Open the **...** action menu in the category's table row and select **Remove category**, then confirm. Members lose access to the category through this team, but keep it if they have it directly or through another team.

> [!NOTE]
>
> Team access adds to the category access members already have. To limit a member to their team's categories, edit their category access in the **Members** tab on the **People** page. Learn more in [Add and manage organization members](/platform-config/people/add-and-manage-org-members#manage-existing-members).

### Manage a team's members

* **Add members**: In the **Members** section, select **Add member**. Search for and select the users to add, then select **Add users**. Only members of your organization can join a team, so [invite new people to the organization](/platform-config/people/add-and-manage-org-members#add-a-member) first.
* **Remove a member**: Open the **...** action menu in the member's table row and select **Remove member**, then confirm. They lose the categories and shared resources they had through this team.

## Delete a team

In the **Teams** tab, open the team's **...** action menu and select **Delete team**, then confirm. Members lose the categories, shared resources, and wallet they had through the team.

> [!WARNING]
>
> Deleting a team can't be undone. Before you delete it, make sure its members have the access they need directly or through another team.

## Request a team wallet

Profound sets up team wallets. To give a team its own [credits](/getting-started/credits), send your Profound account team the team name and its monthly credit budget. Once the wallet is set up, the team's page shows the credits left and the date the wallet resets. Learn how wallets are used in [Team wallets](/platform-config/people/teams/overview#team-wallets).

## Share a resource with a team

1. Open a Dashboard, Document, Sheet, Agent, or Workflow, then open its sharing options.
2. In the search field, type the team's name and select it from the **Teams** group.
3. Choose the access level, then select **Share**.

The list includes every team in your organization.


# How teams work in Profound
Source: https://docs.tryprofound.com/platform-config/people/teams/overview

Learn what teams are in Profound and how they control category access, shared resources, and credits

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen > People > Teams tab
>
> **Required role**: Admin or custom

A team is a named group of members inside your [organization](/getting-started/organizations-and-teams). Where [roles](/platform-config/people/roles-and-permissions) control what a member can do, such as run Agents or manage integrations, teams control what a member can see and which credits they spend.

Use teams to mirror how your company works, for example one team per business unit, brand, or region. Each team can have its own [categories](/getting-started/categories#category-structure-and-how-it-relates-to-other-profound-building-blocks), its own shared resources, and its own [credit](/getting-started/credits) wallet. To set one up, see [Create and manage teams](/platform-config/people/teams/create-and-manage-teams).

> [!NOTE]
>
> Teams are optional. They help organize access and resources within your organization, but if everyone in your organization works on the same categories, you may not need teams.

## What problem teams solve

Large companies often have several groups working in Profound at once, each with its own brands, budget, and private material. Without teams, keeping those groups apart takes workarounds:

* **Separate organizations for each group.** Data stays apart, but reporting can't roll up across organizations, and people switch between organizations to manage their work.
* **One shared credit pool.** Any group can use up credits that another group needs.
* **Access set one person at a time.** An admin grants category access to each new member individually.

With teams, you keep everyone in one organization and draw the lines inside it. Each group sees its own categories, uses its own credits, and gets the resources shared with it, while your organization keeps one place for reporting and administration.

## Category access

When you add a [category](/getting-started/categories) to a team, every member of the team can access it. The access that comes from the team adds to the access a member already has, and never removes from it.

For example, a member has direct access to Category 1 and joins a team that has Categories 2, 3, and 4. They can now access Categories 1–4. If you remove them from the team, they go back to Category 1 only. A member of several teams can access the categories visible to all of them.

A team can access only categories your organization uses, including categories another organization shares with yours. Learn more about shared categories in [Categories in Profound](/getting-started/categories#when-you-have-multiple-organizations-and-categories).

## Sharing resources with a team

Resources, such as Dashboards, Documents, Sheets, Agents, or Workflows, are private to the person who created them until they're shared. You can share them with a whole team instead of inviting members one by one. The sharing dialog lists every team in your organization.

Every member of the team gets the access level you pick when sharing, such as viewer or editor, for as long as they're on the team. The share decides their access to that resource, whatever their role says. When someone leaves the team, they lose access that came through the team, but keep anything shared with them directly.

## Team wallets

A team can have its own credit wallet. Credits in a team wallet are reserved for that team, and no other team can spend them. If your contract sets aside credits for specific business units, give each unit a team wallet so its credits stay separate without moving credits between groups by hand. Profound sets up team wallets for you: contact your Profound account team with your team names and monthly credits per team.

Every member spends from one active wallet at a time:

| Member's situation | Where the credits come from |
| - | - |
| Not on any team | The organization's credits. Admins can still set a [credit limit for individual members](/getting-started/credits#estimate-and-track-credit-usage). |
| Joins a team without a wallet | No change. The team's **Team Wallet** panel shows that no wallet is configured. |
| Joins a team with a wallet | The team wallet, once Profound assigns them to it. Until then, they keep spending from the organization's credits. Individual credit limits don't apply while they spend from a team wallet. |
| On more than one team with a wallet | One of the team wallets, set by Profound. Contact your Profound account team to change it. |
| Removed from a team | Back to the organization's credits. |

When the active wallet [runs out](/getting-started/credits#when-you-run-low-or-run-out), Profound doesn't switch to another wallet. The member can't use credits until the wallet is topped up or an admin moves them to another team.

> [!NOTE]
>
> Whoever runs an Agent, a Workflow, or a Sheet pays for the run from their own active wallet. Sharing one with someone doesn't charge the person who shared it.

## Who can manage teams

The following permissions control teams:

* **Manage teams**: create, rename, and delete teams, and change their categories.
* **Manage team members**: add members to teams and remove them.
* **View people directory**: see the **People** page, including the **Teams** tab.

Admins have all of them by default, and you can add them to any [custom role](/platform-config/people/custom-roles).

These permissions apply to every team in the organization. Members who can't manage teams see only the teams they belong to.

## Limitations

* **No dedicated team admins.** Anyone with the team permissions manages every team in the organization. You can't make someone an admin of only their own team.
* **One role per member.** A member's [role](/platform-config/people/roles-and-permissions) applies across the whole organization, no matter which teams they're on.
* **No individual credit limits inside a team wallet.** Members who need a personal credit limit should stay on the organization's credits.
* **Teams don't span organizations.** A team lives in one organization. Members of one organization can't be added to a team owned by another organization.


# About Projects
Source: https://docs.tryprofound.com/projects/about

How Projects give your team and AI Marketer (Aim) a shared place to work

> [!NOTE]
>
> **Available on:** the Enterprise plan
>
> **Find it in:** Workspace > Projects

Projects gives your team a place for work that spans more than one task, such as a product launch, a category push, or a sentiment fix.

A Project is a container for [AI Marketer (Aim)](/aim/about) chats that share a goal, context, and members. Every chat in a Project uses the same files, Knowledge Bases, and instructions, and the Project collects every Artifact those chats produce. Aim works across your data and context to suggest new Projects and help with existing ones.

[https://player.mux.com/O6lsZlleRWXm5I33QLXx00sM9WSkvY8Fumzica4IfX028?metadata-video-title=Profound+Projects+Walkthrough\&video-title=Profound+Projects+Walkthrough](https://player.mux.com/O6lsZlleRWXm5I33QLXx00sM9WSkvY8Fumzica4IfX028?metadata-video-title=Profound+Projects+Walkthrough\&video-title=Profound+Projects+Walkthrough)

## The problem Projects solve

Outside a Project, each chat with Aim starts from scratch. You explain the goal every time, outputs from earlier chats are hard to find, and your team's work on the same initiative is scattered across chats.

A Project keeps the goal, context, chats, and Artifacts for an initiative in one place, where your teammates can work on it with you.

## Key features

### Work in chats

The Project page opens to a chat box. Ask Aim a question or give it a task. Aim saves every new chat to the Project. The **Your chats** and **Shared chats** tabs list each chat with its status (such as **Needs review**), the Artifacts it produced, and when it was last active.

### Instructions and context

The right sidebar of each Project has **Instructions** and **Context** sections. Instructions set goals, tone, and guidance for how Aim responds in the Project. Context is what Aim knows about the Project: files or photos you add, documents from your[ Knowledge Bases](/context-manager/knowledge-bases-and-brand-kits), and brand information already stored in[ Context Manager](/context-manager/about).

### Suggestions

Aim regularly analyzes your data, including Answer Engine Insights, Prompt Volumes, and Agent Analytics, and suggests new Projects and tasks.

Project suggestions appear at the top of the **Projects** page and propose new initiatives for your organization. The Profound home page also shows recent suggestions in the **Suggestions from Aim** panel. Learn more in [Suggestions from Aim](/projects/suggestions-from-aim).

### Artifacts

Each Project chat can produce Artifacts (outputs), such as Documents, Sheets, or images. They appear as chips in the chat and in the **Artifacts** section of the Project page. Everyone with access to the Project can open them.

### Members and sharing

New Projects are private, but you can invite team members or share a Project with your organization at any time. Members see the Project's details, context, and chats, and can start chats and add Artifacts of their own.

> [!TIP]
>
> Learn how to create your first Project in [Get started with Projects](/projects/getting-started).


# Projects FAQ
Source: https://docs.tryprofound.com/projects/faq

Frequently asked questions about Projects

> **What happened to my existing Projects after the new version of Projects came out?**
>
> Tracked Projects from the previous version moved to the **Archived** tab in Projects, where they're read-only. Migrating them to the new version is on our roadmap.

> **Can I revert to the old version of Projects?**
>
> No. All new Projects use the new version. Opening a previous-version Project from the **Archived** tab shows it in the legacy UI.

> **Do I need to ask for suggestions in a Project?**
>
> No. AI Marketer (Aim) starts generating Suggestions as soon as you create a Project, and adds more as the Project's context and chats grow. Learn more in [Suggestions from Aim](/projects/suggestions-from-aim).

> **How is a Project different from a folder of chats?**
>
> A folder doesn't learn from its contents and tell you what to do next. A Project adds context that every chat inherits, collaboration with teammates in your organization, and Suggestions from Aim based on the Project's context and chats.

> **How are Profound Projects different from Claude Projects in Claude for Enterprise?**
>
> Projects in Claude is a general-purpose workspace: you bring the documents and instructions, and you decide what work to do. Profound Projects runs on your AEO data. Aim has access to your Visibility, Citations, and Sentiment data, so it suggests new Projects and tasks on its own and updates those suggestions as your data changes.

> **Can I run Agents inside a Project?**
>
> Yes. Agent runs are chats, so they can live in a Project. In a Project chat, type `/` or select **+** to add an Agent and run it there.


# Get started with Projects
Source: https://docs.tryprofound.com/projects/getting-started

Create a Project, add context, work in chats, and start tasks from Aim suggestions

This guide takes you from a blank Projects page to a shared Project with context, chats, and a first task started from AI Marketer (Aim) suggestion.

## Step 1: Create a Project

### From the Projects page

1. In Profound, go to **Projects** in the left sidebar and select **+ New Project**. This opens a new Project form.

   ![New Project option highlighted on the Projects page](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=fbf9e26c0786cd4a4722021a2392a81a)
2. Enter a **Project name** and describe the goal under **Tell Aim what to work on**.
3. Choose the **Visibility** for the Project: \
   \- **Private**: only you and invited members can view and use the Project.\
   \- **Your organization**: everyone in your Profound organization can view and use the Project.
4. Select **Create Project**.

![New Project form with fields for the project name, Aim instructions, and visibility.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=eb3ea89c6bca545815f29f9a8dba7ae6)

After that, Aim starts generating suggestions for the new Project right away.

### From a Project Suggestion

On the **Projects** page, open a suggestion card under **New Project Suggestions**. Review its description and tasks, then select **Create Project**, or select **Archive** if it isn't useful. To learn more, see [Suggestions from Aim](/projects/suggestions-from-aim).

### From Profound’s home page

Under the Aim chat box on the home page, select **Start a project** > **Create a new Project**.

![Home page Start a project menu with the Create a new Project option highlighted.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=67526927ebfa8a88882547e215b118cc)

Or ask Aim in the chat box to start a Project, and describe its goal. For example:

```txt wrap theme={null}
Start a Project for our launch of Instant Payouts, a new feature that lets small-business customers move money from their account to their bank in under a minute. It goes live on November 12.

The goal is to make sure that when someone asks an AI assistant about instant payouts for small businesses, or compares payout speeds between us and other payment providers, we show up and the answer describes the feature correctly.
```

That's enough for Aim to analyze your context and suggest tasks. If you already know which tasks you want, describe them in the prompt.

If you're not sure what to ask, select **Start a project** under the chat box and choose a template, such as **Create new content** or **Perform a competitor analysis**. The template pre-fills a brief in the chat box. Tailor it to your goals and send it. Aim asks for anything it still needs, proposes a plan for you to review, and creates the Project.

New Projects appear on the **Projects** page under **Your Projects**. Select a Project to open its page.

![Projects page showing project cards under Your Projects.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-4.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=eb404c40be7ad02645ef83df050a59da)

## Step 2: Add instructions and context

The right sidebar of the Project page has **Instructions** and **Context** sections.

![Project page with the right sidebar showing Instructions and Context sections.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-5.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=c9cb01ef52e0cd543a5bd88b1b5e933c)

### Add instructions

Select **+** next to **Instructions**, and set the goals, tone, and guidance Aim follows in this Project. For example:

```txt wrap theme={null}
Write in plain, confident English. Use short sentences, no jargon, and no exclamation marks.
```

or

```txt wrap theme={null}
Only make claims about speed, fees, and availability that appear in the launch brief attached to this Project.
```

### Add context

Select **+** next to **Context** to add files, photos, or documents from your Knowledge Bases. Aim uses this context in every chat in the Project and to shape its suggestions.

## Step 3: Work on tasks in chats

The Project page opens to a chat box. Ask Aim a question or give it a task. Aim answers using the Project's instructions and context, and saves the chat to the Project so you can return to it anytime.

The chats you start appear in the **Your chats** tab, and chats shared with you appear in the **Shared chats** tab. If Aim is waiting for your reply, the chat's status shows **Needs review**.

![Project chat page with chat tabs and a message box for Aim.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-6.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=a22146f3c23f6f5a3aaa0085ae637455)

### Add files, skills, or Agents

In the chat box, select **+** to add files, images, Knowledge Base documents, or Agents, or to connect external tools. You can also type `/` in a chat to add files or skills, or to reference an Agent.

![Chat attachment menu showing options to add files, skills, Agents, and tools.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-7.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=880bca4c2f542b52d01f5cab942ae917)

## Step 4: Start a task from a Suggestion

In the right sidebar of the Project page, open the **Suggestions for you** panel.

![Project sidebar showing the Suggestions for you panel.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-8.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=bb52929a9b169f85e602e803805569e2)

Each Suggestion is a task Aim recommends based on the Project's context and previous chats. Open one to see its summary, plan, category, sources, and why it’s useful.

To act on a Suggestion, select **Start task** on its card. Aim opens a new chat and starts on the plan. To dismiss a Suggestion, select **Archive**.

![Suggestion card showing details and actions to start or archive the task.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-9.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=9132492319c648de8481221636ee5c18)

> [!TIP]
>
> Add feedback when you archive a Suggestion to improve future Suggestions.

To check for new Suggestions, select the refresh icon on the panel.

![Refresh control for checking for new project suggestions.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-10.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=6286196b51ddec8bf647e3e344741f6d)

## Step 5: Share the Project with your team

At the top of the Project page, select **Share**. Set general access to **Your organization**, or choose **Only people invited** and search for people in your organization to invite. Invited members appear under **Details** in the right sidebar on the Project page.

![Project sharing menu with access settings and invited members.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/getting-started-11.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=50d627aeba77bef81be6328ef110a9f6)

Members see the Project, its context, and its chats, and can start chats and add Artifacts of their own. Each chat is view-only for everyone except the member who started it. Artifacts from every chat appear in the Project's **Artifacts** section, where any member can open them.


# Suggestions from Aim
Source: https://docs.tryprofound.com/projects/suggestions-from-aim

How AI Marketer (Aim) recommends work in Projects and how to act on it

Suggestions are pieces of work that [AI Marketer (Aim)](/aim/about) recommends based on your data, such as a page that's losing citations, a publication that describes your product incorrectly, or a topic where a competitor leads in visibility.

Suggestions appear on Profound's home page (**Workspace > AI Marketer**), on the **Projects** page, and inside each Project. The **Suggestions** page lists all of them.

## Types of Suggestions

### Tasks and Projects

Aim suggests new tasks and Projects.

A task Suggestion is one piece of work: refresh a page, pitch a publication, or add prompts to a topic. A Project Suggestion is an initiative with a goal and a set of tasks.

A task can belong to a Project or run as a stand-alone chat with Aim. From a stand-alone task chat, ask Aim to start a new Project.

> [!TIP]
>
> Learn more about Projects and tasks in [About Projects](/projects/about).

### What you see when you open a Suggestion

A task Suggestion shows:

* **Category**: Owned, Earned, Social, or Paid.
* **Sources**: the data Aim drew on, such as Visibility, Citations, Prompt inventory, or Web research.
* **Summary**: what the data shows.
* **Suggestion**: the plan and why it matters.

![A Task Suggestion details showing category, sources, summary, and recommended action.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=ae3427e1f9d5ddae227ebd2138500239)

A **Project** Suggestion shows its description and a list of recommended tasks. Open any recommendation to see the same details.

To accept a Suggestion, select **Start task** or **Create Project**. To dismiss it, select **Archive**.

## Where to find Suggestions

### Profound's home page

The **Suggestions from Aim** panel is below the Aim chat box.

![Suggestions from Aim panel on the Profound home page.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=f9e5c72817cb438d61de3f0d25053677)

The **For you** tab lists tasks that aren't tied to a Project. The **Existing Projects** tab lists tasks for Projects you have access to, each labeled with its Project.

Select **Refresh** to check for new Suggestions, or select [**Customize Suggestions**](/projects/suggestions-from-aim#customize-suggestions) to tell Aim about your priorities.

### Projects page

Select Projects in the left sidebar to see the **New Project Suggestions** section. Each card is an initiative Aim recommends for your organization, with a description and an initial set of tasks.

### Inside a Project

The **Suggestions for you** panel in a Project's right sidebar lists tasks for that Project, based on its context and chats. Learn more in [Get started with Projects](/projects/getting-started).

## View all Suggestions on the Suggestions page

Wherever you find Suggestions, select **See all suggestions** to open a Suggestions page.

### Filter and search

![Suggestions page with tabs for All, Suggestions, Existing Projects, New Projects, and Archived.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=fdf20ca669ae63fc73021d39efb5d152)

Use the tabs to narrow the list:

* **All**: every open Suggestion.
* **Suggestions**: stand-alone tasks that aren't tied to a Project.
* **Existing Projects**: tasks for Projects you have access to.
* **New Projects**: suggested new Projects.
* **Archived**: Suggestions you or your teammates archived.

Use Search to find a Suggestion by title. Each row shows the Project the Suggestion belongs to, its category, and the date Aim added it. Select a row to open the Suggestion, or select the archive icon at the end of the row to archive it without opening it.

![Suggestions table with the search box and archive option highlighted.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-4.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=c7c68dff5821d18db8c1761458661f40)

### Refresh

Aim adds Suggestions automatically as it finds them. To check for new ones:

* Select **Refresh suggestions** at the top of the Suggestions page.

  ![Refresh suggestions button at the top of the Suggestions page.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-5.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=a11f04963380ad3a3df6874b872abd01)

* Select **Refresh** in **Suggestions from Aim** on the home page.

  ![Refresh button in the Suggestions from Aim panel on the Profound home page.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-6.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=9be438f225e50304ae8fe47fe69bd03e)

* Select Refresh in **New Project Suggestions** on the **Projects** page.

  ![Refresh option in the New Project Suggestions section on the Projects page.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-7.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=9293f9b3a709de58982b607ed789c0f5)

Refreshing runs a new analysis and adds new Suggestions to the list.

## Customize Suggestions

Tell Aim about your role and priorities so its Suggestions match how you work.

To open the **Customize your AI Marketer** drawer, select **Settings** at the top of the Suggestions page, or select **Customize Suggestions** in **Suggestions from Aim** on the home page.

![Customize your AI Marketer drawer with role, focus area, priority, and custom instruction settings.](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/projects/suggestions-from-aim-8.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=ce0e9f8d4092797b3aa9c4d651d336e3)

1. Under **About you**, enter your **Job title** and choose your **Focus areas**: Owned (your site and content), Earned (third-party coverage), Social, Paid, or any combination. To see Suggestions across every area, leave them all selected.
2. Under **Priorities**, select the **Topics**, **Regions**, **Personas**, and **Tags** from your Profound configuration that matter most to you. Aim weights its Suggestions toward them.
3. Under **Custom Instructions**, write rules or goals in plain language, for example, "Prioritize pages that are losing citation share" or "Don't suggest Reddit or YouTube work."
4. Select **Save**, then refresh Suggestions to see the changes.

These settings apply only to you and to the brand you've selected. Your teammates set their own.


# Prompt setup guides
Source: https://docs.tryprofound.com/resources/prompt-guides/overview

Select your industry to learn how to build a prompt set for it

Each guide covers the topics to track, example prompts, refinement steps, and tags for one industry.

> **[Automotive](/resources/prompt-guides/automotive)**
>
> Models and trims, financing, trade-ins, EVs, and recalls

> **[B2B software](/resources/prompt-guides/b2b-software)**
>
> Products and modules, integrations, pricing, security, and alternatives

> **[Consumer packaged goods](/resources/prompt-guides/consumer-packaged-goods)**
>
> Brand lines, occasions, ingredients, packaging, and where to buy

> **[Financial services](/resources/prompt-guides/financial-services)**
>
> Cards, savings, loans, investing, and seasonal promotions

> **[Healthcare](/resources/prompt-guides/healthcare)**
>
> Service lines, insurance, telehealth, availability, and open enrollment

> **[Pharma and life sciences](/resources/prompt-guides/pharma-life-sciences)**
>
> Conditions, treatments, coverage, safety, and prescriber questions

> **[Publishing and media](/resources/prompt-guides/publishing-media)**
>
> Beats, newsletters, podcasts, subscriptions, and news cycles

> **[Retail and e-commerce](/resources/prompt-guides/retail)**
>
> Sales events, gifting, launches, private label, and shipping claims

> **[Telecom and communications](/resources/prompt-guides/telecom)**
>
> Plans, coverage by city, home internet, switching, and bundles

> **[Travel and hospitality](/resources/prompt-guides/travel-hospitality)**
>
> Destinations, trip types, loyalty, and peak dates


# Get started with Sheets
Source: https://docs.tryprofound.com/sheets/getting-started

Run Workflows across hundreds of inputs at once with Sheets

> [!NOTE]
>
> **Available on:** all plans
>
> **Find it in:** Workspace > Sheets

Profound Sheets let you run [Workflows](/agents/workflows/about-workflows) across large sets of inputs simultaneously. Rather than triggering Workflows one input at a time, Sheets lets you process hundreds or thousands of inputs in parallel. A Sheet row can represent a page, topic, keyword, or prompt as a unique input, and collect Workflow run outputs in a single table.

Sheets supports a range of content operation workflows. Common use cases include:

* **AEO health reports:** Pull visibility, citation, and sentiment data across all tracked topics, run competitive gap analysis on each, and generate a prioritized action list on a recurring schedule.
* **Bulk blog page optimization:** Import a CSV of underperforming blogs and run an AEO Content Refresh Workflow across all of them simultaneously.
* **Localization at scale:** Adapt and translate content across markets in parallel, preserving AEO structure for each region's prompt patterns.

## Run Workflows at scale with Sheets

A Sheet is structured as a spreadsheet. Each row contains an individual input, such as a URL, keyword, or tracked prompt. Each column is either a data field, a Workflow step that processes the row's input, or the output of the run.

When you run a Workflow from Sheets, every row executes in parallel. Workflow Agent columns operate on their row's input data and write outputs back into the table automatically. Review, edit, and export results to your CMS from within the platform.

### Workflow outputs in Sheets

When a Workflow Agent column finishes processing, the output lands in the sheet table automatically. From there, review and edit results directly.

Pass the output into another Workflow downstream within the same sheet, push it directly to a CMS, or edit the data within the sheet.

### Create a new Sheet

To create a new Sheet, navigate to the Sheets section and select **+ Create new**.

Select **+ Blank sheet** for an empty sheet, or import existing data from a file, Google Sheets, or other supported services.

![Sheets page with the Create new menu highlighted showing Blank sheet and import options](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/sheets/getting-started-1.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=99a29838b50d3d91095a7d371cf623a5)

Once created, the Sheet opens with a default set of empty rows and columns, as well as an **Add an Agent** dialog that lets you select a published Workflow or create a new one to add to your Sheet.

Rename the Sheet at any time by clicking its title at the top left of the screen. Each Sheet operates independently, so you can maintain separate Sheets for different projects, workflows, or datasets.

### Add and manage columns

To add a column, click **+ Add Column** next to the last used column header in the header row. You'll be prompted to choose a column type.

![Screenshot of the blank sheet view with Add column option highlighted and a dropdown menu open](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/sheets/getting-started-2.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=5acbfaa72c82cca75e5e873b39ac851f)

Sheets support the following column types:

* **Agent** — Runs a Profound Workflow against each row's input and returns the result. Configure Workflows to use specific prompts and tools.
* **Text** — A short text field for entering or displaying brief string values.
* **Long text** — A larger text field suited for multi-sentence or paragraph-length content.
* **Number** — Stores numeric values for quantitative data or calculations.
* **List** — A collection of items, such as numbers or strings of text. Useful for iterating over a set of values.
* **Date range** — Captures a start and end date, useful for scheduling or time-bound data.
* **Dropdown** — Presents a fixed list of options for each cell, keeping values consistent across rows.
* **Profound Topics** — A dropdown containing predefined topics within Profound, enabling topical categorization of rows.
* **Profound Prompts** — A dropdown with saved prompts from your Profound prompt library.
* **Boolean** — A true/false toggle for flagging rows or tracking binary states.

### Edit or remove columns

Once you select a column type, name it and configure any type-specific settings. From a column's settings, you can also add new columns to the left or right, pin it, hide it, or rename it.

![A screenshot of the list view with a column settings menu open and highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/sheets/getting-started-3.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=7b3c4dbd28e81ac08b223a6428d6267d)

To remove a column, open its settings and select **Delete Column**.

> [!TIP]
>
> If you delete a Workflow Agent column that references other columns as inputs or outputs, you'll have the option to delete only the Workflow Agent column or all associated columns.

## Import and export Sheet data

Use the **Import** and **Export** buttons at the top right of the screen to move data in and out of your Sheet.

![A sheet view with Import and Export buttons highlighted](https://mintcdn.com/profound-37face47/d3tYUJ4mfdayQ4Zm/images/sheets/getting-started-4.png?fit=max\&auto=format\&n=d3tYUJ4mfdayQ4Zm\&q=85\&s=6b019328596e68a021cdb16fe6806ed0)

### Import

Importing lets you populate rows from a file, Google Sheets, and other supported services. It's the fastest way to load large datasets into a Sheet without manual entry.

### Export Sheets

Exporting saves your current Sheet data as a specified file type. Exports preserve all column values and row content, including any completed Workflow outputs, making it easy to share results or perform further analysis outside the platform.

Supported file types for export include:

* Microsoft Excel (.xlsx)
* OpenDocument (.ods)
* PDF (.pdf)
* Web Page (.html)
* Comma Separated Values (.csv)
* Tab Separated Values (.tsv)

## Settings

The **Settings** button in the top toolbar provides access to sheet-level configuration options. From here, edit Sheet details, hide columns while preserving their data, or duplicate the current Sheet.


# Run Workflows from Sheets
Source: https://docs.tryprofound.com/sheets/run-workflows-from-sheets

Set up a Sheet, map inputs, and run a Workflow over every row

Profound Sheets lets you run [Workflows](/agents/workflows/about-workflows) (Workflow Agents) across large sets of inputs simultaneously. Rather than triggering Workflows one input at a time, Sheets lets you process hundreds or thousands of inputs in parallel.

## Adding a Workflow Agent to a Sheet

To add a Workflow to a Sheet:

1. Click **+ Add Column** next to the last used column header in the header row.
2. From the column type menu, select **Agent**.
3. In the **Add an Agent** dialog, select which Workflow Agent you want to run or create a new one.
4. Follow the prompts to configure Workflow Agent inputs and output columns and add them to the Sheet.

The Workflow Agent column appears in the Sheet grid alongside your other columns. Each row in that column displays a **Run** button, indicating the Workflow is ready to process that row's input when triggered.

## Run a Workflow from a Sheet

### Single run

To [run a Workflow](/agents/workflows/workflow-runs) on a single row, click the **Run** button within that row's Workflow Agent column cell. The Workflow processes that row's input and returns its output directly in the cell once complete.

![Screen Shot 2026-04-03 at 12.00.59 PM.png](https://mintcdn.com/profound-37face47/2mGao8qF1TMHCUpR/images/sheets/run-agents-from-sheets-2.png?fit=max\&auto=format\&n=2mGao8qF1TMHCUpR\&q=85\&s=ceabbef6a2632c3bb250f56fb038e619)

The cell status updates to show it's running, then displays a green **Completed** indicator when finished. Single runs are useful for testing your Workflow Agent setup on a specific input before committing to a full column run.

### Bulk run

To run the Workflow Agent across all rows at once, click **Run** at the top of the Workflow Agent column header. This triggers the Workflow for every row with input data in the Sheet simultaneously, processing all inputs in parallel.

Monitor progress as individual cells update their status in real time, moving from **Run** to a running state, then to **Completed** as each Workflow finishes.

> [!WARNING]
>
> Workflows only run **when the necessary inputs are present**. For example, if you have 100 rows but only 13 contain the necessary inputs, the Workflow runs for those 13 rows only.

## Handle outputs

Once a Workflow Agent run is complete, its output appears in the output cell for that row (configured in the **Add an Agent** dialog). Add a separate Text or Long text column alongside the Workflow Agent column to capture, label, or annotate those outputs.

To use Workflow Agent outputs downstream, reference that column's values as inputs in other columns or workflows.


# Get an agent
Source: https://docs.tryprofound.com/api-reference/agents/get-an-agent

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents/{agent_id}
Retrieve an agent and its schema details.

Agents can have both a live published version and a draft version with newer
unpublished changes. Use the `version` parameter to choose which state to return.



# Get an agent run
Source: https://docs.tryprofound.com/api-reference/agents/get-an-agent-run

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents/{agent_id}/runs/{run_id}
Retrieve the current status and result details for an agent run.



# List agents
Source: https://docs.tryprofound.com/api-reference/agents/list-agents

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents
List agents available to your organization.

Agent status reflects whether an agent has ever been published. `published`
agents have a live published version. `draft` agents have not been
published yet.



# Run an agent
Source: https://docs.tryprofound.com/api-reference/agents/run-an-agent

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/agents/{agent_id}/runs
Start a new run for an agent.

Runs always execute the agent's live published version, so the agent must be
published first with `POST /v1/agents/{agent_id}/publish`. Unpublished drafts
cannot be run.



# Optimization Analysis
Source: https://docs.tryprofound.com/api-reference/content/optimization-analysis

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/content/{asset_id}/optimization/{content_id}



# Optimization List
Source: https://docs.tryprofound.com/api-reference/content/optimization-list

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/content/{asset_id}/optimization



# Add Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/add-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/documents
Add a document to a knowledge base using JSON text or multipart file upload.



# Add Folder
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/add-folder

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/folders
Create an empty folder at the requested knowledge base path.



# Delete Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/delete-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed delete /v1/knowledge-bases/{knowledge_base_id}/documents
Delete an existing document from a knowledge base.



# Delete Folder
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/delete-folder

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed delete /v1/knowledge-bases/{knowledge_base_id}/folders
Delete a folder. With recursive=false, non-empty folders return 409 and no contents are deleted.



# List Knowledge Bases
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/list-knowledge-bases

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/knowledge-bases
List knowledge bases accessible to the API key.



# Search Knowledge Base
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/search-knowledge-base

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/search
Search a knowledge base and return matching snippets or pages.



# Update Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/update-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed put /v1/knowledge-bases/{knowledge_base_id}/documents
Overwrite a knowledge base document using JSON text or multipart file upload.



# Get Account Insights
Source: https://docs.tryprofound.com/api-reference/openai-ads/get-account-insights

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/ads/openai-ads/ad-account/insights
Get ad account insights for the organization's OpenAI Ads partner brand.

`aggregation_level=campaign` returns one row per campaign (with `campaign_id`
/ `campaign_name` and all metrics), so every campaign's insights come back in
a single call; `time_granularity=daily` gives per-day rows (e.g. daily spend).



# Create prompts
Source: https://docs.tryprofound.com/api-reference/organization/create-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/org/categories/{category_id}/prompts
Create one or more prompts in a category. Topics and tags are auto-created if referenced by name and not yet existing. Use dry_run to preview without persisting.



# Get Assets
Source: https://docs.tryprofound.com/api-reference/organization/get-assets

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/assets
Get the organization assets, one row per (asset, organization) pair.

An asset's category can belong to multiple organizations; one asset row is
emitted per owning org so no association is silently dropped.



# Get Categories
Source: https://docs.tryprofound.com/api-reference/organization/get-categories

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories
Get the organization categories, one row per (category, organization) pair.



# Get Category Assets
Source: https://docs.tryprofound.com/api-reference/organization/get-category-assets

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/assets



# Get Category Citation Categories
Source: https://docs.tryprofound.com/api-reference/organization/get-category-citation-categories

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/citation-categories
Get the citation categories for a category: the built-in buckets plus any custom categories.



# Get Category Citation Tags
Source: https://docs.tryprofound.com/api-reference/organization/get-category-citation-tags

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/citation-tags
Get the custom citation tags defined for a category.



# Get Category Personas
Source: https://docs.tryprofound.com/api-reference/organization/get-category-personas

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/personas



# Get Category Regions
Source: https://docs.tryprofound.com/api-reference/organization/get-category-regions

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/regions
Get the regions for a specific category.



# Get Category Tags
Source: https://docs.tryprofound.com/api-reference/organization/get-category-tags

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/tags
Get the tags for a specific category.



# Get Category Topics
Source: https://docs.tryprofound.com/api-reference/organization/get-category-topics

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/topics
Get the topics for a specific category.



# Get Domains
Source: https://docs.tryprofound.com/api-reference/organization/get-domains

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/domains
Get the organization domains.



# Get Models
Source: https://docs.tryprofound.com/api-reference/organization/get-models

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/models
Get the organization models.



# Get Personas
Source: https://docs.tryprofound.com/api-reference/organization/get-personas

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/personas
Get the organization personas, one row per (persona, organization) pair.

Same (item, org) fan-out as ``get_assets``: a persona's category can be
owned by multiple orgs, and each owning org gets its own row so no
association is silently dropped.



# Get Regions
Source: https://docs.tryprofound.com/api-reference/organization/get-regions

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/regions
Get the organization regions.



# List prompts
Source: https://docs.tryprofound.com/api-reference/organization/list-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/prompts
Retrieve prompts in a category with optional filtering by type, topic, tag, region, platform, or persona. Supports cursor-based pagination.



# Update prompt status
Source: https://docs.tryprofound.com/api-reference/organization/update-prompt-status

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed patch /v1/org/categories/{category_id}/prompts/status
Bulk-update the status of one or more prompts. Prompts already in the target status are skipped. Use dry_run to preview without persisting.

Status options:
- 'active': Prompts will run daily.
- 'disabled': Prompts will not run moving forward, but historical data is preserved.
- 'deleted': Prompts are deleted along with historical data



# Update prompts
Source: https://docs.tryprofound.com/api-reference/organization/update-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed patch /v1/org/categories/{category_id}/prompts
Update one or more existing prompts. Only provided fields are changed. Dimension fields (regions, platforms, personas, tags) replace the full set when provided. Use dry_run to preview without persisting.



# Get Answers
Source: https://docs.tryprofound.com/api-reference/prompts/get-answers

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/prompts/answers



# Get Bots Report V1
Source: https://docs.tryprofound.com/api-reference/reports/get-bots-report-v1

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/bots
Get bot traffic report from the daily aggregated materialized view.

This endpoint queries pre-aggregated daily bot data, making it efficient
for large date ranges and high-traffic sites.

Metrics:
- count: unique bot visits
- citations: unique citation events
- indexing: unique indexing events
- training: unique training events
- last_visit: most recent visit timestamp



# Get Bots Report V2
Source: https://docs.tryprofound.com/api-reference/reports/get-bots-report-v2

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v2/reports/bots
Get bot traffic report from the hourly aggregated materialized view (UTC-based).

Supports date_interval="hour", calendar intervals through "year", "quarter", and "relative_week".
When `view_id` is provided, the query is scoped to that domain segment's hosts and paths.

Metrics:
- count: unique bot visits
- citations: unique citation events (ai_assistant bot type)
- indexing: unique indexing events (index bot type)
- training: unique training events (ai_training bot type)
- last_visit: most recent visit timestamp

Dimensions:
- date, path, bot_name, bot_provider, bot_type



# Get Referrals Report V1
Source: https://docs.tryprofound.com/api-reference/reports/get-referrals-report-v1

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/referrals
Get referral traffic report from the daily aggregated materialized view.

This endpoint queries pre-aggregated daily referral data, making it efficient
for large date ranges and high-traffic sites.



# Get Referrals Report V2
Source: https://docs.tryprofound.com/api-reference/reports/get-referrals-report-v2

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v2/reports/referrals
Get referral traffic report from the hourly aggregated materialized view (UTC-based).

Supports date_interval="hour", calendar intervals through "year", "quarter", and "relative_week".
When `view_id` is provided, the query is scoped to that domain segment's hosts and paths.



# Query Citations
Source: https://docs.tryprofound.com/api-reference/reports/query-citations

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/citations
Get citations for a given category.

The ``mentioned`` filter supports ``is true`` and ``is false``. It uses the
latest page analysis available at or before ``end_date``; pages without an
analysis by then are excluded from both values. ``citation_share`` keeps all
otherwise eligible citations in its denominator when this filter is used.



# Query Fanouts
Source: https://docs.tryprofound.com/api-reference/reports/query-fanouts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/query-fanouts



# Query Sentiment
Source: https://docs.tryprofound.com/api-reference/reports/query-sentiment

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/sentiment
Get citations for a given category.



# Query Visibility
Source: https://docs.tryprofound.com/api-reference/reports/query-visibility

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/visibility
Query visibility report.



# Introduction
Source: https://docs.tryprofound.com/bots/introduction

An overview of Profound bots

Profound operates a small number of automated HTTP agents that fetch publicly available data from the web on behalf of our customers. This section documents each bot, what it does, when it runs, and how site operators can identify, allow, or block it.

## Active bots

| Name | User-Agent | Trigger | Purpose |
| - | - | - | - |
| [ProfoundBot](/bots/profoundbot) | `ProfoundBot/1.0 (+https://docs.tryprofound.com/bots)` | User-initiated | Fetches a single public web page on demand so you can read its content as Markdown, HTML, or both |
| [Profound Sheets Sitemap Importer](/bots/sitemap-importer) | `ProfoundSheetsSitemapImporter/1.0` | User-initiated | Fetches a website's `sitemap.xml` and `robots.txt` so you can import sitemap URLs into a Profound Sheet |
| [Profound Knowledge Base Fetcher](/bots/knowledge-base-fetcher) | `ProfoundKnowledgeBase/1.0 (+https://docs.tryprofound.com/bots)` | User-initiated, or auto-sync when enabled | Fetches a website's sitemap files when you import the site into a Knowledge Base, and re-checks imported PDF files for changes |

## Bot operating rules

Every Profound bot follows the same operating rules:

* **Public data only:** Our bots fetch resources that are intended to be publicly readable (for example, `sitemap.xml`, `robots.txt`, public HTML pages). They never attempt authenticated areas, never submit forms, and never bypass access controls.
* **User-Agent identification:** Every bot sends a stable, descriptive User-Agent header with a name that begins with `Profound`. Site operators can match on that prefix to identify all Profound traffic.
* **No persistent crawling without disclosure:** Any bot that runs on a schedule, recursively follows links, or maintains an index has a dedicated page in this section documenting its behaviour in detail.

> [!NOTE]
>
> Reach us at [security@tryprofound.com](mailto:security@tryprofound.com) to report abuse, request blocking, or ask questions about the traffic you have observed. We treat security seriously and respond to all reports.


# Profound Knowledge Base
Source: https://docs.tryprofound.com/index

Guides, references, and recipes for using Profound and building on its data

# Profound Knowledge Base

Guides, references, and recipes for using Profound and building on its data

> **[Analyze your AI search presence](/getting-started/glossary)**
>
> Answer Engine Insights, Prompt Volumes, dashboards, and the basics to get started

> **[Get the work done with AI Marketer (Aim) and Agents](/aim/about)**
>
> Your AI marketing teammate, Projects, Agents, Sheets, and the brand context behind them

> **[Set up your organization](/platform-config/authentication/configure-sso)**
>
> Authentication, Activity Logs, Analytics Setup, team access, and agency mode

> **[Connect your tools](/integrations/overview)**
>
> Integrations and connectors that make your CMS, analytics, and messaging tools available to Agents and Aim

> **[Build with the API and SDKs](/rest-api/introduction)**
>
> Pull Profound data into your own applications, reports, and pipelines

> **[Work with Profound from AI tools](/mcp/overview)**
>
> Give ChatGPT, Copilot, and coding assistants access to your Profound data through MCP

> **[Follow reporting recipes](/cookbook/introduction)**
>
> Ready-made API queries for visibility, citations, and metrics

## More ways to learn

> **[Blog](https://www.tryprofound.com/blog)**
>
> Product news and perspectives on AI search

> **[Community](https://community.tryprofound.com)**
>
> Meet other Profound users, swap tips, and ask questions

> **[Profound University](https://university.tryprofound.com)**
>
> Courses and resources to build your expertise

> **[Research hub](https://www.tryprofound.com/research)**
>
> Studies and insights on AI search from the Profound team

> **[The Marketing Engineer Podcast](https://www.tryprofound.com/podcasts)**
>
> Conversations with the people shaping marketing in the AI era

> **[Webinars](https://www.tryprofound.com/webinars)**
>
> Live and recorded sessions on the platform and AI search strategy

> **[White papers](https://www.tryprofound.com/reports-guides)**
>
> In-depth reports and guides on AI search

## Need help?

> **[Status](https://status.tryprofound.com)**
>
> Check service health and current incidents

> **[Contact support](mailto:support@tryprofound.com)**
>
> Reach the Profound team for help with your account or setup


# Authentication
Source: https://docs.tryprofound.com/mcp/authentication

Learn how to authenticate with the hosted Profound MCP server

## OAuth

Profound MCP uses OAuth 2.1. When your client prompts for authentication, complete the browser sign-in with your Profound account. Every tool call runs as the authenticated user and returns only the data that user can access.

> [!NOTE]
>
> OAuth is the recommended authentication method for all Profound MCP connections.

## Long-lived Bearer token (API key)

Profound MCP supports a long-lived Bearer token authentication based on an API key. This is useful for setups that can't complete a per-user OAuth flow each time, such as service accounts or internal tools.

### Setup

**1. Generate an API key**

Follow the instructions in [Getting Your API Key](/rest-api/authentication#getting-your-api-key) section of Profound API documentation.

**2. Configure your MCP client's authentication settings**

In your MCP client's authentication settings for the Profound connection, select **Bearer Token** and enter:

```text theme={null}
Bearer <your_api_key>
```

### Security notes

* Your API key is sensitive information. Treat it like a password: store it securely, never commit it to version control, and set an expiration date when you create it.

* Revoke keys you no longer use from the **API Keys** page in your Profound account settings.


# Analytics and reporting
Source: https://docs.tryprofound.com/mcp/capabilities/analytics-capabilities

Understand the analytics and reporting tools available through Profound MCP

Profound MCP gives AI assistants read-only access to Profound's Answer Engine Optimization (AEO), brand visibility, citation, sentiment, shopping, prompt volume, and Agent Analytics data. Use this page to understand what the hosted MCP server can do before connecting it to an MCP client. When you're ready to connect, refer to the [Connection guides](/mcp/common-mcp-clients).

## How the tools work together

Most conversations start with the [discovery tools](/mcp/capabilities/discovery-tools), then move into reports. Here's what that typically looks like:

**1. Confirm access**

Ask what data you can see. The assistant confirms the signed-in user, available organizations, regions, and entitlements with [`whoami`](/mcp/capabilities/discovery-tools#whoami).

**2. Find the right scope**

Ask about the brand or site you care about. The assistant resolves it with [`list_organizations`](/mcp/capabilities/discovery-tools#list-organizations), then [`list_categories`](/mcp/capabilities/discovery-tools#list-categories) for visibility reports or [`list_domains`](/mcp/capabilities/discovery-tools#list-domains) for traffic reports.

**3. Add optional filters**

Ask to narrow the question to a market, AI engine, prompt set, topic, or tag. The assistant resolves those filters with [`list_regions`](/mcp/capabilities/discovery-tools#list-regions), [`list_models`](/mcp/capabilities/discovery-tools#list-models), [`list_tags`](/mcp/capabilities/discovery-tools#list-tags), [`list_topics`](/mcp/capabilities/discovery-tools#list-topics), and [`list_prompts`](/mcp/capabilities/prompts-capabilities#list-prompts).

**4. Run a report**

Ask about what you want reported. The assistant retrieves visibility, sentiment, citations, prompt answers, shopping, prompt volume, referrals, or bot crawl data for the date range you name.

## Behavior and safety

All tools are read-only. They retrieve analytics data, but don't create, update, or delete anything in Profound.

| Behavior | What it means |
| - | - |
| Non-destructive | No tool performs destructive updates |
| Live data | Tools read from the Profound API, so results reflect the caller's current access and data |

All report tools set the following MCP hints:

| Hint | Value | What it means |
| - | - | - |
| `readOnlyHint` | `true` | The tool only reads data |
| `idempotentHint` | `true` | The tool is safe to retry with the same arguments |

Dates are ISO 8601 strings in `YYYY-MM-DD` format. Reports validate `start_date` and `end_date` and return an actionable error if the date window is invalid.

## Visibility reports

These tools are scoped to a `category_id` and a date range. Use them to understand how brands appear in AI answers, both in regular text answers and in [shopping mode](https://help.tryprofound.com/articles/8862399588-about-shopping) results.

### Brand visibility reports

Use these tools to understand how brands appear in AI answers, how those answers feel, how accurate they are, and which sources AI engines cite.

| Tool | Usage |
| - | - |
| `get_visibility_report` | How visible a brand is in AI answers and how that varies by model, topic, region, prompt, or persona |
| `get_sentiment_report` | What sentiment AI answers express about one brand or competitor and which themes and claims drive it |
| `get_citations_report` | Which domains, pages, and citation categories AI engines cite for this category |
| `get_prompt_answers` | What raw AI answers Profound observed behind the metrics |
| `get_factcheck_report` | FactCheck scores for a category over a date range |
| `get_factcheck_claims` | Inaccurate claims identified in a category over a date range |

> **get\_visibility\_report**
>
> Measures how often and how prominently a brand appears in AI answers for a category over a date range.
>
> **Example prompts**:
>
> * "How visible were we in AI answers last month, broken down by model?"
> * "Are we gaining or losing visibility against competitors this quarter?"
>
> Default metric: `visibility_score`. The other metrics are `share_of_voice` and `average_position`.
>
> `visibility_score` is a raw decimal. Multiply it by 100 to match the percentage the Profound platform shows, so `0.42` is 42%.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `group_by` | No | `null` | Group results by `date`, `model`, `topic`, `region`, `prompt`, or `persona` |
> | `metrics` | No | `visibility_score` | Metrics to return: `visibility_score`, `share_of_voice`, or `average_position` |
> | `interval` | No | `day` | Size of each time bucket when grouping by `date`: `day`, `week`, or `month` |
> | `scope` | No | `owned` | `owned` for your tracked brands, or `all` to include competitors |
> | `assets` | No | `null` | Narrow to one or more brands, by name |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `model_filter` | No | `null` | Narrow to one or more AI models, by ID from `list_models` |
> | `persona_filter` | No | `null` | Narrow to one or more personas, by ID |
> | `filter` | No | `null` | Filter expression for conditions the other inputs can't express, such as excluding one region or matching either of two topics. Combined with those inputs using `and` |
> | `limit` | No | `null` | Top-level groups per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes**:
>
> * Every row carries an `asset` object of `{name, owned}`. Results are always grouped by brand as well as by `group_by`, so you get one row per brand per bucket.
> * `limit` caps top-level groups rather than rows, so a grouped query returns more rows than the limit you set.

> **get\_sentiment\_report**
>
> Measures the sentiment AI answers express about one brand or competitor in a category over a date range, or the sentiment of the pages those answers cite.
>
> **Example prompts**:
>
> * "How do AI answers feel about our pricing?"
> * "Which cited pages are most negative about our brand?"
>
> Sentiment comes from either the answers themselves or the pages they cite, and `source` selects which.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `asset` | Yes | - | Brand or competitor name to analyze |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `source` | No | `response` | `response` for tone in AI answers, `citation` for tone on cited pages |
> | `group_by` | No | `[]` | Group response rows by `date`, `model`, `topic`, `region`, `prompt`, `persona`, `tag`, `theme`, or `claim`. Takes at most two non-date groupings, plus `date` for a time series |
> | `theme` | No | `null` | Narrow to one sentiment theme, such as pricing |
> | `claim` | No | `null` | Narrow to one claim |
> | `model` | No | `null` | Narrow to one or more AI models |
> | `citation_category` | No | `null` | Narrow to one or more citation categories, including custom category names |
> | `page_contains` | No | `null` | Narrow to pages whose URL contains this text. Case doesn't matter |
> | `citation_sort_by` | No | `citation_share` | Sort citation rows by `citation_share` or `positive_sentiment` |
> | `citation_sort_direction` | No | `desc` | Sort direction, `asc` or `desc` |
> | `limit` | No | `10` | Rows per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes**:
>
> * `citation_category`, `page_contains`, `citation_sort_by`, and `citation_sort_direction` are available only when `source` is `citation`, and return an error otherwise.
> * Filters narrow results, and never group them. When `source` is `citation`, the `model` changes which pages qualify and their citation share, but the sentiment is still for the page, not for the model.

> **get\_citations\_report**
>
> Shows which sources AI engines cite for a category, and how often, over a date range.
>
> **Example prompts**:
>
> * "Which sites do AI engines cite most in our category?"
> * "How often do AI answers cite our own domain?"
> * "How are citations split across citation categories?"
>
> Default metrics: `count` and `citation_share`. `count` is raw citation frequency, and `citation_share` is averaged per AI model so it stays comparable across models. `citation_coverage` is the fraction of answers, from 0 to 1, that cite the domain or page at least once, also averaged per AI model.
>
> `entity` sets what each row represents. With no `entity`, each row is one cited domain, ranked most cited first. Use that default to find which sites AI engines cite most. `entity: "page"` returns one row per cited page URL, and `entity: "citation_category"` returns one row per citation category, with its `id` and `name`, ordered by citation share.
>
> > [!NOTE]
> >
> > `scope: "owned"` and `domain_filter` work with domain rows only, and `page_filter` and `first_cited_at` work with page rows only. Citation category rows take no `group_by` and only the `count` and `citation_share` metrics. Other combinations return an error.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `entity` | No | `domain` | What each row represents: `domain`, `page`, or `citation_category` |
> | `group_by` | No | `null` | Group results by `date`, `model`, `topic`, `region`, `persona`, or `prompt`. Domain rows take one value, or `topic` and `model` together, and page rows take one value. Add `date` to either for a time series |
> | `metrics` | No | `count`, `citation_share` | Metrics to return: `count`, `citation_share`, `citation_coverage`, `rank`, or `first_cited_at` |
> | `interval` | No | `day` | Size of each time bucket when grouping by `date`: `day`, `week`, or `month` |
> | `scope` | No | `all` | `all` to cover every cited domain, or `owned` to narrow to your own domains |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `model_filter` | No | `null` | Narrow to one or more AI models, by ID from `list_models` |
> | `persona_filter` | No | `null` | Narrow to one or more personas, by ID |
> | `domain_filter` | No | `null` | Narrow to one or more domains, including their subdomains |
> | `page_filter` | No | `null` | Narrow to one or more page URLs |
> | `citation_category_filter` | No | `null` | Narrow to one or more citation categories, such as `owned` or `social` |
> | `analysis_type_filter` | No | `null` | Narrow to one analysis type: `visibility`, `sentiment`, `factcheck`, or `all` |
> | `filter` | No | `null` | Filter expression for conditions the other inputs can't express, such as excluding one region or matching either of two topics. Combined with other filter inputs using `and`. A `domain` or `page` field inside it follows the same entity rules as `domain_filter` and `page_filter` |
> | `limit` | No | `null` | Top-level groups per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes**:
>
> * Domain and page rows carry a `rank`, where `1` is the most cited. `first_cited_at` is returned on page rows only.
> * `citation_coverage` is returned on domain and page rows only, and its value can be `null`. It's enabled per category. When it isn't enabled for the category you query, the report still returns the other metrics you asked for, and `info.metrics` leaves out `citation_coverage`. Your Profound account team can enable it.
> * Citation category rows carry no `rank`, and their `citation_share` values can total more than 100%, because a citation can count under both its page's category and its domain's category.
> * `info.entity` in the response reports which entity the rows represent.
> * `group_by: ["page"]` still works as before: it selects page rows when `entity` is omitted. Setting `entity` to anything other than `page` alongside it returns an error.
> * `limit` caps top-level groups rather than rows, so a grouped query returns more rows than the limit you set.

> **get\_prompt\_answers**
>
> Retrieves the answers AI engines gave for a category's prompts over a date range: the raw text behind the visibility metrics.
>
> **Example prompts**:
>
> * "What did AI engines answer for our savings prompts last week?"
> * "Pull the answers behind last month's visibility drop."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to retrieve prompt answers for |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `prompt_id` | No | `null` | Narrow the results to a single prompt |
> | `country` | No | `null` | Narrow to one or more countries, by name or ID. Can't be combined with `region` |
> | `region` | No | `null` | Narrow to one or more regions, by name or ID. Can't be combined with `country` |
> | `topic_ids` | No | `null` | Narrow to one or more topics, by ID from `list_topics` |
> | `model_ids` | No | `null` | Narrow to one or more answering AI models, by ID from `list_models` |
> | `limit` | No | `100` | Rows per page |
> | `offset` | No | `0` | Row offset for pagination |
>
> **Notes**:
>
> * Several IDs in one list match any of those IDs. When both inputs are set, an answer has to match both.
> * Filters apply before pagination, so a filtered call pages through matching answers only.

> **get\_factcheck\_report**
>
> Reports how accurately AI answers describe a category over a date range: the FactCheck score, its trend, and the accurate and inaccurate counts behind it.
>
> **Example prompts**:
>
> * "What's our FactCheck score for last month?"
> * "Is accuracy improving, and which models are least accurate?"
>
> Rows always return `accuracy`, `accurate`, and `inaccurate`. With no `group_by`, the report returns the headline score for the category.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `group_by` | No | `[]` | Group results by `date`, `model`, `region`, `persona`, `prompt`, `topic`, `tag`, `citation`, or `theme`. Takes up to two values, and `citation` can't be combined with another value |
> | `model` | No | `null` | Narrow to one or more AI models, by exact name |
> | `topic` | No | `null` | Narrow to one or more topics, by exact name from `list_topics` |
> | `topic_negate` | No | `false` | When `true`, exclude the topics in `topic` instead of narrowing to them |
> | `region` | No | `null` | Narrow to one or more regions, by exact name |
> | `persona` | No | `null` | Narrow to one or more personas, by exact name |
> | `prompt` | No | `null` | Narrow to one or more prompts, by exact name |
> | `tag` | No | `null` | Narrow to one or more tags, by exact name |
> | `limit` | No | `100` | Rows per page, up to 100 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes**:
>
> * FactCheck is scoped to a category, so there's no brand or competitor input.
> * Filters match names exactly.
> * `topic_negate` applies to `topic` only, and leaves the other filters narrowing as usual.

> **get\_factcheck\_claims**
>
> Lists the inaccurate claims AI answers made about a category over a date range.
>
> **Example prompts**:
>
> * "Which inaccurate claims came up in our category last month?"
> * "What's the evidence behind the most common inaccurate claim, and which pages are cited for it?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `group_by` | No | `[]` | Group results by one of `model`, `region`, `persona`, `prompt`, `topic`, `tag`, or `theme` |
> | `include` | No | `[]` | Extra detail per claim: `theme`, `reasoning`, `models`, `evidence`, or `citation_sources` |
> | `model` | No | `null` | Narrow to one or more AI models, by exact name |
> | `topic` | No | `null` | Narrow to one or more topics, by exact name from `list_topics` |
> | `topic_negate` | No | `false` | When `true`, exclude the topics in `topic` instead of narrowing to them |
> | `region` | No | `null` | Narrow to one or more regions, by exact name |
> | `persona` | No | `null` | Narrow to one or more personas, by exact name |
> | `prompt` | No | `null` | Narrow to one or more prompts, by exact name |
> | `tag` | No | `null` | Narrow to one or more tags, by exact name |
> | `limit` | No | `25` | Rows per page, up to 100, or up to 5 when `include` contains `citation_sources` |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes**:
>
> * `group_by` takes one value, and `group_by: ["tag"]` can't be combined with a `tag` filter. Grouping by date, citation, or claim isn't supported.
> * `citation_sources` adds the pages cited for each claim, so it can't be combined with `group_by`, and `limit` has to be 5 or less. A higher limit returns an error instead of being reduced automatically.
> * The `include` values add why each claim is inaccurate, which models repeated it, and which pages are cited for it. Those values can be combined, so one ungrouped call can return a claim's theme, reasoning, models, evidence, and citation sources together.
> * For the score, its trend, or accurate and inaccurate counts, use [`get_factcheck_report`](#get-factcheck-report) instead.

### Legacy visibility and citations reports

`get_visibility_report` and `get_citations_report` each have an earlier version that returns the legacy response shape: rows as positional values, grouped with `dimensions` rather than `group_by`, and without the `scope` and `assets` brand scoping. Both remain available for integrations built against the earlier shape.

> [!NOTE]
>
> These tools take a half-open date window: `start_date` is inclusive and `end_date` is exclusive. To cover all of April 2026, pass `start_date: 2026-04-01` and `end_date: 2026-05-01`. The current tools treat both ends as inclusive, so the same dates cover a different window.

| Tool | Usage |
| - | - |
| `get_visibility_report_v1` | Visibility for a category, in the legacy response shape |
| `get_citations_report_v1` | Citations for a category, in the legacy response shape |

> **get\_visibility\_report\_v1**
>
> Measures how often and how prominently a brand appears in AI answers for a category over a date range, and returns the legacy response shape. For new integrations, use [`get_visibility_report`](#get-visibility-report) instead.
>
> Default metric: `visibility_score`.
>
> Other useful metrics include `share_of_voice`, `mentions_count`, `executions`, and `average_position`. Useful dimensions include `date`, `region`, `topic`, `model`, `prompt`, `tag`, `persona`, and `asset_name`.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (exclusive) in `YYYY-MM-DD` format |
> | `metrics` | No | `visibility_score` | Metrics to return |
> | `dimensions` | No | `null` | Group-by fields |
> | `filters` | No | `null` | Advanced `{ "field", "operator", "value" }` predicates |
> | `limit` | No | `null` | Top-N row cap. When set, `next_cursor` is always `null` |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `model_filter` | No | `null` | Narrow to one or more AI models |
> | `persona_filter` | No | `null` | Narrow to one or more personas |
> | `asset_filter` | No | `null` | Narrow to one or more brand or competitor assets |
> | `cursor` | No | `null` | Pagination token from a previous response |
> | `page_size` | No | `500` | First-page size, up to 10,000 |
>
> **Notes**:
>
> * `visibility_score` is a raw decimal. Multiply it by 100 to match the percentage the Profound platform displays.
> * Without an asset filter or `asset_name` in `dimensions`, `visibility_score` is summed across every tracked brand in the category, so the total can exceed 1. To scope the report to one brand, pass `asset_filter` or add `asset_name` to `dimensions`.

> **get\_citations\_report\_v1**
>
> Shows which sources AI engines cite for a category, and how often, over a date range, and returns the legacy response shape. For new integrations, use [`get_citations_report`](#get-citations-report) instead.
>
> Default metrics: `count` and `citation_share`. Dimensions include `hostname`, `path`, `root_domain`, `url`, `model`, `topic`, `prompt`, `tag`, and `persona`.
>
> > [!NOTE]
> >
> > `root_domain_filter` must be paired with `dimensions: ["root_domain"]`.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to report on |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (exclusive) in `YYYY-MM-DD` format |
> | `metrics` | No | `count`, `citation_share` | Metrics to return |
> | `dimensions` | No | `null` | Group-by fields |
> | `filters` | No | `null` | Advanced `{ "field", "operator", "value" }` predicates |
> | `limit` | No | `null` | Top-N row cap. When set, `next_cursor` is always `null` |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `model_filter` | No | `null` | Narrow to one or more AI models |
> | `persona_filter` | No | `null` | Narrow to one or more personas |
> | `root_domain_filter` | No | `null` | Narrow to one or more root domains. Requires `root_domain` in `dimensions` |
> | `hostname_filter` | No | `null` | Narrow to one or more hostnames |
> | `citation_category_filter` | No | `null` | Narrow to one or more citation categories |
> | `cursor` | No | `null` | Pagination token from a previous response |
> | `page_size` | No | `500` | First-page size, up to 10,000 |

### Shopping visibility reports

> [!NOTE]
>
> Shopping analysis data is currently available only for ChatGPT.

Use these tools to understand how brands, products, and retailers appear when AI answers a prompt with shopping mode results.

| Tool | Usage |
| - | - |
| `get_shopping_brands_report` | Brand visibility inside AI shopping results |
| `get_shopping_products_report` | Per-product visibility, position share, and offers |
| `get_shopping_merchants_report` | Which retailers ChatGPT surfaces |
| `get_shopping_trigger_rate_report` | How often prompts return shopping results |

> [!TIP]
>
> Start by asking the assistant how often the category's prompts return shopping results (`get_shopping_trigger_rate_report`). If they rarely do, the other shopping reports have little data to show.

> **get\_shopping\_brands\_report**
>
> Measures how often each brand appears when ChatGPT returns shopping results. This is the shopping counterpart to `get_visibility_report`.
>
> **Example prompts**:
>
> * "Which brands appear most in ChatGPT shopping results for our category?"
> * "How does our shopping visibility compare to competitors, day by day?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category, from `list_categories` |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `group_by` | No | `null` | Group results by `date`, `topic`, `region`, or `prompt` |
> | `metrics` | No | all | Metrics to return |
> | `interval` | No | `day` | Size of each time bucket when grouping by `date`: `day`, `week`, or `month` |
> | `scope` | No | `null` | `owned` for your tracked brands, or `all` to include competitors. Left unset, the report covers your tracked brands |
> | `assets` | No | `null` | Narrow to one or more brands, by name. An `assets` selection returns one page and ignores `limit` |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `persona_filter` | No | `null` | Narrow to one or more personas |
> | `prompt_filter` | No | `null` | Narrow to one or more prompts |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `filter` | No | `null` | Filter expression for conditions the other inputs can't express, such as excluding one region or matching either of two topics. Combined with those inputs using `and` |
> | `limit` | No | `null` | Top-level groups per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes**:
>
> * Every row carries an `asset` object of `{name, owned}`, and asset is always an implicit grouping key: results are one row per asset and group-by bucket.

> **get\_shopping\_products\_report**
>
> Measures individual product visibility inside AI shopping results, one row per product plus any group-by bucket.
>
> **Example prompts**:
>
> * "How visible are our products in ChatGPT shopping results?"
> * "Where is our flagship product sold, and at what price?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category, from `list_categories` |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `group_by` | No | `null` | Group the results by `date`, `topic`, or `prompt`. Can't be combined with `target_product` or `include_merchants` |
> | `metrics` | No | all | Metrics to return |
> | `interval` | No | `day` | Size of each time bucket when grouping by `date`: `day`, `week`, or `month` |
> | `target_product` | No | `null` | Switch to the item view: one product plus its closest competitors. Can't be combined with `group_by` or `include_merchants` |
> | `include_merchants` | No | `false` | `true` adds each product's offers as `merchants`, a list of `{name, price}`, plus `product_url` and `product_image_urls` on the product row. Can't be combined with `group_by` or `target_product` |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `persona_filter` | No | `null` | Narrow to one or more personas |
> | `prompt_filter` | No | `null` | Narrow to one or more prompts |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `brand_filter` | No | `null` | Narrow to one brand's products |
> | `merchant_filter` | No | `null` | Narrow to the products one retailer surfaces |
> | `filter` | No | `null` | Filter expression for conditions the other inputs can't express, such as excluding one region or matching either of two topics |
> | `limit` | No | `null` | Top-level groups per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes and tips**:
>
> * The default metrics are `visibility_score`, `average_position`, `visibility_rank`, `position1_percentage`, `position2_percentage`, `position3_percentage`, `position_above3_percentage`, `product_rating`, and `product_num_reviews`.
> * The `position` metrics are how your assistant tells "always shown, always fourth" from "sometimes shown first". Each one is a raw 0–1 fraction of the product's appearances at that slot, and together they sum to about 1, so 0.30 is 30%.
> * `include_merchants` mode accepts no `group_by` or `target_product`, and the position-frequency metrics aren't available in it.

> **get\_shopping\_merchants\_report**
>
> Measures which retailers appear in a category's shopping results.
>
> **Example prompt**:
>
> * "Which retailers does ChatGPT surface for our category?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category, from `list_categories` |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `view` | No | `distribution` | `distribution`, `brand_share`, or `top_products` |
> | `by_date` | No | `false` | `true` returns a time series. Supported in the `distribution` view only |
> | `metrics` | No | the view's full set | Metrics to return: ones valid for the chosen view |
> | `interval` | No | `day` | Size of each time bucket when grouping by `date`: `day`, `week`, or `month` |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `persona_filter` | No | `null` | Narrow to one or more personas |
> | `prompt_filter` | No | `null` | Narrow to one or more prompts |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `filter` | No | `null` | Filter expression for conditions the other inputs can't express, such as excluding one region or matching either of two topics |
> | `limit` | No | `null` | Top-level groups per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> The `view` input decides what each row represents and which metrics are available.
>
> | View | Rows | Answers |
> | - | - | - |
> | `distribution` | One row per merchant | Which retailers dominate this category |
> | `brand_share` | One row per merchant and brand | Whose products a retailer shows |
> | `top_products` | One row per merchant and product | What a retailer lists most |
>
> Each view accepts its own metrics, and the server rejects anything outside the set:
>
> | View | Metrics |
> | - | - |
> | `distribution` | `merchant_share`, `merchant_share_rank`, `merchant_visibility`, `merchant_visibility_rank` |
> | `brand_share` | `brand_share`, `merchant_share`, `visibility_rank` |
> | `top_products` | `merchant_visibility`, `product_visibility`, `product_rank` |
>
> **Notes and tips**:
>
> * This report has no merchant or product filter, so you can't narrow the results to one retailer. Instead, narrow by topic, region, persona, prompt, or tag, then look up the retailer you care about in the returned rows.
> * The `info.view` metadata field names the view the server applied. `view` is optional and defaults to `distribution`, so if the rows don't look like what you expected, check `info.view` to see which view produced them.

> **get\_shopping\_trigger\_rate\_report**
>
> Measures how often prompts return shopping mode results. This is the denominator behind the other shopping reports, the assistant can use it to explain a thin or empty result.
>
> **Example prompts**:
>
> * "How often do our prompts trigger shopping mode results?"
> * "Which topics trigger shopping mode most often?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category, from `list_categories` |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `group_by` | No | `null` | Group results by `date`, `topic`, `region`, `persona`, or `prompt` |
> | `metrics` | No | all | Metrics to return |
> | `interval` | No | `day` | Size of each time bucket when grouping by `date`: `day`, `week`, or `month` |
> | `topic_filter` | No | `null` | Narrow to one or more topics |
> | `region_filter` | No | `null` | Narrow to one or more regions |
> | `persona_filter` | No | `null` | Narrow to one or more personas |
> | `prompt_filter` | No | `null` | Narrow to one or more prompts |
> | `tag_filter` | No | `null` | Narrow to one or more tags |
> | `filter` | No | `null` | Filter expression for conditions the other inputs can't express, such as excluding one region or matching either of two topics |
> | `limit` | No | `null` | Top-level groups per page, up to 50 |
> | `cursor` | No | `null` | Pagination token from `info.next_cursor` |
>
> **Notes and tips**:
>
> * The default metrics are `total_runs`, `shopping_triggered_runs`, and `trigger_rate_percentage`.
> * Despite its name, `trigger_rate_percentage` is a decimal fraction between 0 and 1, so 0.17 means 17%.
> * Group results by `prompt` or `topic` to see which prompts or topics trigger shopping results most often. Those are the places where shopping visibility is worth optimizing.

## Prompt volume reports

These tools measure demand for a keyword across everything people ask AI platforms. Ask about one keyword at a time; the assistant makes one call per keyword. To see how you appear in the prompts you track, ask for a [visibility report](#get-visibility-report) instead.

| Tool | Usage |
| - | - |
| `get_prompt_volume` | How often people ask AI platforms about a keyword, by week or month |
| `get_prompt_volume_intents` | Why people ask about a keyword, as a share of conversations per intent |

> [!NOTE]
>
> When the assistant reports no estimate for a keyword, that doesn't mean nobody asks about it. Either nothing matched the keyword in that date range, or every slice had two or fewer distinct users and Profound removed it for privacy.

> **get\_prompt\_volume**
>
> Measures projected weekly and monthly volume for a keyword, by country and platform.
>
> **Example prompts**:
>
> * "How many people asked ChatGPT about running shoes last month?"
> * "Is demand for 'project management software' going up this quarter?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `keyword` | Yes | - | Keyword to measure |
> | `matching_type` | Yes | - | `phrase_match` matches all tokens in any order. `exact_match` also requires the tokens to be adjacent and in the order you typed them, so it returns the same results or fewer |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `frequency` | No | `null` | `week` or `month` returns only that frequency. Omit it to get both |
> | `regions` | No | `null` | Narrow to one or more three-letter country codes, such as `USA` |
> | `platforms` | No | `null` | Narrow to one or more platform hostnames: `chatgpt.com`, `gemini.google.com`, or `perplexity.ai` |
> | **Notes**: | | | |
>
> * Figures come back as weekly and monthly projections, one per country and platform. Weekly figures are dated by the Monday of that week, and monthly figures by the first day of the month. A monthly figure exists only for a calendar month that falls fully inside the date range you ask about, never the current month, so a question about this month gets weekly figures only. A monthly figure isn't the sum of its weeks, because a week that crosses a month boundary is split between the two months. When you need a monthly number, ask for one rather than adding up weeks.
> * A broad question, such as every country and platform over several months, can exceed the 10,000-row limit. The assistant then gets incomplete totals and `info.truncated` set to `true`. Narrow the question to a country, a platform, or a shorter date range to get complete figures.
> * Each new keyword counts toward your organization's limit of 1,000 distinct keywords per UTC day. Asking again about a keyword you already looked up that day doesn't count. If you reach the limit, wait and ask about the same keyword again. Rephrasing it uses more of the allowance.

> **get\_prompt\_volume\_intents**
>
> Measures the share of conversations about a keyword that fall under each intent, so you can tell whether people are researching or ready to buy. It doesn't measure how many people ask; `get_prompt_volume` does.
>
> **Example prompts**:
>
> * "Are people researching CRM software or ready to buy?"
> * "What intents drive AI conversations about electric bikes?"
>
> Each intent comes back as a share of the classified conversations, as a fraction from 0 to 1 in `pct_conversation`, so `0.42` is 42%. Shares across all intents add up to 100%. Where an intent has subcategories, each subcategory is its own row (`main_classification` plus `sub_category_classification`), and the share for the main intent is the sum of its subcategories.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `keyword` | Yes | - | Keyword to measure |
> | `matching_type` | Yes | - | `phrase_match` matches all tokens in any order. `exact_match` also requires the tokens to be adjacent and in the order you typed them, so it returns the same results or fewer |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `regions` | No | `null` | Narrow to one or more three-letter country codes, such as `USA` |
> | `platforms` | No | `null` | Narrow to one or more platform hostnames: `chatgpt.com`, `gemini.google.com`, or `perplexity.ai` |
> | **Notes**: | | | |
>
> * Asking about intents doesn't count toward the daily keyword limit, but it shares a burst limit with `get_prompt_volume`. If the assistant is rate-limited, asking about the same keyword again a little later works.

## Traffic reports

These tools are scoped to a tracked domain, not a category. The assistant resolves the domain with [`list_domains`](/mcp/capabilities/discovery-tools#list-domains) first and passes the exact hostname Profound returns.

| Tool | Usage |
| - | - |
| `get_referrals_report` | How many visits did a domain receive from AI engines, and which referrers drove them |
| `get_bots_report` | Which AI crawlers are visiting a domain, and how often |

> **get\_referrals\_report**
>
> Measures visits a domain received from AI engines, such as ChatGPT and Perplexity, over a date range.
>
> Default metric: `visits`. Useful dimensions include `referral_type`, `referral_source`, and `date`.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `domain` | Yes | - | Tracked domain, using the exact hostname from `list_domains` |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `metrics` | No | `visits` | Metrics to return |
> | `dimensions` | No | `null` | Group-by fields, such as `referral_type`, `referral_source`, or `date` |
> | `organization_id` | No | `null` | Disambiguates the domain when the caller belongs to multiple organizations |
> | `filters` | No | `null` | Advanced `{ "field", "operator", "value" }` predicates |
> | `limit` | No | `null` | Top-N row cap. When set, `next_cursor` is always `null` |
> | `referral_source_filter` | No | `null` | Narrow to one or more referrer vendors, such as `openai` |
> | `referral_type_filter` | No | `null` | Narrow to one or more referral categories: `internal`, `referer`, `utm`, or `none` |
> | `cursor` | No | `null` | Pagination token from a previous response |
> | `page_size` | No | `500` | First-page size, up to 10,000 |

> **get\_bots\_report**
>
> Measures AI crawler activity against a domain over a date range, including bots such as GPTBot and PerplexityBot.
>
> **Example prompts**:
>
> * "Which AI crawlers visit our domain?"
> * "Is GPTBot crawling us more since the site update?"
>
> Default metrics: `count` and `citations`. Useful dimensions include `bot_provider`, `bot_name`, `bot_type`, and `date`.
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `domain` | Yes | - | Tracked domain, using the exact hostname from `list_domains` |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `metrics` | No | `count`, `citations` | Metrics to return |
> | `dimensions` | No | `null` | Group-by fields, such as `bot_provider`, `bot_name`, `bot_type`, or `date` |
> | `organization_id` | No | `null` | Disambiguates the domain when the caller belongs to multiple organizations |
> | `filters` | No | `null` | Advanced `{ "field", "operator", "value" }` predicates |
> | `limit` | No | `null` | Top-N row cap. When set, `next_cursor` is always `null` |
> | `bot_provider_filter` | No | `null` | Narrow to one or more providers, such as `openai` or `anthropic` |
> | `bot_name_filter` | No | `null` | Narrow to one or more individual bots, such as `GPTBot` |
> | `bot_type_filter` | No | `null` | Narrow to one or more bot categories: `ai_assistant`, `ai_training`, `index`, or `ai_agent` |
> | `cursor` | No | `null` | Pagination token from a previous response |
> | `page_size` | No | `500` | First-page size, up to 10,000 |

## Resources

Profound MCP also exposes read-only MCP resources: static reference material that an MCP client can load into context.

| Resource URI | Audience | Use it for |
| - | - | - |
| `file:///profound/glossary` | Assistant | Compact index of Profound-specific terms to load once per session |
| `file:///profound/glossary/full` | User | Full glossary with definitions and examples, larger than the index |
| `file:///profound/glossary/{term}` | Assistant | Full definition of one glossary term |
| `file:///profound/sentiment-guide` | Assistant | How to choose a sentiment source and read its metrics, filters, and sorting, before calling `get_sentiment_report` |

The `{term}` slot accepts any slug from the glossary index, which includes the metrics and report concepts behind this page's tools:

| Report | Term resources |
| - | - |
| Visibility | `visibility-score`, `visibility`, `mentions`, `share-of-voice` |
| Citations | `citations`, `citation-share` |
| Sentiment | `sentiment`, `aggregated-sentiment-score`, `sentiment-theme` |
| FactCheck | `factcheck-score`, `inaccurate-claim`, `factcheck-theme` |
| Shopping | `shopping-trigger-rate`, `product-visibility`, `merchant-share` |
| Bots | `bot-tracker` |
| Any report | `date-range`, `prompt-volume`, `executions` |


# Discovery tools
Source: https://docs.tryprofound.com/mcp/capabilities/discovery-tools

Understand the discovery tools available through Profound MCP

Discovery tools resolve human-readable context, such as an organization, category, domain, region, AI model, tag, or topic, into the IDs the other Profound MCP tools take as inputs. Most conversations start here: the assistant confirms what you can access, finds the right scope, and then passes the resolved IDs to other tools. To connect your MCP client to the hosted server, refer to the [Connection guides](/mcp/common-mcp-clients).

## Behavior and safety

All discovery tools are read-only. They retrieve reference data, but don't create, update, or delete anything in Profound.

| Behavior | What it means |
| - | - |
| Non-destructive | No tool performs destructive updates |
| Live data | Tools read from the Profound API, so results reflect the caller's current access and data |

All discovery tools set the following MCP hints:

| Hint | Value | What it means |
| - | - | - |
| `readOnlyHint` | `true` | The tool only reads data |
| `idempotentHint` | `true` | The tool is safe to retry with the same arguments |

## Discovery tools

| Tool | Usage |
| - | - |
| `whoami` | Confirm the authenticated principal, organizations, regions, and entitlements |
| `list_organizations` | List organizations the user can access |
| `list_regions` | List geographic regions, optionally scoped to an organization |
| `list_models` | List tracked AI models such as ChatGPT, Perplexity, Google AI Overviews, and Gemini |
| `list_categories` | List tracked categories for an organization |
| `list_domains` | List tracked domains for an organization |
| `list_tags` | List tags in a category |
| `list_topics` | List topics in a category |
| [`list_prompts`](/mcp/capabilities/prompts-capabilities#list-prompts) | List configured prompts in a category, with optional status, tag, and topic filters. Documented with the other [prompt tools](/mcp/capabilities/prompts-capabilities) |

> **whoami**
>
> Confirms the authenticated user, organizations, regions, and entitlements available to this MCP session.
>
> **Example prompts**:
>
> * "Which Profound account am I signed in as?"
> * "What organizations and regions can I see?"
>
> **Inputs:** none.

> **list\_organizations**
>
> Lists the organizations the authenticated user can access. Returned IDs feed category, domain, and report tools.
>
> **Example prompts**:
>
> * "Which organizations can I pull Profound data from?"
>
> **Inputs:** none.

> **list\_regions**
>
> Lists geographic regions configured for an organization.
>
> **Example prompts**:
>
> * "Which regions do we track?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `org_id` | No | `null` | Organization to scope to. When omitted, shows regions across all accessible organizations |

> **list\_models**
>
> Lists the AI models Profound tracks. Returned model IDs filter reports to a single engine.
>
> **Example prompts**:
>
> * "Which AI models does Profound track?"
>
> **Inputs:** none.

> **list\_categories**
>
> Lists tracked categories, markets, or segments in an organization. Most brand visibility reports are scoped to a category.
>
> **Example prompts**:
>
> * "What categories are we tracking?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `org_id` | Yes | - | Organization whose categories to list |

> **list\_domains**
>
> Lists tracked domains for an organization. Domains are exact hostnames, so `www.example.com` and `example.com` are distinct.
>
> **Example prompts**:
>
> * "Which of our domains does Profound track traffic for?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `org_id` | Yes | - | Organization whose tracked domains to list |

> **list\_tags**
>
> Lists tags available within a category for filtering prompts and reports.
>
> **Example prompts**:
>
> * "What tags exist in the Banking category?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category whose tags to list |

> **list\_topics**
>
> Lists topics available within a category for filtering prompts and reports.
>
> **Example prompts**:
>
> * "What topics do we track under Banking?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to list topics for |

## Resources

Profound MCP also exposes read-only MCP resources: static reference material that an MCP client can load into context.

| Resource URI | Description |
| - | - |
| `file:///profound/glossary` | Compact index of Profound-specific terms to load once per session |
| `file:///profound/glossary/full` | Full glossary with definitions and examples, larger than the index |
| `file:///profound/glossary/{term}` | Full definition of one glossary term |


# Knowledge Bases
Source: https://docs.tryprofound.com/mcp/capabilities/knowledge-base-capabilities

Understand the Knowledge Base tools available through Profound MCP

Profound MCP gives AI assistants tools to list your Knowledge Bases, search their contents, and add new documents to them. To connect your MCP client to the hosted server, refer to the [Connection guides](/mcp/common-mcp-clients).

> [!TIP]
>
> Learn more about Knowledge Bases in the [Profound Help Center](https://help.tryprofound.com/articles/5775538586-knowledge-bases-overview).

## How the tools work together

**1. Find a Knowledge Base**

Ask which Knowledge Bases you have. The assistant lists them with [`list_knowledge_bases`](#list-knowledge-bases), showing each one's name and description.

**2. Search a Knowledge Base**

Ask a question your Knowledge Base can answer. The assistant searches its contents with [`search_knowledge_base`](#search-knowledge-base) and returns the best-matching passages.

**3. Add a document**

Ask to save new content to the Knowledge Base. The assistant drafts the document, previews the plan with [`add_knowledge_base_document`](#add-knowledge-base-document), and adds it once you confirm.

## Behavior and safety

### Behaviors

| Behavior | What it means |
| - | - |
| Preview before apply | `add_knowledge_base_document` defaults to preview mode and adds nothing until the assistant applies the plan with `preview: false` |
| Add-only writes | MCP can only add documents. To revise or delete one, use the Profound platform |
| Duplicate names rejected | Adding a document under a name that already exists fails, and the response suggests retrying under a versioned name |
| Retries create copies | Each applied add creates a new document, so a retry creates another copy unless its name is rejected as a duplicate |
| Live data | Tools read from the Profound API, so results reflect your current access and data |

## Knowledge Base tools

> [!NOTE]
>
> Creating a Knowledge Base is not available through MCP. To add one, [use the Profound platform](https://help.tryprofound.com/articles/5090178560-add-and-manage-knowledge-bases).

| Tool | Usage |
| - | - |
| `list_knowledge_bases` | List the Knowledge Bases you can access |
| `search_knowledge_base` | Search the content inside one Knowledge Base |
| `add_knowledge_base_document` | Add one new markdown document to a Knowledge Base |

> **list\_knowledge\_bases**
>
> Lists the Knowledge Bases available to you.
>
> **Example prompts**:
>
> * "Which Knowledge Bases do we have?"
> * "Do we have a Knowledge Base for brand guidelines?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `organization_id` | Yes, when you have access to more than one | `null` | The organization whose Knowledge Bases to list. The assistant resolves it with [`list_organizations`](/mcp/capabilities/discovery-tools#list-organizations) |

> **search\_knowledge\_base**
>
> Runs a semantic search inside a single Knowledge Base. Each result returns `id`, `score`, `content`, and `metadata`, including `folder_path`, `source_filename`, and `filetype`.
>
> **Example prompts**:
>
> * "What do our brand guidelines say about tone of voice for product launches?"
> * "Find the pricing FAQ in the product Knowledge Base."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `knowledge_base` | Yes | - | The Knowledge Base to search, by ID, name, or slug |
> | `query` | Yes | - | What to search for, in plain language |
> | `top_k` | No | `10` | Maximum number of results to return |
> | `organization_id` | Yes, when you have access to more than one | `null` | The organization the Knowledge Base belongs to. The assistant resolves it with [`list_organizations`](/mcp/capabilities/discovery-tools#list-organizations) |

> **add\_knowledge\_base\_document**
>
> Adds one new document to a Knowledge Base as markdown text.
>
> **Example prompts**:
>
> * "Save this campaign brief to the marketing Knowledge Base."
> * "Add the new positioning doc to the brand guidelines Knowledge Base, under the messaging folder."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `knowledge_base` | Yes | - | The Knowledge Base to add to, by ID, name, or slug |
> | `name` | Yes | - | Document name: lowercase, hyphenated, and descriptive, such as `brand-guidelines` |
> | `text` | Yes | - | The full document content as markdown, with YAML frontmatter at the top |
> | `folder` | No | `null` | Folder to save the document in. When omitted, the document is saved to the Knowledge Base root folder |
> | `preview` | No | `true` | When `true`, return the plan without adding the document; when `false`, add it |
> | `create_missing_parents` | No | `false` | Set to `true` only to confirm creating several nested folder levels at once, after the tool asks to confirm a deeply nested path |
> | `organization_id` | Yes, when you have access to more than one | `null` | The organization the Knowledge Base belongs to. The assistant resolves it with [`list_organizations`](/mcp/capabilities/discovery-tools#list-organizations) |
>
> **Notes**:
>
> * The tool accepts text only. If your source is a URL, the assistant reads the page first, then submits its content as markdown.
> * For whole-site scraping, Notion connections, and Google Drive linking, use the [Profound platform](https://help.tryprofound.com/articles/5090178560-add-and-manage-knowledge-bases).
> * New folders: saving into one new folder inside an existing one works automatically, and the response reports `folder_created`. Saving into several missing folder levels at once fails first, and the assistant retries with `create_missing_parents: true` after you confirm the nesting is intended.
> * Knowledge Base documents are identified by name, so adding a second document with a name that already exists fails.
> * Newly added content can take a moment to appear in search results. It's a good practice to confirm with the assistant that the document was added before performing further actions with it.

## Resources

Profound MCP provides reference material for Knowledge Bases as read-only MCP resources.

| Resource URI | Use it for |
| - | - |
| `file:///profound/knowledge-base-guide` | The authoring playbook: document intents, the minimum quality bar, naming rules, and folder placement |
| `file:///profound/knowledge-base-guide/templates` | A collection of markdown and YAML document templates |
| `file:///profound/glossary/knowledge-base` | The definition of a Knowledge Base |


# Profound Docs
Source: https://docs.tryprofound.com/mcp/capabilities/profound-docs-capabilities

Understand the Profound Docs tools available through Profound MCP

Profound MCP gives AI assistants tools to find, read, write, and manage documents in Profound Docs, the collaborative documents in the Profound platform, such as memos and briefs. To connect your MCP client to the hosted server, refer to the [Connection guides](/mcp/common-mcp-clients).

> [!NOTE]
>
> Documents in Profound Docs are not Knowledge Base documents. To add content that AI assistants and Agents search and retrieve, use the `add_knowledge_base_document` tool instead. Refer to [Knowledge Base](/mcp/capabilities/knowledge-base-capabilities).

## How the tools work together

**1. Find a document**

Ask whether a document exists or which ones changed recently. The assistant searches document names with `list_docs` and links you to each result.

**2. Read it**

Ask for a summary or the full text. The assistant reads the document with `get_doc`, skimming the opening by default and loading the whole body when the task needs it.

**3. Create or revise**

Ask for a new document and the assistant drafts and saves it with `create_doc`. Ask for a change to an existing one and it rewrites the document with `replace_doc_content`.

**4. Rename, reshare, or delete**

Ask to rename a document or change who sees it (`update_doc`), or to delete one the assistant created earlier (`delete_doc`).

## Behavior and safety

Write tools default to preview mode: the assistant receives a plan of the change first, and nothing changes until it applies that plan. The exception is `delete_doc`, which has no preview and deletes on the first call.

### Behaviors

| Behavior | What it means |
| - | - |
| Preview before apply | `create_doc`, `replace_doc_content`, and `update_doc` return a plan first, and report `applied: true` only once a change is applied |
| Whole-document writes | `replace_doc_content` overwrites the entire body and clears comments and tabs on every call, including full-body and partial rewrites |
| Last write wins | The replace tool doesn't check versions. When two writers replace the same document, the later write overwrites the earlier one, with no rollback |
| Rename and reshare restrictions | `update_doc` works only for documents created through MCP or ones you created yourself in the Profound platform. Only a document's creator can change its visibility |
| Delete restrictions | `delete_doc` works only for documents created through MCP. Asking to delete one you made in the Profound platform returns `deleted: false` |
| Live data | Tools read from the Profound API, so results reflect your current access and data |

### MCP hints

The following MCP hints are set on the write tools.

| Tool | Hints | What it means |
| - | - | - |
| `create_doc` | `destructiveHint: false`, `idempotentHint: true` | Creating twice with the same `name` returns the existing document instead of a second copy, and creating never overwrites an existing document |
| `replace_doc_content` | `destructiveHint: true`, `idempotentHint: true` | The call overwrites the entire body and clears comments and tabs |
| `update_doc` | `destructiveHint: true`, `idempotentHint: true` | A rename permanently stops the title from following the first heading, and re-sending `visibility: organization` downgrades existing edit access to view-only |
| `delete_doc` | `destructiveHint: true`, `idempotentHint: true` | A deleted document can't be restored through any API, and a repeat call reports `deleted: false` |

Every write tool also sets `openWorldHint: true`, which tells your MCP client that the writes land in the live, shared environment, so the client may ask you to confirm before running them.

## Profound Docs tools

| Tool | Usage |
| - | - |
| `list_docs` | Find documents by name and read their details |
| `get_doc` | Read one document, as a preview or in full |
| `create_doc` | Create a new document from drafted content |
| `replace_doc_content` | Replace a document's entire body, clearing comments and tabs |
| `update_doc` | Rename a document, change who can see it, or both |
| `delete_doc` | Delete a document this integration created |

> **list\_docs**
>
> Finds documents and their details, such as the name, owner, visibility, and link (`url`). Results are ordered by most recently modified first and never include the document body.
>
> **Example prompts**:
>
> * "Do we already have a doc about our AEO strategy?"
> * "Which Profound docs changed most recently?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `q` | No | `null` | Approximate match on the document name. It doesn't search the document body |
> | `sort` | No | `null` | `recency` is the only accepted value, and it matches the default ordering. Any other value returns an error |
> | `organization_id` | No | `null` | Organization whose documents to list. Defaults to the session's organization (if resolved). Required when the account has access to more than one organization |
> | `limit` | No | `20` | Documents returned per page of results, 1–50. Values outside the range return an error |
> | `cursor` | No | `null` | The `next_cursor` from a previous call, used to fetch the next page. It expires one hour after it's issued |
>
> **Limitations**:
>
> * The tool doesn't filter by document author or owner. To find one person's documents, the assistant lists all documents and reads each row's owner, which can take a while.
> * The tool doesn't return a total document count.

> **get\_doc**
>
> Reads one document. The default preview returns only the opening of the body. When the task needs the full text or the discussion on it, the assistant loads the complete body or the comments.
>
> **Example prompts**:
>
> * "Summarize the AEO strategy memo."
> * "Read the Q3 retro in full, including the comments people left on it."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `doc_id` | Yes | - | The document's ID, from `list_docs` or a document URL |
> | `organization_id` | Yes | - | The organization that owns the document |
> | `preview` | No | `true` | `true` returns the opening of the body; `false` returns the complete body |
> | `include_comments` | No | `false` | `true` also returns comments, each with the comment text and the passage it was left on |

> **create\_doc**
>
> Creates a new document from content the assistant has drafted. The tool takes the complete body in one call.
>
> **Example prompts**:
>
> * "Write a Q3 AEO retro based on this quarter's visibility data and save it as a Profound doc."
> * "Draft a competitor brief and store it in Profound so I can share it later."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `name` | Yes | - | Title for the document, at most 256 bytes. Also its identity within an organization |
> | `content_markdown` | Yes | - | The complete body as Markdown, at most 1,000,000 bytes |
> | `organization_id` | No | `null` | Organization to create the document in. Defaults to the session's organization |
> | `preview` | No | `true` | When `true`, return the plan without creating the document; when `false`, create it |
>
> **Notes**:
>
> * Document titles are unique within an organization: creating a document with a title that already exists returns the existing document and discards the new body.
> * New documents are visible only to their creator.
> * When the call fails between creating the document and storing its body, the document exists but is empty, and the response reports `content_written: false`. A retry returns the same empty document instead of filling it.

> **replace\_doc\_content**
>
> > [!WARNING]
> >
> > Use with caution: this tool replaces the entire document body, clears all tabs except the main tab, and clears all comments.
>
> Replaces a document's content with new text. Any change to an existing document, from ticking a box to rewriting a section, goes through this tool.
>
> **Example prompts**:
>
> * "Add a section on shopping prompts to the AEO strategy memo."
> * "Fix the typos in the Q3 retro."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `doc_id` | Yes | - | The document to replace, from `list_docs` or its URL |
> | `content` | Yes | - | The complete new body as Markdown. An empty value clears the document |
> | `organization_id` | No | `null` | Defaults to the session's organization |
> | `skip_title_sync` | No | `false` | `true` keeps the current title. By default the title follows the new content's first heading |
> | `preview` | No | `true` | When `true`, report what the overwrite would change; when `false`, write |
>
> **Notes**:
>
> * When the document content is replaced, the title updates to match the new content's first heading, unless `skip_title_sync` is `true`.

> **update\_doc**
>
> Renames a document, changes who can see it, or both in one call.
>
> **Example prompts**:
>
> * "Rename the strategy memo to 'AEO strategy H2'."
> * "Make the Q3 retro visible to the whole organization."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `doc_id` | Yes | - | The document to rename or reshare |
> | `name` | No | `null` | New title, at most 256 bytes. Omit when the title isn't changing |
> | `visibility` | No | `null` | `invited_only` or `organization`, which is view-only. Omit when sharing isn't changing |
> | `organization_id` | No | `null` | Defaults to the session's organization |
> | `preview` | No | `true` | When `true`, return the plan without changing anything; when `false`, apply |
>
> **Notes**:
>
> * At least one of `name` or `visibility` is required. A call with neither returns an error.
> * A document rename doesn't change the document's first heading, and permanently stops the title from following it. For example, if you rename "AEO strategy" to "AEO strategy H2", the first heading of the document still says "AEO strategy".
> * View-only is the sole access level that `update_doc` can grant to everyone at an organization. If everyone at an organization already has edit access, setting `visibility` to `organization` through MCP downgrades that access to view-only.
> * The tool works only for documents created through MCP or ones you created yourself in the Profound platform. A document shared with you, even with edit access, can't be renamed or reshared, and only its creator can change its visibility.

> **delete\_doc**
>
> Deletes a document the assistant created through an MCP connection earlier, such as a draft that turned out wrong or a duplicate. This is the one write tool with no preview mode.
>
> **Example prompts**:
>
> * "Delete the draft you created earlier, we're keeping the final version."
> * "That brief turned out to be a duplicate, remove it."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `doc_id` | Yes | - | The document to delete, from `list_docs` or its URL |
> | `organization_id` | No | `null` | Defaults to the session's organization |
>
> **Notes**:
>
> * Deletion works only for documents created through MCP, and only by their creator.
> * Deletions are final: you can't undo them.

## Resources

| Resource URI | Description |
| - | - |
| `file:///profound/docs-guide` | The authoring playbook: how a document differs from a Knowledge Base document, when to create and when to replace, how to safely overwrite a whole document, and who can see a new document |
| `file:///profound/glossary/doc` | The definition of a Profound Doc |


# Projects
Source: https://docs.tryprofound.com/mcp/capabilities/projects-capabilities

Understand the Projects tools available through Profound MCP

Profound MCP gives AI assistants tools to read, generate, and manage Projects and their tasks. To connect your MCP client to the hosted server, refer to the [Connection guides](/mcp/common-mcp-clients).

> [!NOTE]
>
> Profound MCP works with the legacy version of Projects. Projects you create in the Profound platform aren't available through MCP yet. Learn more about Projects in [Profound Help Center](https://help.tryprofound.com/articles/8581387448-about-projects).

## How the tools work together

**1. Resolve the Project category**

Projects are scoped to a category, so before doing anything with Projects, the assistant resolves the category with [`list_categories`](/mcp/capabilities/discovery-tools#list-categories).

**2. Find a Project**

Ask what work exists in a category. The assistant lists its Projects with [`list_projects`](#list-projects), narrowing to active or archived work when you ask for it.

**3. Read it**

Ask about one Project. The assistant reads the goal, the resolved identity, and the full task list with [`get_project`](#get-project).

**4. Generate a new Project**

Ask for a new Project. The assistant previews the generation plan with [`new_project`](#new-project), starts generation once you confirm, and follows progress with [`get_project_status`](#get-project-status). Generation runs for around 10 minutes.

**5. Manage tasks**

Ask to add, edit, or progress tasks. The assistant uses [`create_project_task`](#create-project-task), [`update_project_task`](#update-project-task), and [`set_project_task_status`](#set-project-task-status). When a Project is finished or no longer relevant, ask to archive it: [`archive_project`](#archive-project) sets it aside without deleting it.

## Behavior and safety

### Behaviors

Every write tool except `delete_project_task` defaults to preview mode: the assistant receives a plan of the change first, shows it to you, and applies the change only after you confirm it.

| Behavior | What it means |
| - | - |
| Preview before apply | Write tools return a plan of the change and change nothing until the assistant applies it |
| Archive over delete | `archive_project` marks a Project `retired` and preserves its history |
| Full-replace briefs | A `brief` passed to `update_project_task` replaces the existing brief instead of merging into it |
| Destructive deletes | `delete_project_task` permanently deletes a task, with no preview and no undo |
| Clear text writes | The tools that take text reject raw metrics, so assistants replace them with human-readable values, for example `33%` instead of `0.33` for a Visibility Score. The saved task wording can differ from the text you provided |
| Live data | Tools read from the Profound API, so results reflect your current access and data |

### MCP hints

The read tools (`list_projects`, `get_project`, and `get_project_status`) are marked `readOnlyHint: true` and `idempotentHint: true`: they only read data and are safe to retry. The following MCP hints are set on the write tools:

| Tool | Hints | What it means |
| - | - | - |
| `new_project` | `destructiveHint: false`, `idempotentHint: false` | Applying twice starts two generations and creates two Projects |
| `archive_project` | `destructiveHint: false`, `idempotentHint: true` | Archiving an already-retired Project changes nothing |
| `create_project_task` | `destructiveHint: false`, `idempotentHint: false` | Applying twice adds the task twice |
| `update_project_task` | `destructiveHint: false`, `idempotentHint: true` | Applying the same change again leaves the task as it is |
| `set_project_task_status` | `destructiveHint: false`, `idempotentHint: true` | Setting the same status again changes nothing |
| `delete_project_task` | `destructiveHint: true`, `idempotentHint: true` | The delete is permanent, and a repeat call has nothing left to delete |

Every write tool also sets `openWorldHint: true`, which tells your MCP client that the writes land in the live, shared environment, so the client may ask you to confirm before running them.

## Projects tools

| Tool | Usage |
| - | - |
| `list_projects` | List a category's Projects |
| `get_project` | Read one Project with its goal and full task list |
| `get_project_status` | Check a Project's lifecycle and live generation status |
| `new_project` | Generate a new Project in a category |
| `archive_project` | Retire a Project without deleting it |
| `create_project_task` | Add a task to a Project |
| `update_project_task` | Update fields on an existing task |
| `set_project_task_status` | Change a task's status |
| `delete_project_task` | Permanently delete a task |

> **list\_projects**
>
> Lists a category's Projects and their details, including title, number of tasks, and the last updated date.
>
> **Example prompts**:
>
> * "What projects do we have in the Banking category?"
> * "Show me the projects we're actively tracking."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | The Profound category, from `list_categories` |
> | `status` | No | `null` | Filter to `suggested`, `tracked`, or `retired`. Leaving it blank returns all Projects |
> | `limit` | No | `20` | Maximum Projects to return |
> | `offset` | No | `0` | Offset for pagination |
>
> **Notes**
>
> * The tool doesn't return a total Project count.

> **get\_project**
>
> Reads one Project, including the title, the goal, the description, and the tasks.
>
> **Example prompts**:
>
> * "Walk me through the 'Rank top three on savings accounts' project."
> * "What tasks are still open on that project?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project ID, from `list_projects` |
> | `category_id` | Yes | - | The category of the Project |
>
> > [!TIP]
> >
> > Ask your assistant to compare live metrics against the Project target. The data `get_project` returns pairs well with the [analytics tools](/mcp/capabilities/analytics-capabilities).

> **get\_project\_status**
>
> Gets a Project's lifecycle status:
>
> * `suggested` for proposed work
> * `tracked` for active work measured against the goal
> * `retired` for archived work
>
> For Projects that were recently generated or are currently being generated, this tool also returns the `live_generation` state: `queued`, `running`, `completed`, or `failed`.
>
> **Example prompts**:
>
> * "Is the project you started generating ready yet?"
> * "Did that project generation finish or fail?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project ID, from `list_projects` or `new_project`. Not a generation `run_id` |
> | `category_id` | Yes | - | The category of the Project |

> **new\_project**
>
> Starts generating a Project. The preview shows the generation plan, and once you confirm it, Profound builds the Project in the background and returns a link to it. Building takes around 10 minutes, and the assistant can check whether it's done with `get_project_status`.
>
> **Example prompts**:
>
> * "Create a project to improve our visibility on savings accounts prompts."
> * "Generate a project for earned media on the checking topic, focused on the US market."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | The category to generate in, from `list_categories` |
> | `focus` | Yes | - | One concrete sentence naming the Project to generate. Shares a 2,000-character limit with `context` |
> | `topics` | No | `null` | Topic names the Project should focus on |
> | `project_name` | No | `null` | Title for the Project. If left blank, Profound generates the name |
> | `work_types` | No | `null` | Kind of work to produce: `publish`, `refresh`, `earned`, `creative`, or `social` |
> | `date_range` | No | `null` | Time period the generated work should cover |
> | `platforms` | No | `null` | Platforms or channels to prioritize |
> | `regions` | No | `null` | Markets or regions to prioritize |
> | `tags` | No | `null` | Category tags to guide what gets generated |
> | `context` | No | `null` | Extra detail to guide what gets generated: objective, audience, competitors, evidence, desired output, and constraints. Shares a 2,000-character limit with `focus` |
> | `preview` | No | `true` | When `true`, show the plan without creating anything; when `false`, start generation |

> **archive\_project**
>
> Archives a Project, which sets it aside without deleting it. The Project and its history are preserved and marked `retired`, so the change is reversible.
>
> **Example prompts**:
>
> * "Archive the holiday campaign project, it's done."
> * "Retire the projects we're no longer working on."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project ID |
> | `category_id` | Yes | - | The category of the Project |
> | `reason` | No | `null` | Short reason for archiving, for example no longer relevant, already done, or superseded |
> | `preview` | No | `true` | When `true`, return the plan without changing the Project; when `false`, apply |

> **create\_project\_task**
>
> Adds a task to a Project.
>
> **Example prompts**:
>
> * "Add a task to refresh the pricing page to the savings project."
> * "Add a task to the 'Rank top three on savings accounts' project: promote the new rates announcement on LinkedIn. Mark it high impact."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project to add the task to |
> | `category_id` | Yes | - | The category of the Project |
> | `title` | Yes | - | The task title, 5–12 words |
> | `summary` | No | `null` | The task's summary |
> | `brief` | No | `null` | The task's working instructions, sent as a structured brief object |
> | `type` | No | `null` | Task type: `publish`, `refresh`, `earned`, `competitor_copy`, `creative`, or `social` |
> | `topic` | No | `null` | The topic this task targets |
> | `impact` | No | `null` | Impact score, 1–5, where 5 is the highest |
> | `reference_url` | No | `null` | A supporting link, such as the page to work on or a source to draw from |
> | `reference_label` | No | `null` | Display name for the reference link |
> | `position` | No | `null` | The task's position in the Project's task list |
> | `preview` | No | `true` | When `true`, show the planned task without creating it; when `false`, create it |

> **update\_project\_task**
>
> Updates the fields on an existing task.
>
> **Example prompts**:
>
> * "Update the pricing page task: the new rates page is live now, link it as the reference."
> * "Reword the summary of the LinkedIn task so it focuses on the new savings rates."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project ID |
> | `task_id` | Yes | - | The task ID, from `get_project`'s task list |
> | `category_id` | Yes | - | The category of the Project |
> | `title` | No | `null` | New title |
> | `summary` | No | `null` | New summary paragraph |
> | `brief` | No | `null` | New task instructions, sent as a structured brief object |
> | `type` | No | `null` | New task type |
> | `topic` | No | `null` | New target topic |
> | `impact` | No | `null` | New impact score, 1–5 |
> | `reference_url` | No | `null` | New supporting link for the task |
> | `reference_label` | No | `null` | New display name for that link |
> | `preview` | No | `true` | When `true`, show the planned changes; when `false`, apply |
>
> **Notes**:
>
> * A new `brief` replaces the old one entirely, so the assistant submits the complete brief even for a small change.

> **set\_project\_task\_status**
>
> Changes a task's status. The status options are `not_started`, `in_progress`, `done`, and `abandoned`. The `abandoned` status doesn't delete tasks.
>
> **Example prompts**:
>
> * "Mark the pricing refresh task as done."
> * "We're pausing the earned media task, shelve it."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project ID |
> | `task_id` | Yes | - | The task ID |
> | `category_id` | Yes | - | The category of the Project |
> | `status` | Yes | - | The new task status |
> | `note` | No | `null` | Note recorded with the status change |
> | `preview` | No | `true` | When `true`, return the proposed status change; when `false`, apply |

> **delete\_project\_task**
>
> Permanently deletes a task from a Project. Destructive and irreversible, with no preview step.
>
> > [!TIP]
> >
> > To archive a task without losing it, ask to change its status to `abandoned` instead.
>
> **Example prompts**:
>
> * "Delete the duplicate task you just added."
> * "Remove that test task for good."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `project_id` | Yes | - | The Project ID |
> | `task_id` | Yes | - | The task ID to delete |
> | `category_id` | Yes | - | The category of the Project |

## Resources

| Resource URI | Description |
| - | - |
| `file:///profound/projects-guide` | The Projects playbook: goals and measurement, the resolved identity block, task types and brief schemas, the impact rubric, and the `new_project` interview |
| `file:///profound/projects-guide/brief-examples` | Complete examples of a Project brief and a task brief |


# Prompts
Source: https://docs.tryprofound.com/mcp/capabilities/prompts-capabilities

Understand the prompt tools available through Profound MCP

Profound MCP gives AI assistants tools to see which prompts a category tracks, read the answers AI engines gave to them, create new prompts, update existing ones, and activate, disable, or delete them in bulk. To connect your MCP client to the hosted server, refer to the [Connection guides](/mcp/common-mcp-clients).

## How the tools work together

**1. Resolve the category**

Prompts are scoped to a category, so before doing anything with prompts, the assistant resolves the category with [`list_categories`](/mcp/capabilities/discovery-tools#list-categories).

**2. See what you track**

Ask which prompts the category runs. The assistant lists them with [`list_prompts`](#list-prompts), narrowing by status, topic, or tag when you ask for it.

**3. Read the answers**

Ask what AI engines say. The assistant retrieves the raw answers behind the metrics with [`get_prompt_answers`](#get-prompt-answers).

**4. Create or revise**

Ask to track new prompts and the assistant drafts and previews them with [`create_prompts`](#create-prompts). Ask for a change to existing prompts and it previews field-by-field changes with [`update_prompts`](#update-prompts).

**5. Activate, disable, or delete**

Ask to pause prompts, resume them, or remove them for good. The assistant transitions them in bulk with [`update_prompts_status`](#update-prompts-status).

## Behavior and safety

Write tools default to preview mode: the assistant receives a plan of the change first, and nothing persists until it applies that plan with `preview: false`.

### Behaviors

| Behavior | What it means |
| - | - |
| Preview before apply | `create_prompts`, `update_prompts`, and `update_prompts_status` return a plan first and persist nothing until the change is applied |
| Creates aren't idempotent | Each applied `create_prompts` call creates new prompts, so applying twice with the same input creates duplicates |
| Collections replace | On update, `tags`, `regions`, `platforms`, and `personas` replace the whole set instead of merging into it |
| Deletion is permanent | `update_prompts_status` with `deleted` removes prompts and their historical data |
| Auto-created references | When a prompt names a topic or tag that doesn't exist yet, the tool creates it instead of returning an error |
| Live data | Tools read from the Profound API, so results reflect your current access and data |

### MCP hints

The read tools (`list_prompts` and `get_prompt_answers`) only retrieve data and change nothing in Profound. The following MCP hints are set on the write tools:

| Tool | Hints | What it means |
| - | - | - |
| `create_prompts` | `idempotentHint: false` | Retrying an applied call creates more prompts |
| `update_prompts` | `destructiveHint: true`, `idempotentHint: true` | `tags`, `regions`, `platforms`, and `personas` are replaced, and the same input applied twice yields the same state |
| `update_prompts_status` | `destructiveHint: true`, `idempotentHint: true` | `deleted` is permanent, and prompts that already have the target status are skipped |

## Prompt tools

| Tool | Usage |
| - | - |
| `list_prompts` | List a category's prompts, with status, tag, and topic filters |
| `get_prompt_answers` | Read the answers AI engines gave for a category's prompts |
| `create_prompts` | Create one or more prompts in a category |
| `update_prompts` | Update fields on existing prompts |
| `update_prompts_status` | Activate, disable, or delete prompts in bulk |

> **list\_prompts**
>
> Lists the prompts configured in a category, with their IDs and details. The other prompt tools take prompt IDs from this list.
>
> **Example prompts**:
>
> * "Which prompts are we tracking in the Banking category?"
> * "Show me the disabled prompts under the savings accounts topic."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to list prompts for |
> | `status` | No | `all` | One of `active`, `disabled`, or `all` |
> | `tag_ids` | No | `null` | Include only prompts carrying these tags |
> | `topic_ids` | No | `null` | Include only prompts under these topics |
> | `exclude_tag_ids` | No | `null` | Exclude prompts carrying these tags |
> | `exclude_topic_ids` | No | `null` | Exclude prompts under these topics |
> | `combine` | No | `AND` | How `tag_ids` and `topic_ids` combine when both are set: `AND` or `OR` |
> | `limit` | No | `100` | Maximum prompts to return |
> | `cursor` | No | `null` | Pagination token from a previous response's `next_cursor` |

> **get\_prompt\_answers**
>
> Retrieves the answers AI engines gave for a category's prompts over a date range: the raw text behind the visibility metrics.
>
> **Example prompts**:
>
> * "What did AI engines answer for our savings prompts last week?"
> * "Pull the answers behind last month's visibility drop."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Category to retrieve prompt answers for |
> | `start_date` | Yes | - | Window start (inclusive) in `YYYY-MM-DD` format |
> | `end_date` | Yes | - | Window end (inclusive) in `YYYY-MM-DD` format |
> | `prompt_id` | No | `null` | Narrow the results to a single prompt |
> | `country` | No | `null` | Narrow to one or more countries, by name or ID. Can't be combined with `region` |
> | `region` | No | `null` | Narrow to one or more regions, by name or ID. Can't be combined with `country` |
> | `topic_ids` | No | `null` | Narrow to one or more topics, by ID from `list_topics` |
> | `model_ids` | No | `null` | Narrow to one or more answering AI models, by ID from `list_models` |
> | `limit` | No | `100` | Number of prompts returned per page of results |
> | `offset` | No | `0` | Offset for pagination |
>
> **Notes**:
>
> * Several IDs in one list match any of those IDs. When both inputs are set, an answer has to match both.
> * Filters apply before pagination, so a filtered call pages through matching answers only.

> **create\_prompts**
>
> Creates one or more prompts in a category. Each prompt names the question to ask, the topic it belongs to, and the platforms and regions to run it on.
>
> **Example prompts**:
>
> * "Start tracking 'What are the best running shoes for flat feet?' on ChatGPT in the US."
> * "Create prompts for these questions under the savings accounts topic."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category, from `list_categories` |
> | `prompts` | Yes | - | The prompts to create |
> | `preview` | No | `true` | When `true`, show what would happen; when `false`, create |
>
> Each entry in `prompts` accepts:
>
> | Field | Required | Default | Description |
> | - | - | - | - |
> | `prompt` | Yes | - | The prompt text |
> | `topic` | Yes | - | Topic ID or name. A name that doesn't exist is created |
> | `platforms` | Yes | - | Platform IDs or names to run the prompt on |
> | `regions` | Yes | - | Region IDs or names to run the prompt in |
> | `language` | No | `en-US` | Prompt language |
> | `tags` | No | `[]` | Tag IDs or names. A name that doesn't exist is created |
> | `analysis_types` | No | `["visibility"]` | Analysis types to run: `visibility`, [`sentiment`](https://help.tryprofound.com/articles/3189907319-about-sentiment), or [`accuracy`](https://help.tryprofound.com/articles/5793584301-about-factcheck) |
> | `personas` | No | `[]` | Persona IDs or names |
> | `asset` | No | `null` | Asset the prompt is associated with |
>
> **Notes**:
>
> * Topic and tag names are matched exactly: a misspelled name creates a new topic or tag instead of joining the existing one, splitting your data between the two. The preview counts these in `topics_created` and `tags_created`, so check it for unintended new topics or tags before confirming.
> * Creates aren't idempotent: applying the same call twice creates duplicate prompts.

> **update\_prompts**
>
> Updates one or more existing prompts. The preview returns a field-by-field comparison per prompt: old and new values for single fields, and added and removed values for collections.
>
> **Example prompts**:
>
> * "Move the pricing prompts to the 'Fees' topic."
> * "Tag the savings prompts with 'high priority'."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category that owns the prompts |
> | `prompts` | Yes | - | The prompt updates to apply |
> | `preview` | No | `true` | When `true`, return the planned changes; when `false`, apply |
>
> Each entry in `prompts` accepts:
>
> | Field | Required | Default | Description |
> | - | - | - | - |
> | `id` | Yes | - | The prompt to update, from `list_prompts` |
> | `prompt` | No | `null` | New prompt text |
> | `topic` | No | `null` | New topic |
> | `tags` | No | `null` | New tag set |
> | `regions` | No | `null` | New region set |
> | `platforms` | No | `null` | New platform set |
> | `personas` | No | `null` | New persona set |
> | `analysis_types` | No | `null` | New analysis types: `visibility`, [`sentiment`](https://help.tryprofound.com/articles/3189907319-about-sentiment), or [`accuracy`](https://help.tryprofound.com/articles/5793584301-about-factcheck) |
> | `language` | No | `null` | New language |
> | `asset` | No | `null` | New associated asset |
>
> **Notes**:
>
> * For `tags`, `regions`, `platforms`, and `personas` fields, passing `null` doesn't change the value, `[]` clears it, and a non-empty list replaces the whole set. To add one tag, the assistant submits the complete tag set including the existing tags.

> **update\_prompts\_status**
>
> Changes the status of prompts:
>
> * `active` - resumes daily runs
>
> * `disabled` - stops future runs and preserves history
>
> * `deleted` - permanently deletes the prompts and their historical data
>
> > [!WARNING]
> >
> > Use the `deleted` status with caution: it permanently removes the prompts and their historical data.
>
> **Example prompts**:
>
> * "Disable the holiday prompts, we'll bring them back in November."
> * "Delete the test prompts you created earlier."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `category_id` | Yes | - | Tracked category that owns the prompts |
> | `prompt_ids` | Yes | - | IDs of the prompts to change status for |
> | `status` | Yes | - | `active`, `disabled`, or `deleted` |
> | `preview` | No | `true` | When `true`, return the plan; when `false`, apply |
>
> **Notes**:
>
> * Applying the status to prompts that already have this same status doesn't change anything.

## Resources

Profound MCP provides reference material for prompts as read-only MCP resources.

| Resource URI | Description |
| - | - |
| `file:///profound/glossary/prompt` | The definition of a prompt, the individual query sent to an AI answer engine |
| `file:///profound/glossary/prompt-generation-model` | The definition of the Prompt Generation Model, which varies phrasing and context in the prompts a topic runs |
| `file:///profound/glossary/responses` | The definition of responses, the full answers AI engines gave to a prompt |


# Workflows
Source: https://docs.tryprofound.com/mcp/capabilities/workflows-capabilities

Understand the Workflow tools available through Profound MCP

Profound MCP gives AI assistants tools to build, manage, and run Profound Workflows directly from an MCP client. To connect your MCP client to the hosted server, refer to the [Connection guides](/mcp/common-mcp-clients).

> [!NOTE]
>
> The tool names say agent, but they build and run [Workflows](/agents/workflows/about-workflows), the node-based automations from the Workflow Builder. Agents built with AI Marketer (Aim) aren't available through MCP yet.

## How the tools work together

### Run a Workflow

**1. Find a Workflow**

Ask which Workflows exist. The assistant browses the organization's Workflows with [`list_agents`](#list-agents), or pre-built templates with [`list_agent_definition_templates`](#list-agent-definition-templates).

**2. Inspect inputs**

Ask what a Workflow needs to run. The assistant reads its `input_schema` with [`get_agent`](#get-agent) before starting a run.

**3. Run the Workflow**

Ask to run the Workflow. The assistant starts the run with [`run_agent`](#run-agent), passing inputs that match the Workflow's `input_schema`.

**4. Check the results**

Ask whether the run is done. The assistant checks the run state with [`get_agent_run`](#get-agent-run).

### Build a Workflow

**1. Open a build session**

Ask to build a Workflow. The assistant opens a build session with [`start_agent_build_session`](#start-agent-build-session) and includes the returned `agent_build_session_id` in every later build call, so the whole build attempt is tracked as one session.

**2. Explore building blocks**

The assistant assembles a workflow graph with [`list_agent_node_types`](#list-agent-node-types) and [`get_agent_node_schema`](#get-agent-node-schema), or starts from a template with [`list_agent_definition_templates`](#list-agent-definition-templates).

**3. Bind connected accounts**

When the graph includes a node that runs on a connected account, the assistant fills that node's `integration_id` with an ID from [`list_integrations`](#list-integrations).

**4. Create or update a draft**

The assistant previews the plan with [`create_agent_definition`](#create-agent-definition) or [`update_agent_definition`](#update-agent-definition), and saves the draft once you confirm.

**5. Validate**

The assistant catches issues in the Workflow's structure early with [`validate_agent_definition`](#validate-agent-definition) and fixes them with `update_agent_definition`. This is a fast check before publishing the Workflow.

**6. Publish**

Ask to make the Workflow live. The assistant previews the publish plan with [`publish_agent_definition`](#publish-agent-definition) and applies it once you confirm.

#### Use natural language when working with Claude

If you're using Claude as the client, you can prompt it in natural language to build Workflows in one go, without making it step through various tools manually.

Here's an example prompt and a Workflow structure Claude may produce with it:

```text wrap theme={null}
Build a Workflow in Profound that creates a structured, AEO-optimized article based on provided inputs. It should analyze top-cited pages, live Google results, and existing brand content to produce a well-researched article
```

![An example diagram outlining the structure of a Workflow Claude generated with the prompt above](https://mintcdn.com/profound-37face47/tkHhxzpFTsYeVbgR/images/mcp/example-agent-diagram.png?fit=max\&auto=format\&n=tkHhxzpFTsYeVbgR\&q=85\&s=e18bad1c1a9a47296b4ff4fcb73e3bf3)

## Behavior and safety

Workflow tools can create and update Workflow definitions and start Workflow runs. The create, update, and publish tools default to preview mode: the assistant receives a plan of the change first and applies it once you confirm.

| Behavior | What it means |
| - | - |
| Preview before apply | Create, update, and publish tools default to preview mode |
| Live data | Tools read from the Profound API, so results reflect the caller's current access and data |

## Workflow tools

| Tool | Usage |
| - | - |
| `list_agents` | List Workflows available in an organization |
| `get_agent` | Get details of a specific Workflow, including its input schema |
| `run_agent` | Start a Workflow run |
| `get_agent_run` | Check the status and output of a previously started run |
| `start_agent_build_session` | Open a build session to build or revise a Workflow |
| `list_agent_node_types` | List the node types available for building a Workflow graph |
| `get_agent_node_schema` | Get the configuration schema for a specific node type |
| `list_integrations` | List the integrations your organization has connected, to fill a node's `integration_id` |
| `list_agent_definition_templates` | Browse pre-built Workflow templates to use as starting points |
| `get_agent_definition` | Read back a Workflow's full workflow graph |
| `create_agent_definition` | Create a new draft Workflow definition |
| `update_agent_definition` | Update an existing draft Workflow definition |
| `validate_agent_definition` | Check whether a draft Workflow definition is valid and publishable |
| `publish_agent_definition` | Publish a draft Workflow definition so it goes live |

> **list\_agents**
>
> Lists the Workflows defined in your organization.
>
> **Example prompts**:
>
> * "Which Workflows do we have published?"
> * "Show me our draft Workflows."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `statuses` | No | `["published"]` | Workflow statuses to include: `published`, `draft`, or both |
> | `cursor` | No | - | Pagination cursor from a previous page |
> | `limit` | No | - | Maximum number of Workflows to return |

> **get\_agent**
>
> Gets details of a specific Workflow, including its `input_schema`. The assistant reads this before calling `run_agent` to confirm which inputs the Workflow expects.
>
> **Example prompts**:
>
> * "What inputs does the article writer Workflow need?"
> * "What does the weekly report Workflow produce?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the Workflow to retrieve |
> | `version` | No | `published` | `published` for the live version, `draft` for the latest unpublished changes |

> **run\_agent**
>
> Starts a Workflow run. Returns a run ID the assistant can use in `get_agent_run`.
>
> **Example prompts**:
>
> * "Run the article writer Workflow on the topic 'AI search trends'."
> * "Kick off the weekly report Workflow."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the Workflow to run |
> | `inputs` | Yes | - | Input values from the Workflow's `input_schema` |
>
> **Notes**:
>
> * Each key in `inputs` is a property ID from the Workflow's `input_schema`. The human-readable label for each property lives in that property's `title`.

> **get\_agent\_run**
>
> Gets the status and outputs of a previously started Workflow run. The assistant checks this repeatedly until the run ends, whether it succeeds or fails.
>
> **Example prompts**:
>
> * "Is that Workflow run finished yet?"
> * "What did the article writer Workflow produce?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the Workflow the run belongs to |
> | `run_id` | Yes | - | Run ID returned by `run_agent` |
> | `verbose` | No | `false` | When `true`, include each step's raw output in the response |

> **start\_agent\_build\_session**
>
> Opens a Workflow build session, the first step before building or revising a Workflow. The assistant calls it before any other build tool, then includes the returned `agent_build_session_id` in every later build call.
>
> **Example prompts**:
>
> * "Build a Workflow that drafts AEO-optimized articles from our citation data."
> * "Let's rework the weekly report Workflow."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `intent` | Yes | - | Short, plain-language description of what the Workflow should do |

> **list\_agent\_node\_types**
>
> Lists the node types available for building a Workflow. Returns each type's `node_type` identifier, display name, and a one-line description. The assistant calls this after opening a build session, when assembling a new Workflow.
>
> **Example prompts**:
>
> * "What building blocks can the Workflow use?"
> * "Is there a node that runs a web search?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `kind` | No | `null` | Narrow the list by which credential the node runs on: `third_party`, `platform_key`, or `native`. Omit for the full list |
>
> **Notes**:
>
> * `third_party` nodes run on an account your organization connects, such as WordPress or Slack. `platform_key` nodes run on an external research vendor Profound has an agreement with, such as Exa or Perplexity, so there's nothing for you to connect. `native` nodes are Profound's own: the structural nodes, such as `llm` and `code`, and the `profound_*` nodes that read your Profound data.

> **get\_agent\_node\_schema**
>
> Gets the configuration schema and examples for a specific node type. The assistant uses the returned schema to fill a node's `config` correctly when building or editing a Workflow.
>
> **Example prompts**:
>
> * "How is the LLM node configured?"
> * "What settings does the code node take?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `node_type` | Yes | - | Node type identifier from `list_agent_node_types` |

> **list\_integrations**
>
> Lists the integrations your organization has connected and made available to Workflows. Returns an `integration_id` for each connection, which the assistant uses for integration nodes before publishing the Workflow.
>
> **Example prompts**:
>
> * "Which integrations can I use in my Workflow?"
> * "Is our WordPress connected, so the Workflow can publish to it?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `provider` | No | `null` | Filter results by an integration provider, such as `wordpress` or `google_gmail`. An empty result means no integrations from that provider are connected |
>
> **Notes**:
>
> * The tool returns only active connections.
> * Each connection returns `integration_id`, `provider`, `account`, `label`, `status`, and `level`, which is `org` or `user`.
> * If you have several connections from the same integration provider (for example, two different WordPress sites for a blog and for documentation), the `account` field tells the connections apart.
> * Google Docs, Google Slides, and Google Sheets use the organization's Google Drive connection, so a returned `google_drive` connection covers nodes from all three. Gmail and Google Search Console each have a connection of their own.

> **list\_agent\_definition\_templates**
>
> Browses the public catalog of pre-built Workflow templates. Each template includes a plain-language goal, the inputs it needs, what it produces, and a skeleton workflow to use as a starting point. The assistant calls this before building a new Workflow from scratch.
>
> **Example prompts**:
>
> * "What Workflow templates can we start from?"
> * "Is there a template for a content refresh workflow?"
>
> **Inputs:** none.

> **get\_agent\_definition**
>
> Reads a Workflow's definition graph in the same format that `create_agent_definition` and `update_agent_definition` accept. The assistant uses this when copying an existing Workflow or inspecting a node type you want to replicate.
>
> **Example prompts**:
>
> * "Show me how the article writer Workflow is built."
> * "Copy the weekly report Workflow as a starting point for a new one."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the Workflow to read |
> | `version` | No | `published` | `published` for the live version, `draft` for the latest unpublished changes |

> **create\_agent\_definition**
>
> Creates a new draft Workflow definition. The preview returns a plain-language plan and workflow diagram without saving anything, and the assistant saves the draft once you confirm.
>
> **Example prompts**:
>
> * "Save that workflow as a draft Workflow."
> * "Create the Workflow we just designed."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `name` | Yes | - | Display name for the Workflow |
> | `description` | Yes | - | What the Workflow does |
> | `organization_id` | Yes | - | Organization to create the Workflow in, from `list_organizations` |
> | `graph` | Yes | - | The workflow graph as a `{ nodes, edges }` object |
> | `agent_build_session_id` | Yes | - | Session ID from `start_agent_build_session` |
> | `preview` | No | `true` | When `true`, return a plan without saving; when `false`, save the draft |

> **update\_agent\_definition**
>
> Updates an existing draft Workflow definition. The assistant uses this tool to fix validation issues after `create_agent_definition` or to iterate on a saved draft. Displays the preview first and applies it after you confirm.
>
> **Example prompts**:
>
> * "Fix the validation issues and update the draft."
> * "Add a web search step to the draft Workflow."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the draft Workflow to update |
> | `graph` | Yes | - | The updated definition graph |
> | `agent_build_session_id` | Yes | - | Session ID from `start_agent_build_session` |
> | `preview` | No | `true` | When `true`, return a plan without saving; when `false`, save |

> **validate\_agent\_definition**
>
> Checks whether a saved draft Workflow definition is valid. Returns `valid` (boolean), `issues` (list of actionable errors), and, when the draft is publishable, the `input_schema` and `output_schema` the Workflow exposes once live.
>
> **Example prompts**:
>
> * "Check whether the draft Workflow is publishable."
> * "Anything wrong with the Workflow before we publish it?"
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the draft Workflow to validate |
> | `agent_build_session_id` | Yes | - | Session ID from `start_agent_build_session` |
>
> **Notes:**
>
> * A draft Workflow can't be run, so this is the only check available before you publish.
> * The check covers the Workflow's structure only, so a draft that passes it can still be rejected when you publish. Publishing applies the stricter check.
> * The check also reports a required field that's set but unusable, such as an empty `integration_id` on a node that runs on a connected account, or a missing `data.category_id` on a node that reads Profound data. Publishing allows both, so a Workflow published with either one unresolved fails when it runs.
> * If the check reveals any issues, the assistant fixes them with `update_agent_definition`.

> **publish\_agent\_definition**
>
> Publishes a Workflow definition so it goes live and is visible across the organization. Displays the preview first and applies it after you confirm.
>
> **Example prompts**:
>
> * "Publish the Workflow so the team can use it."
> * "Make the article writer Workflow live."
>
> **Inputs**
>
> | Input | Required | Default | Description |
> | - | - | - | - |
> | `agent_id` | Yes | - | ID of the draft Workflow to publish |
> | `agent_build_session_id` | Yes | - | Session ID from `start_agent_build_session` |
> | `preview` | No | `true` | When `true`, return a plan without publishing; when `false`, publish |
>
> **Notes**:
>
> * Publishing checks the Workflow's structure and output wiring before anything changes, so a draft with structural problems is rejected and stays a draft. This check is stricter than the one `validate_agent_definition` runs.
> * Publishing doesn't check what the Workflow needs in order to run. A Workflow missing a node's `integration_id` or `data.category_id` publishes and then fails when it runs, so resolve both before you publish.

## Resources

Profound MCP provides reference material for Workflows as read-only MCP resources.

| Resource URI | Description |
| - | - |
| `file:///profound/agent-builder-guide` | The Workflow-building playbook, from intake and building through preview, validation, publishing, and running |
| `file:///profound/glossary/agent` | The definition of a Profound Workflow |
| `file:///profound/glossary/agent-run` | The definition of a Workflow run |


# Connection guides
Source: https://docs.tryprofound.com/mcp/common-mcp-clients

Connect your AI tool to Profound MCP server

> [!NOTE]
>
> If your tool is not listed here, check its MCP documentation for support for remote hosted MCP servers. See the [AI tools connection guide](/mcp/connection-tutorials/connect-ai-coding-tools) for the server URL and JSON configuration.

> **[Claude Code](/mcp/connection-tutorials/connect-ai-coding-tools#claude-code)**

> **[Claude Desktop](/mcp/connection-tutorials/connect-ai-coding-tools#claude-desktop)**

> **[ChatGPT](/mcp/connection-tutorials/connect-chatgpt)**

> **[Cursor](/mcp/connection-tutorials/connect-ai-coding-tools#cursor)**

> **[Gemini CLI](/mcp/connection-tutorials/connect-gemini-cli)**

> **[Microsoft Copilot Studio](/mcp/connection-tutorials/connect-ms-copilot)**

> **[OpenCode](/mcp/connection-tutorials/connect-ai-coding-tools#opencode)**

> **[VS Code](/mcp/connection-tutorials/connect-ai-coding-tools#vs-code)**

> **[Windsurf](/mcp/connection-tutorials/connect-ai-coding-tools#windsurf)**

> **[WRITER AI Studio](/mcp/connection-tutorials/connect-writer)**


# Connect AI coding tools to Profound MCP
Source: https://docs.tryprofound.com/mcp/connection-tutorials/connect-ai-coding-tools



This guide walks you through connecting your AI tool to Profound using our hosted MCP server.

```text Streamable HTTP theme={null}
https://mcp.tryprofound.com/mcp
```

## Claude Code

**Step 1**

Run this command in your terminal:

```bash theme={null}
claude mcp add --transport http profound https://mcp.tryprofound.com/mcp
```

**Step 2**

Use `/mcp` in Claude Code to confirm the server is installed and complete any authentication flow if prompted.

> **Using Different Installation Scopes**
>
> * `--scope local` installs the server only for the current project
> * `--scope project` shares the configuration with your team in the project
> * `--scope user` installs the server across your projects
>
> For more details, see [Claude Code MCP Installation Scopes](https://code.claude.com/docs/en/mcp#mcp-installation-scopes).

## Claude Desktop

**Step 1**

Open **Settings** -> **Connectors** in Claude Desktop.

**Step 2**

Search for **Profound** on the connectors list, select **Connect**.

**Step 3**

Complete any authentication flow if prompted.

## Cursor

**Step 1**

Open **Cursor Settings** -> **Tools & MCPs** -> **New MCP Server**.

**Step 2**

Add the following configuration:

```json theme={null}
{
  "mcpServers": {
    "profound": {
      "url": "https://mcp.tryprofound.com/mcp"
    }
  }
}
```

**Step 3**

Save changes to the config.

**Step 4**

Open **Cursor Settings** -> **Tools & MCPs** -> **New MCP Server**.

**Step 5**

Look for the `profound` MCP server entry, click the **Connect** button, and complete any authentication flow if prompted.

## OpenCode

Add Profound MCP to your OpenCode config:

```json theme={null}
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "profound": {
      "type": "remote",
      "url": "https://mcp.tryprofound.com/mcp",
      "enabled": true
    }
  }
}
```

If OpenCode prompts for authentication, run `opencode mcp auth profound`.

## VS Code

**Step 1**

Create a `.vscode/mcp.json` file in your workspace:

```json theme={null}
{
  "servers": {
    "profound": {
      "type": "http",
      "url": "https://mcp.tryprofound.com/mcp"
    }
  }
}
```

**Step 2**

Open the Command Palette and run **MCP: List Servers**.

**Step 3**

Start the Profound server and complete any authentication flow if prompted.

## Windsurf

**Step 1**

Open **Windsurf Settings** and search for **MCP**.

**Step 2**

Open the raw `mcp_config.json` file and add the following configuration:

```json theme={null}
{
  "mcpServers": {
    "profound": {
      "serverUrl": "https://mcp.tryprofound.com/mcp"
    }
  }
}
```

**Step 3**

Save the configuration and restart Windsurf.

## Other Tools

### JSON Configuration Format

Most MCP clients accept a JSON configuration.

```json Streamable HTTP theme={null}
{
  "mcpServers": {
    "profound": {
      "url": "https://mcp.tryprofound.com/mcp"
    }
  }
}
```

## Troubleshooting

### Authentication Issues

If your client prompts for authentication and the connection fails, try restarting the client and reconnecting. If your client offers a disconnect option for MCP servers, disconnect the existing connection and try again.

If you need a long-lived credential instead of the OAuth flow, see [Authentication](/mcp/authentication) for the Bearer token (API key) alternative.


# Connect ChatGPT to Profound MCP
Source: https://docs.tryprofound.com/mcp/connection-tutorials/connect-chatgpt



This guide walks you through connecting Profound to ChatGPT using the Profound MCP server. Once connected, ChatGPT can access Profound data and capabilities directly in conversation.

## Before you start

You'll need:

* A [ChatGPT](https://chatgpt.com/) account on a plan that supports developer mode. Learn more in the [OpenAI developer mode documentation](https://developers.openai.com/api/docs/guides/developer-mode)
* A [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Turn on developer mode**

In [ChatGPT](https://chatgpt.com/), open your account menu in the lower-left corner of the screen, and select **Settings**.

![ChatGPT landing page screenshot with the account menu expanded and Settings option highlighted](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-account-settings.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=15d832371fcd4a500beda7bea53d53d3)

Under **Security and login**, enable **Developer mode**.

![ChatGPT Settings screenshot with Security and login menu option and Developer mode toggle highlighted](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-settings-dev-mode.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=2e87eb0c9213d906185b75c44a9f48e8)

**2. Create the Profound plugin**

Go to the ChatGPT [Plugins page](https://chatgpt.com/plugins) and select **+**.

![ChatGPT Plugins page screenshot with an add plugin option highlighted](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-plugins-add.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=ada57bb06d16db73acc2200ae2bab1fc)

In the New Plugin dialog, fill in the fields as follows:

* **Name:** profound
* **Description:** Provides AEO and GEO marketing analytics and allows agents to operate the Profound platform.
* **MCP server URL:** `https://mcp.tryprofound.com/mcp`
* **Authentication:** OAuth

  ![ChatGPT New Plugin dialog screenshot with Profound server details filled in](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-new-plugin.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=2dd338e6fbd94044b5b194b903fce32a)

Under the custom MCP servers risk notice, select **I understand and want to continue**, then select **Create**.

**3. Authenticate with Profound**

Profound uses OAuth 2.0. Complete the flow with your Profound account.

![ChatGPT Add profound to ChatGPT authorization modal screenshot with Sign in with profound button](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-oauth.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=c5536c8191f821d96b33c196f687b199)

If you have access to multiple Profound organizations, select the one you want to use with the Profound MCP server.

![Profound OAuth dialog screenshot with Organization access dropdown and Allow access button highlighted](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-connect-account.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=e25bd610914f5471d698658e8db151d6)

**4. Install the plugin**

After you configure the Profound plugin, it appears among the available plugins in **Plugins** > **Personal** in ChatGPT.

Select the plugin, then select **Install** on the plugin page.

![Profound plugin page screenshot with Install button highlighted](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-install-plugin.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=7c67768045dc92fac587a06453198cb7)

**5. Verify the connection**

* Open **Settings** > **Plugins**.
* Confirm that **profound** appears in your list of installed plugins.

  ![ChatGPT Settings Plugins screenshot with profound plugin highlighted](https://mintcdn.com/profound-37face47/dP9Ca1npe_by_Wf3/images/mcp/chatgpt/chatgpt-success.png?fit=max\&auto=format\&n=dP9Ca1npe_by_Wf3\&q=85\&s=9826868e4752b3282456b9fd75bb83db)

## Troubleshooting

If the plugin fails to create or ChatGPT can't reach Profound:

* Remove the plugin and add it again as instructed in [Step 2](#step2).
* If authentication fails, sign out of Profound in your browser, sign back in, and retry.


# Connect Gemini CLI to Profound MCP
Source: https://docs.tryprofound.com/mcp/connection-tutorials/connect-gemini-cli



This guide shows you how to connect Gemini CLI to the Profound MCP server, so Gemini can pull your Profound visibility data and run Profound actions directly from the terminal.

## Before you start

Make sure you have:

* [Gemini CLI](https://geminicli.com/) installed
* A [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Authenticate with Gemini CLI**

1. Run Gemini CLI in your terminal:
   ```bash theme={null}
   gemini
   ```

2. When asked **How would you like to authenticate for this project?**, select **Sign in with Google**. Gemini will launch a browser session.

3. In the browser, select your Google account and sign in.

If successful, you'll see tips to get started and the command line.

![Gemini CLI startup view screenshot](https://mintcdn.com/profound-37face47/1cWSg0aIvHViHQFW/images/mcp/gemini-cli-startup-view.png?fit=max\&auto=format\&n=1cWSg0aIvHViHQFW\&q=85\&s=76d40e31a87646b8b53c7540d7bcd79c)

**2. Add the Profound MCP server**

1. MCP servers are configured in the `~/.gemini/settings.json` configuration file. To add the Profound server, create the `mcpServers` block with the Profound server details in `settings.json`:
   ```json theme={null}
     "mcpServers": {
       "profound": {
         "url": "https://mcp.tryprofound.com/mcp"
       }
     }
   ```
   You can do this in your preferred code editor, or directly from Gemini CLI by entering the prompt:
   ```text theme={null}
   Add an MCP server called profound with url https://mcp.tryprofound.com/mcp
   ```

2. Review the changes and apply them if the `mcpServers` block looks correct.

   ![Gemini CLI screenshot with the prompt to add a new MCP server, a change in settings.json and Apply this change - Allow once option highlighted](https://mintcdn.com/profound-37face47/1cWSg0aIvHViHQFW/images/mcp/gemini-cli-add-server-prompt.png?fit=max\&auto=format\&n=1cWSg0aIvHViHQFW\&q=85\&s=3fb9455f76c783cecbbfef87b017f109)

3. For the changes to take effect, quit and reopen your Gemini CLI.

4. To verify the server was added, run the command to list the MCP servers inside the CLI.

   ```bash theme={null}
     /mcp list
   ```

   It should show `profound` among the configured MCP servers.

   ![Gemini CLI screenshot with profound server on the list of configured servers](https://mintcdn.com/profound-37face47/1cWSg0aIvHViHQFW/images/mcp/gemini-cli-server-added.png?fit=max\&auto=format\&n=1cWSg0aIvHViHQFW\&q=85\&s=c3e025c8441f3ebb942e06ba1bf11905)

**3. Sign in to Profound**

Profound authenticates with OAuth. Gemini CLI discovers the OAuth endpoints automatically, so no API keys, client ID, or secret are needed.

1. Inside the CLI, run the command to authenticate with Profound:

   ```bash theme={null}
   /mcp auth profound
   ```

2. Select **Yes** when asked to authenticate in the browser. This will open the browser to the Profound sign-in page.

   ![Gemini CLI screenshot with authentication command and an authentication dialog with the Yes option highlighted](https://mintcdn.com/profound-37face47/1cWSg0aIvHViHQFW/images/mcp/gemini-cli-auth-profound.png?fit=max\&auto=format\&n=1cWSg0aIvHViHQFW\&q=85\&s=ed557cad3e990f807e7046cc7bc02b64)

3. If you're not logged in, follow the prompts to log in to your Profound account. Then approve access.

   ![Gemini CLI oauth dialog screenshot with Allow access option highlighted](https://mintcdn.com/profound-37face47/1cWSg0aIvHViHQFW/images/mcp/gemini-cli-oauth.png?fit=max\&auto=format\&n=1cWSg0aIvHViHQFW\&q=85\&s=9a1874de714a746cb59a8224d20439b8)

When you see the success message, return to the Gemini CLI session in your terminal.

**4. Verify the connection**

Still inside Gemini CLI, run the MCP list command again.

```bash theme={null}
  /mcp list
```

If the Profound server shows as **Ready** with a green circle and has available tools listed, it's connected and ready to go.

![Gemini CLI screenshot with an mcp list command ran and profound server listed as ready ](https://mintcdn.com/profound-37face47/1cWSg0aIvHViHQFW/images/mcp/gemini-cli-profound-ready.png?fit=max\&auto=format\&n=1cWSg0aIvHViHQFW\&q=85\&s=5e602e680f2ba3321a39c9b8b9e1923f)

> [!TIP]
>
> For more details about configuring MCP server connections in Gemini CLI, see Gemini CLI's [MCP server documentation](https://geminicli.com/docs/tools/mcp-server).

## Troubleshooting

### The browser doesn't open during sign-in

Make sure you have a local browser installed. OAuth won't work in headless or remote sessions without browser access. Run the setup where you can open a browser, then retry signing in or authenticating with Profound.

### Profound server shows as disconnected

Re-run `/mcp auth profound` to refresh the token. If it still fails, sign out of Profound in your browser, sign back in, and retry.


# Connect Microsoft Copilot Studio agent to Profound MCP
Source: https://docs.tryprofound.com/mcp/connection-tutorials/connect-ms-copilot



This guide walks you through connecting Profound to a Microsoft Copilot Studio agent using the Profound MCP server. Once connected, your agent can access Profound data and capabilities directly.

## Before you start

Before you get started, you'll need:

* [Access to a Microsoft Copilot Studio account](https://learn.microsoft.com/en-us/microsoft-copilot-studio/requirements-licensing-subscriptions) with an [agent set up](https://learn.microsoft.com/en-us/microsoft-copilot-studio/fundamentals-get-started)
* A [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Open your agent in Copilot Studio**

* Sign in to [Microsoft Copilot Studio](https://copilotstudio.microsoft.com/)
* Go to the **Agents** tab in the left navigation bar

  ![Microsoft Copilot Studio landing page screenshot with Agents tab highlighted](https://mintcdn.com/profound-37face47/HEiJHUZHHXuTyMES/images/mcp/ms-copilot-landing-page.png?fit=max\&auto=format\&n=HEiJHUZHHXuTyMES\&q=85\&s=ef868da28d5cdb10436cfad79b424c0f)
* Select and open the agent you want to connect Profound to
* Go to the **Tools** tab for the agent

  ![Microsoft Copilot Studio agent page screenshot with Tools tab highlighted](https://mintcdn.com/profound-37face47/HEiJHUZHHXuTyMES/images/mcp/ms-copilot-agent.png?fit=max\&auto=format\&n=HEiJHUZHHXuTyMES\&q=85\&s=c3d447c939ddb934bdbcb5eb5bc94236)

**2. Add a new tool**

* Select **+ New tool**

  ![Microsoft Copilot Studio tools page screenshot with Tools tab and + New Tool button highlighted](https://mintcdn.com/profound-37face47/HEiJHUZHHXuTyMES/images/mcp/ms-copilot-tools.png?fit=max\&auto=format\&n=HEiJHUZHHXuTyMES\&q=85\&s=f71de8ef668600bbde9ec4d71ffbc16a)
* In the New tool dialog window, select **Model Context Protocol**

  ![Microsoft Copilot Studio New Tool dialog screenshot with Model Context Protocol option highlighted](https://mintcdn.com/profound-37face47/HEiJHUZHHXuTyMES/images/mcp/ms-copilot-new-tool.png?fit=max\&auto=format\&n=HEiJHUZHHXuTyMES\&q=85\&s=f0998ed1c48abacb5fb348bdfa6b5a41)

**3. Enter the Profound server details**

After the setup wizard appears, fill in the fields as follows:

![Microsoft Copilot Studio Model Context Protocol setup wizard screenshot](https://mintcdn.com/profound-37face47/Dq3YtFbUVP-dqF6H/images/mcp/ms-copilot-setup-wizard.png?fit=max\&auto=format\&n=Dq3YtFbUVP-dqF6H\&q=85\&s=99c713d660d441f9a5b19413c608e6fa)

* **Server name:** Profound
* **Server description:** Provides AEO and GEO marketing analytics as well as allows agents to operate the Profound platform.
* **Server URL:** `https://mcp.tryprofound.com/mcp`
* **Authentication:** OAuth 2.0

**4. Set up authentication**

> [!NOTE]
>
> Profound uses OAuth 2.0 to verify your identity. This is the standard "sign in with your account" method used by most modern apps.

After you select OAuth 2.0 as the authentication method in the setup wizard form:

* Select **Dynamic discovery**: This is the simplest option that lets Copilot Studio configure the sign-in flow automatically

  ![Microsoft Copilot Studio Model Context Protocol setup wizard Authentication section screenshot with Dynamic discovery option selected](https://mintcdn.com/profound-37face47/HEiJHUZHHXuTyMES/images/mcp/ms-copilot-dynamic-discovery.png?fit=max\&auto=format\&n=HEiJHUZHHXuTyMES\&q=85\&s=4075536a3bc52c845485c632a866113e)
* Select **Create**

After that, the **Add tool** dialog appears.

**5. Create a connection and add the server**

* On the Add tool screen, select **Create a new connection**
* Follow any sign-in prompts to authenticate with your Profound account
* After the connection is created, select **Save** for the MCP server, then refresh the tools list

  ![Microsoft Copilot Studio Tools section screenshot with refresh tools button and Allow all toggle highlighted](https://mintcdn.com/profound-37face47/Dq3YtFbUVP-dqF6H/images/mcp/ms-copilot-refresh-tools.png?fit=max\&auto=format\&n=Dq3YtFbUVP-dqF6H\&q=85\&s=c299819cb903ab412a189a68fa97a1c7)
* Select **Add to agent** to finish

Your agent now has access to the Profound MCP server.

**6. Verify the connection**

To confirm everything is working:

* Go to the **Tools** tab for your agent
* Look for **Profound** in the list of connected tools

  ![Microsoft Copilot Studio Model Tools tab screenshot with the newly connected Profound tool highlighted](https://mintcdn.com/profound-37face47/HEiJHUZHHXuTyMES/images/mcp/ms-copilot-setup-success.png?fit=max\&auto=format\&n=HEiJHUZHHXuTyMES\&q=85\&s=76b7a209d48ba2c83cae204a6b4f6662)

If it shows as connected, you're ready to go.

If the connection doesn't appear or the agent can't reach Profound, try removing the tool and adding it again from step 2. If authentication fails during setup, sign out of Profound in your browser, sign back in, and retry.

For more details about configuring MCP servers in Copilot Studio, see Microsoft's [Add an existing MCP server to an agent](https://learn.microsoft.com/en-us/microsoft-copilot-studio/mcp-add-existing-server-to-agent) guide.


# Connect WRITER AI Studio to Profound MCP
Source: https://docs.tryprofound.com/mcp/connection-tutorials/connect-writer



Connect Profound MCP to WRITER's AI Studio so your WRITER agents can pull live data directly from Profound.

## Before you start

You'll need:

* Access to [WRITER AI Studio](http://app.writer.com/aistudio)
* A [Profound](https://www.tryprofound.com/) account

## Setup guide

**1. Open Connectors & Tools**

In WRITER AI Studio, go to **Connectors & Tools > Connectors** in the left sidebar and select **+ Custom connector**.

![WRITER AI Studio landing page screenshot with Connectors & Tools tab and Custom connector button highlighted](https://mintcdn.com/profound-37face47/Af4z9l8U0tbID14c/images/mcp/writer-landing-page.png?fit=max\&auto=format\&n=Af4z9l8U0tbID14c\&q=85\&s=2c858047a0279625b4c857530c744950)

**2. Choose the connector type**

When prompted to choose a connector type, select **Connect to an MCP server** and enter the Profound server URL: `https://mcp.tryprofound.com/mcp`.

![WRITER AI Studio connector type selection screenshot](https://mintcdn.com/profound-37face47/Af4z9l8U0tbID14c/images/mcp/writer-create-connector.png?fit=max\&auto=format\&n=Af4z9l8U0tbID14c\&q=85\&s=e769023cb5150dbd393512386732cd07)

**3. Enter the Profound connector details**

In the **Edit connector details** screen, fill in the fields as follows:

* **Connector name:** Profound
* **About this connector:** Provides AEO and GEO marketing analytics as well as allows agents to operate the Profound platform

  ![WRITER AI Studio Edit connector details dialog screenshot with relevant fields highlighted](https://mintcdn.com/profound-37face47/Af4z9l8U0tbID14c/images/mcp/writer-edit-connector-details.png?fit=max\&auto=format\&n=Af4z9l8U0tbID14c\&q=85\&s=2aa93f40b9f33f470215055f2e693251)

Select **Next**.

**4. Configure the authentication**

On the **Connect to Profound** screen, scroll to **Define Authentication** and fill in the fields:

* **Authentication type:** OAuth
* **Authorization URL:** `https://auth.tryprofound.com/oauth2/authorize`
* **Access token URL:** `https://auth.tryprofound.com/oauth2/token`
* **DCR:** Yes

  ![Define Authentication section of the WRITER AI Studio Connect to Profound screen with relevant fields highlighted](https://mintcdn.com/profound-37face47/Af4z9l8U0tbID14c/images/mcp/writer-define-authentication.png?fit=max\&auto=format\&n=Af4z9l8U0tbID14c\&q=85\&s=c2c9fa4bd304ede282f74b591e703b01)

Leave the rest of the fields empty or at their default values, and select **Next** to complete the connection.

**5. Verify the connection**

To confirm everything is working:

1. Go to **Connectors & Tools > Connectors** in the left sidebar in the AI Studio UI.
2. Select **+ New connector**.
3. Look for the Profound connector in the list.

![WRITER AI Studio available connectors list screenshot with Profound connector available](https://mintcdn.com/profound-37face47/Af4z9l8U0tbID14c/images/mcp/writer-connection-success.png?fit=max\&auto=format\&n=Af4z9l8U0tbID14c\&q=85\&s=19aad3f719c07fa3e8408c15ea4a3499)

Your Profound MCP connector is now available to use.

> [!TIP]
>
> For more details about configuring custom MCP connectors in AI Studio, see WRITER's [Custom connectors guide](https://dev.writer.com/home/custom-connectors).

## Troubleshooting

### Rate limit exceeded

If you see a `rate limit reached` error, it's coming from WRITER. Wait a few minutes and try creating a connector again.

### Connection failed

If the connection fails after the rate limit error, the connector may have still been created. Check your available connectors and if it's not there, try creating a connector again.


# Profound MCP overview
Source: https://docs.tryprofound.com/mcp/overview

Learn how to connect AI agents to Profound

> [!NOTE]
>
> **Available on**: the Enterprise plan

Connect AI tools and agents to Profound using the [Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) open source standard for connecting AI applications to external systems.

## What is Profound MCP?

Profound MCP is our hosted server for integrating with AI tools that support MCP. Instead of wiring up a custom integration from scratch, you connect an MCP client to Profound and let the client call tools to execute API operations with your Profound account.

```text Hosted server URL theme={null}
https://mcp.tryprofound.com/mcp
```

### Why use Profound MCP?

Profound MCP lets you connect to Profound quickly, without building a custom integration. Your Profound data and [Workflows](/agents/workflows/about-workflows) become available directly inside the AI tool you already use. See how to connect common MCP clients to Profound in [Connection guides](/mcp/common-mcp-clients).

### What can you do with Profound MCP?

* **Visibility Reports** — Retrieve visibility metrics and performance data for companies within specific categories
* **Sentiment Analysis** — Analyze sentiment data and emotional responses across companies and topics
* **Citation Reports** — Track mentions, references, and citation patterns across domains and pages
* **Shopping Reports** — Measure how brands, products, and retailers appear in ChatGPT shopping results
* **Raw Data Access** — Access unprocessed data for custom analysis and reporting
* **Agent Analytics** — Access bot reports, referral reports, and raw analytics data
* **Build and Run Profound Workflows** — Start existing Workflows, create and publish new ones, and poll runs for status and outputs
* **Prompt Management** — Create prompts in a category, update existing ones, and activate, disable, or delete them in bulk
* **Knowledge Bases** — List your Knowledge Bases, search their contents, and add new documents to them
* **Profound Docs** — Find, read, write, and manage collaborative documents such as strategy memos, briefs, and retros
* **Projects** — Read, generate, and manage Projects and their tasks toward a measurable goal

For the full tool and resource catalog, see the [Capabilities](/mcp/capabilities) section.


# Automotive
Source: https://docs.tryprofound.com/resources/prompt-guides/automotive

How to build a prompt set for an automotive brand, with best practices and examples

Car shoppers compare models, trims, and prices before they buy, so the set needs prompts for each model and trim, plus financing, trade-in value, and total cost of ownership. If you sell electric vehicles (EVs), add prompts on range and charging.

## Create your prompt set

### Topics to cover

Replace `[model]` in the example prompts with one of your models.

| Topic | Example prompt |
| - | - |
| Models and trims | Which trim of the `[model]` is worth the extra money? |
| Financing and leasing | Should I lease or buy a midsize SUV with 60-month rates? |
| Head-to-head comparisons | `[model]` vs `[competitor model]` for a family of five |
| Trade-ins and resale value | Which SUVs hold their value best after five years? |
| Model-year launches | What changed in the 2027 `[model]`? |
| Parts and fitment | Will these brake pads fit a 2019 `[model]`? |
| EV and charging | How far can the `[model]` go in winter, and where can I charge it? |
| Service and recalls | Is there a recall on the 2022 `[model]`? |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best midsize SUVs?"*, try *"We have two kids in car seats and a 40-minute commute. What are the best midsize SUVs with a third row and good fuel economy under \$45,000?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Build one set of prompts for each model and trim.
2. Add financing, lease, and total cost of ownership prompts.
3. Use [FactCheck](/answer-engine-insights/factcheck/about) for specs, pricing, and recall information.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which models and trims have the thinnest prompt coverage? | Where to build a set next |
| Which competitors show up instead of us on `[segment]` comparison prompts? | Where [comparison prompts](/getting-started/glossary#competitor-prompt) are lost |
| What are car shoppers asking AI about EVs and charging that my set misses? | New questions to track |
| Which specs, prices, or recall claims about our models are wrong in AI answers? | FactCheck prompts to add |
| What should I add before the model-year changeover and year-end sales? | Seasonal additions |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Model | Sedan, SUV, Truck, EV |
| Buyer need | Family, Commuter, Towing, First car |
| Powertrain | Gas, Hybrid, Electric |
| Campaign | Model-year launch, Financing offer, Certified pre-owned |
| Season | Model-year changeover, Year-end sales, Tax refund season |
| FactCheck | Specs, Pricing, Recalls |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# B2B software
Source: https://docs.tryprofound.com/resources/prompt-guides/b2b-software

How to build a prompt set for a B2B software brand, with best practices and examples

People choosing business software narrow their options by budget and by the tools their company already uses, so the prompt set needs pricing and integration prompts for every product. If IT reviews software purchases, add security and compliance prompts. If a larger competitor leads the category, add prompts that ask for alternatives to it.

## Create your prompt set

### Topics to cover

Replace `[brand]` and `[competitor]` in the example prompts with your brand and a competitor.

| Topic | Example prompt |
| - | - |
| Each product or module | Best reporting tools for marketing teams that pull from Salesforce |
| New integrations | Does `[brand]` integrate with Salesforce and Slack? |
| Marketing ops | Which tool keeps our campaign calendar from slipping? |
| Alternatives to a competitor | Best alternatives to `[competitor]` for a 50-person team |
| IT and security | Which project tools meet SOC 2 and support SSO? |
| Budget planning season | Which tools are worth adding to next year's budget? |
| Finance and procurement | Which project tool has the best pricing for 50 users? |
| Conferences and user events | What did `[brand]` announce at its user conference? |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best project management tools for marketing teams?"*, try *"I run marketing ops at a 50-person SaaS company and our campaigns keep slipping. What are the best project management tools for marketing teams that integrate with Salesforce for under \$15 a user?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Add prompts that check integrations, pricing, and security or compliance.
2. Add "alternatives to `[competitor]`" prompts for the competitors buyers consider instead of you.
3. Run each topic again by company size and by role.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which competitors show up instead of us on integration and pricing prompts? | Where [validation prompts](/getting-started/glossary#validation-prompt) are lost |
| Which product modules and personas have the thinnest prompt coverage? | Business lines to add depth to |
| What are buyers asking AI about alternatives to `[competitor]`, and which sources get cited? | Alternatives prompts to add, and sources to pitch |
| Does our visibility change by company size or role? Compare my personas. | Which variations matter most |
| What do AI answers say about our security and compliance, and is any of it wrong? | FactCheck prompts to add |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Product or module | Project management, Reporting, Automation, Integrations |
| Persona | Marketing ops, IT and security, Finance, Executive |
| Company size | SMB, Mid-market, Enterprise |
| Campaign | New integration, Alternatives to `[competitor]`, Product launch |
| Season | Budget planning, End of quarter, Annual conference |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Consumer packaged goods
Source: https://docs.tryprofound.com/resources/prompt-guides/consumer-packaged-goods

How to build a prompt set for a consumer packaged goods brand, with best practices and examples

Shoppers ask AI to plan meals, parties, and shopping trips, so the set needs prompts about occasions as well as products. Build prompts for each brand line, then add prompts that check ingredients, product claims, and where the products are sold.

## Create your prompt set

### Topics to cover

Replace `[brand]` in the example prompts with one of your brands.

| Topic | Example prompt |
| - | - |
| Each brand line | What are the best granola bars for a busy morning? |
| New product launches | Is the new `[brand]` sparkling water any good? |
| Usage occasions | Easy weeknight dinners my kids will actually eat |
| Store brand vs name brand | Is `[brand]` cereal worth it over the store brand? |
| Ingredients and dietary needs | Best nut-free snacks for school lunches |
| Sustainability and packaging | Which laundry detergents come in recyclable packaging? |
| Where to buy | Where can I buy `[brand]` oat milk near me? |
| Seasonal moments | Best snacks for a summer cookout or a holiday party |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best snacks for kids' lunchboxes?"*, try *"I pack lunches for two kids, one with a nut allergy. What are the best nut-free snacks for kids' lunchboxes under \$5 a box that I can find at Target?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Tag every prompt with a brand line and an occasion.
2. Add ingredient, allergen, and dietary prompts for each product.
3. Put claims like nut-free, non-GMO, and recyclable under [FactCheck](/answer-engine-insights/factcheck/about).

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which brand lines and occasions have the fewest prompts in my set? | Where coverage is thin |
| What are shoppers asking AI about `[product category]` that none of my prompts cover? | New questions to track |
| Which ingredient, allergen, or sustainability claims about us are wrong in AI answers? | FactCheck prompts to add |
| Which retailers and review sites get cited for `[category]`, and are we on them? | Which sources shape visibility in the category |
| Draft 10 back-to-school lunch prompts, tagged by brand line and occasion, and skip anything I already track. | A tagged starter set |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Brand line | Snacks, Beverages, Household, Personal care |
| Occasion | Weeknight dinner, Lunchbox, Party, On the go |
| Dietary need | Nut-free, Gluten-free, Vegan, Low sugar |
| Season | Back to school, Summer grilling, Holiday baking |
| FactCheck | Ingredients, Allergens, Packaging |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Financial services
Source: https://docs.tryprofound.com/resources/prompt-guides/financial-services

How to build a prompt set for a financial services brand, with best practices and examples

People choosing financial products compare rates, fees, and eligibility. Those details change often, so the prompt set needs prompts for each product and each customer segment, with the numbers checked by [FactCheck](/answer-engine-insights/factcheck/about). Add promotion prompts before tax season and the spring home-buying period.

## Create your prompt set

### Topics to cover

| Topic | Example prompt |
| - | - |
| Cards | What are the best travel credit cards for beginners? |
| Students and first-time buyers | Best first credit card for a college student |
| Savings and checking | Best high-yield savings accounts with no minimum |
| Small business | Best business checking for a new LLC |
| Loans and mortgages | How much house can I afford on a \$90K salary? |
| Sign-up bonuses and rate promos | Which travel card has the best sign-up bonus right now? |
| Investing and insurance | Best way to start investing with \$500 |
| Tax season and year-end | Best place to put a tax refund |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best travel credit cards for beginners?"*, try *"I'm a recent grad who travels for work twice a month and I want my flights to earn points. What are the best travel credit cards for beginners with no annual fee or foreign transaction fees?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Add prompts that check fees, rates, and who qualifies.
2. Put rates, fees, and terms under [FactCheck](/answer-engine-insights/factcheck/about), since they change often.
3. Add versions for students, first-time buyers, and small businesses.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which rate, fee, and terms claims about us are inaccurate in AI answers, and where do they come from? | What to fix first |
| Compare my visibility on card prompts for students, first-time buyers, and small business. | Which segments need more prompts |
| What are people asking AI about `[product]` that none of my prompts cover? | New questions to track |
| Which prompts should I add before tax season and spring home-buying? | Seasonal additions |
| Which competitors win on sign-up bonus prompts, and what sources do they cite? | Promo prompts worth tracking |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Product line | Cards, Savings, Loans, Mortgages, Investing, Insurance |
| Segment | Students, First-time buyers, Small business |
| Campaign | Sign-up bonus, Rate promotion, New product |
| Season | Tax season, Year-end investing, Spring home-buying, Holiday travel |
| FactCheck | Fees, Rates, Eligibility, Terms |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Healthcare
Source: https://docs.tryprofound.com/resources/prompt-guides/healthcare

How to build a prompt set for a healthcare brand, with best practices and examples

Patients choose care based on their insurance, their schedule, and their location, so the prompt set needs to cover each service line with those practical details built into the questions. Have your compliance team review the prompts before you publish them.

## Create your prompt set

### Topics to cover

Replace `[brand]` and `[city]` in the example prompts with your own values.

| Topic | Example prompt |
| - | - |
| Recurring conditions | When should I see a doctor for recurring migraines? |
| New clinics and awareness months | Is there a new urgent care near `[city]` that opened this year? |
| Telehealth visits | Best telehealth for migraines that takes Aetna |
| Open enrollment | How do I pick a plan that covers my specialist? |
| Insurance questions | Does `[brand]` accept my insurance for virtual visits? |
| Flu and allergy season | Where can I get a flu shot today without an appointment? |
| Same-day and after-hours care | Which telehealth provider can see me today? |
| Back-to-school and New Year resets | Where can my kid get a sports physical this week? |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best telehealth services for migraine treatment?"*, try *"I'm a teacher with chronic migraines and I can't take time off for appointments. What are the best telehealth services for migraine treatment that take Aetna and offer evening visits?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Add insurance, availability, and location to the prompts.
2. Use [FactCheck](/answer-engine-insights/factcheck/about) to confirm accepted insurance, hours, and services.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which insurance, hours, or services claims about us are wrong or unverified in AI answers? | Facts to put under FactCheck |
| What are patients asking AI about `[specialty]` that my prompts don't cover? | Unanswered patient questions |
| Show my visibility by service line. Which ones are thin? | Where to add prompts first |
| Which prompts should I add before flu season and open enrollment? | Seasonal additions |
| Which sources do AI answers cite for telehealth in our area, and are we on them? | Which sources shape visibility in your area |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Service line | Neurology, Primary care, Behavioral health, Telehealth |
| Constraint | Insurance, Same-day, Evening hours, Location |
| Campaign | New clinic, New service, Awareness month |
| Season | Open enrollment, Flu season, Allergy season, Back-to-school physicals, New Year resolutions |
| Governance | FactCheck: insurance, FactCheck: hours, Compliance reviewed |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Pharma and life sciences
Source: https://docs.tryprofound.com/resources/prompt-guides/pharma-life-sciences

How to build a prompt set for a pharma or life sciences brand, with best practices and examples

Patients, caregivers, and prescribers ask different questions about the same condition, so the prompt set needs to speak to each audience separately and keep branded and unbranded prompts apart. Ask the medical, legal, and regulatory teams to review the set before you track it.

## Create your prompt set

### Topics to cover

Replace `[brand]`, `[condition]`, and `[treatment class]` in the example prompts with your own values.

| Topic | Example prompt |
| - | - |
| Condition education | What causes plaque psoriasis, and when should I see a specialist? |
| Cost and coverage | Does insurance cover `[brand]`, and is there a copay program? |
| Treatment options | What treatments exist for moderate plaque psoriasis beyond creams? |
| Patient support programs | What help is there for starting a new injectable treatment? |
| Brand questions | What is `[brand]` used for, and who shouldn't take it? |
| Prescriber questions | What does the latest data say about `[treatment class]` for adults? |
| Safety and side effects | What are common side effects of biologics for psoriasis? |
| Awareness months and launches | What is new in treatment for `[condition]` this year? |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the treatment options for moderate plaque psoriasis?"*, try *"I was diagnosed with plaque psoriasis last year and creams aren't working. What treatment options are there for moderate plaque psoriasis beyond topical creams, and how do they differ?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Write separate prompts for patients, caregivers, and prescribers.
2. Put indications, safety, and dosing under [FactCheck](/answer-engine-insights/factcheck/about).

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which indication, safety, or dosing claims about our brand are wrong or out of date in AI answers? | Facts to put under FactCheck |
| What are patients asking AI about `[condition]` that my prompts don't cover? | Unanswered patient questions |
| Compare my visibility on branded and unbranded prompts for `[condition]`. | Where the gap is |
| Which sources do AI answers cite for `[condition]` treatments, and are we on them? | Which sources shape visibility for the condition |
| What should I add around `[awareness month]` or a new data release? | Moments worth a prompt set |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Audience | Patient, Caregiver, Prescriber |
| Condition | Psoriasis, Migraine, Diabetes |
| Brand type | Unbranded, Branded, Competitor |
| Campaign | Awareness month, Launch, Patient support |
| FactCheck | Indication, Safety, Dosing, Coverage |
| Governance | Medical review, Legal review |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Publishing and media
Source: https://docs.tryprofound.com/resources/prompt-guides/publishing-media

How to build a prompt set for a publishing or media brand, with best practices and examples

Readers choose a publication by the topics it covers, its price, and how often it publishes, so the prompt set needs prompts for each beat plus prompts that check price and discount claims.

## Create your prompt set

### Topics to cover

Replace `[brand]` in the example prompts with your own value.

| Topic | Example prompt |
| - | - |
| Each beat or vertical | How can I keep up with climate policy news? |
| Promotions and discounts | Is there a holiday deal on a `[brand]` subscription? |
| Newsletters | What are the best newsletters for climate policy coverage? |
| Special reports | Where can I find a deep report on grid reliability? |
| Podcasts | Which podcasts explain energy policy without the jargon? |
| Elections and major events | Where should I follow election results as they come in? |
| Subscription tiers | Does `[brand]` offer a student discount on subscriptions? |
| Awards and earnings season | Best coverage of tech earnings week |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best newsletters for climate policy coverage?"*, try *"I'm a policy analyst at a clean energy nonprofit and I need new regulations in one place. What are the best newsletters for climate policy coverage that publish daily for under \$10 a month?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Build one set of prompts for each beat.
2. Add price, frequency, and discount prompts.
3. Add news-cycle prompts, and retire them when the cycle ends.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which beats have the thinnest prompt coverage in my set? | Where to build a set next |
| What are readers asking AI about `[beat]` that my prompts don't cover? | New questions to track |
| Which price, frequency, or discount claims about us are wrong in AI answers? | FactCheck prompts to add |
| Which sources get cited for `[beat]` recommendations, and are we on them? | Which sources shape visibility for the beat |
| Which of my news-cycle prompts are stale and ready to retire? | Cleanup before the next cycle |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Beat | Climate policy, Tech, Business, Health |
| Format | Newsletter, Podcast, Website, Magazine |
| Campaign | Subscription promo, New product, Special report |
| Season | Elections, Awards, Earnings, Holiday deals |
| Validation | Price, Frequency, Student discount |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Retail and e-commerce
Source: https://docs.tryprofound.com/resources/prompt-guides/retail

How to build a prompt set for a retail or e-commerce brand, with best practices and examples

Shoppers buy around sales events, holidays, and school calendars, so the prompt set needs prompts for each product category and for each peak season, including gift guide and "best X under \$Y" questions. Put shipping, returns, and ingredient claims under [FactCheck](/answer-engine-insights/factcheck/about) so you can see where AI answers get them wrong.

## Create your prompt set

### Topics to cover

Replace `[brand]` and `[product]` in the example prompts with your own values.

| Topic | Example prompt |
| - | - |
| Black Friday and Cyber Monday | Best deals on `[product]` this weekend under \$100 |
| Prime Day-style sales events | What's actually on sale at `[brand]` this week? |
| Holiday gifting | Gift ideas for someone who loves skincare, under \$50 |
| Product launches | How does the new `[brand]` serum compare to the old one? |
| Valentine's Day | Last-minute Valentine's gifts that ship in two days |
| Limited-stock drops | Where can I still buy the `[brand]` holiday set? |
| Back to school | Best backpacks for a college freshman on a budget |
| Private label vs third-party | Is `[brand]`'s own moisturizer as good as the name brands? |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best moisturizers for sensitive skin?"*, try *"I'm in my 30s with rosacea and my moisturizer stings. What are the best moisturizers for sensitive skin under \$40 and fragrance-free?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Tag every prompt with a category and a funnel stage.
2. Add "best X under \$Y" and gift guide prompts before each peak season.
3. Add [FactCheck](/answer-engine-insights/factcheck/about) prompts for shipping, returns, and ingredient claims.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which of my prompts cover gifting and sales events, and which peak moments have nothing? | Seasonal gaps before the rush |
| Show my visibility by topic and funnel stage. Where am I weakest? | Which categories need more prompts |
| What are shoppers asking AI about `[category]` that none of my prompts cover? | New questions to track |
| Which shipping, returns, or ingredient claims about us are wrong in AI answers, and who is the source? | FactCheck prompts worth adding |
| Draft 10 Black Friday prompts, tagged by category and funnel stage, and skip anything I already track. | A tagged starter set |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Category | Skincare, Fragrance, Private label, Third-party brands |
| Campaign | Product launch, Limited-stock drop, Sales event |
| Season | BFCM, Holiday gifting, Valentine's Day, Back to school, Summer sale |
| FactCheck | Shipping, Returns, Ingredients |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Telecom and communications
Source: https://docs.tryprofound.com/resources/prompt-guides/telecom

How to build a prompt set for a telecom or communications brand, with best practices and examples

Phone and internet shoppers compare plans on price and on coverage where they live, so the prompt set needs prompts for each plan and device, plus coverage prompts for each city you serve. Add prompts about switching carriers, contracts, and fees, since those questions come up before a purchase.

## Create your prompt set

### Topics to cover

Replace `[city]` and `[city or zip code]` in the example prompts with the markets you serve.

| Topic | Example prompt |
| - | - |
| Mobile plans | What is the best cell phone plan for a family of four? |
| Coverage by location | Which carrier has the best signal in `[city or zip code]`? |
| Home internet | Which home internet is fastest in `[city]`? |
| Switching carriers | What happens to my phone payments if I switch? |
| Bundles | Does bundling internet and mobile actually save money? |
| Business lines | Best phone plans for a 10-person small business |
| Device launches | Is it worth upgrading to the new phone this fall? |
| Promotions and trade-in deals | What is the best trade-in deal right now? |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best cell phone plans for families?"*, try *"We're a family of four in Columbus and we're paying too much. What are the best cell phone plans for a family of four with unlimited data and good coverage under \$140 a month?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Add city and zip code coverage prompts for your key markets.
2. Add contract, fee, and switching prompts.
3. Use [FactCheck](/answer-engine-insights/factcheck/about) for prices, speeds, and coverage claims.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Which cities or regions have the weakest visibility on coverage prompts? | Where to add local prompts |
| Which competitors show up instead of us on price comparison prompts? | Where [comparison prompts](/getting-started/glossary#competitor-prompt) are lost |
| What are people asking AI about switching carriers that my prompts don't cover? | New questions to track |
| Which price, speed, or coverage claims about us are wrong in AI answers? | FactCheck prompts to add |
| What should I add before fall device launches and Black Friday? | Seasonal additions |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Product | Mobile, Home internet, Bundles, Business |
| Segment | Family, Student, Small business |
| Campaign | Device launch, Switcher offer, Price change |
| Season | Fall device launches, Black Friday, Back to school |
| FactCheck | Pricing, Speeds, Coverage, Fees |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Travel and hospitality
Source: https://docs.tryprofound.com/resources/prompt-guides/travel-hospitality

How to build a prompt set for a travel or hospitality brand, with best practices and examples

Travelers start with a destination, so the prompt set needs prompts for each destination you serve. Add prompts for trip types, such as family or business travel, and for events and peak dates before the busy weeks.

## Create your prompt set

### Topics to cover

Replace `[brand]` and `[destination]` in the example prompts with your own values.

| Topic | Example prompt |
| - | - |
| Destinations | What's the best area to stay in `[destination]` on a first trip? |
| New property openings | Is the new `[brand]` in `[destination]` worth booking? |
| Business trips | Hotels near the `[destination]` convention center with fast wifi |
| Loyalty promos and package deals | Which hotel loyalty program gives the best free nights? |
| Family trips | Family-friendly hotels in `[destination]` with connecting rooms |
| Summer and holiday travel | Where can I book a `[destination]` hotel with breakfast under \$200? |
| Romantic trips | Boutique hotels in `[destination]` under \$250 with a view |
| Festivals, school breaks, shoulder season | Best time to visit `[destination]` to skip the crowds |

> [!TIP]
>
> **Make prompts more specific.** Instead of *"What are the best boutique hotels in Lisbon?"*, try *"We're planning our anniversary trip and want to walk to dinner at night. What are the best boutique hotels in Lisbon under \$250 a night with a view?"*

### Best practices

* **Leave your brand name out of [discovery prompts](/getting-started/glossary#discovery-prompt).** Putting your name in the question inflates your [Visibility Score](/getting-started/glossary#visibility-score). Use it only in [validation](/getting-started/glossary#validation-prompt) and [FactCheck](/answer-engine-insights/factcheck/about) prompts.
* **One prompt = one topic.** Topics shouldn't overlap. Put funnel stage, campaign, and season in tags instead.
* **Ask it like a person would.** Write the full question someone would type or say.
* **Track only prompts that matter to the business.** Thousands of small variations of the same question make trends harder to read.
* **Keep coverage and depth apart.** [Coverage topics](/getting-started/glossary#coverage-topic) include broad prompts for each business line. Add two to five [depth topics](/getting-started/glossary#depth-topic) for your priorities, and keep new prompts out of the coverage topics.
* **Retire seasonal prompts.** Disable seasonal and news-driven prompts once they go stale. If you reword a prompt, you're measuring something new, so don't compare trends across the change.

## Refine your prompt set

The first prompt set you build gives you a baseline. Here's how you can expand it:

1. Build one set of prompts for each destination.
2. Add prompts that check cancellation, amenities, and price caps.
3. Add event prompts before the peak dates.

### Ask AI Marketer

Use [AI Marketer (Aim)](/aim/about) to make your prompts better. Paste one of these questions to an Aim chat to see which prompts to add, cut, or tag. Swap the `[values in brackets]` for your own.

| Ask | What you get |
| - | - |
| Show my visibility by destination. Which places have the thinnest prompt coverage? | Where to build a set next |
| What are travelers asking AI about `[destination]` that my set misses, like cancellation or price caps? | [Validation prompts](/getting-started/glossary#validation-prompt) to add |
| What festivals and events in `[destination]` should I add prompts for before peak dates? | Event-driven prompts |
| Which sources get cited for boutique hotel prompts in `[destination]`, and are we on them? | Which sources shape visibility in the destination |
| Do any AI answers get our cancellation policy or amenities wrong? | FactCheck prompts to add |

### Add tags to your prompts

[Tags](/getting-started/glossary#tag) can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.

| Tag group | Examples |
| - | - |
| Funnel stage | Informational TOFU, Commercial TOFU, Informational BOFU, Commercial BOFU |
| Destination | Lisbon, Porto, Algarve |
| Trip type | Business, Family, Romantic |
| Campaign | Loyalty promotion, New property, Package deal |
| Season | Summer, Holiday travel, School breaks, Events, Shoulder season |

> [!TIP]
>
> **IF YOU'RE STUCK**
>
> 1. **Pick the five topics that matter most to the business.**
> 2. **Write one prompt per funnel stage.** Four stages (Informational and Commercial, each for top and bottom of funnel) give you four base prompts per topic.
> 3. **Add about 10 variations.** Change the [persona](/answer-engine-insights/personas), use case, company size, or wording.

### Advanced tips

* **Use words like "brands" or "stores" in your prompts.** For example, *"What should I look for in a running shoe?"* returns a checklist of features, and *"Which brands make the best running shoes for flat feet?"* returns a list of brands, which is the answer you can measure your brand against.
* **Give every comparison a use case.** A *"`[brand]` vs `[competitor]`"* prompt alone returns an "it depends" answer. Add who is asking and why: *"`[brand]` vs `[competitor]` for a 50-person team that already uses Salesforce"* gives the AI enough context to pick one.
* **Spell out any unclear term the first time.** An acronym or internal product name can derail the answer. Write *"generative engine optimization tools"* rather than *"GEO tools"*, and name the product category rather than using a code word only your team uses for it.
* **Say what you want without pushing toward your own brand.** The question *"Which project management tools, like `[brand]`, are best for marketing teams?"* nudges the AI toward your brand and tells you nothing about whether it would have named you on its own.
* **Test your prompts.** Run each prompt once in an [answer engine](/getting-started/glossary#answer-engine) and read what it returns. A [discovery prompt](/getting-started/glossary#discovery-prompt) should list brands, a [comparison](/getting-started/glossary#competitor-prompt) should pick one, and a [validation](/getting-started/glossary#validation-prompt) should say yes or no. Learn more about different types of prompts in [Profound's AEO guide](https://www.tryprofound.com/aeo-guide/chapter-3#six-types-of-prompts).
* **Tag every prompt you add**. Learn more in [Categorize Your Prompts With Tags](/answer-engine-insights/how-tos/create-topics-and-prompts#categorize-your-prompts-with-tags).

## Next steps

* Read the [Promptologie](https://www.tryprofound.com/aeo-guide/chapter-3) chapter in Profound's AEO guide.
* Learn how to [create, manage, and tag prompts](/answer-engine-insights/prompt-designer) in Profound.
* Review the guide on [how to create effective prompts](/answer-engine-insights/how-tos/create-effective-prompts) using Prompt Volumes and intent data.


# Authentication
Source: https://docs.tryprofound.com/rest-api/authentication

Create a Profound API key and use it to authenticate your REST API requests

> [!NOTE]
>
> **Available on**: the Enterprise plan
>
> **Find it in**: Settings in the organization name menu at the top left of the screen > Account > API Keys

## Overview

To authenticate your API requests, you need an API key. This key identifies and authorizes your requests, ensuring only authorized users can access resources.

Your API key is a personal access token tied to your user account. It provides access to data in the organizations you belong to, including categories, regions, and category reports.

## Getting Your API Key

**1. Navigate to the API key generation portal**

1. Sign into the platform. Select your organization name at the top left of the screen, then select **Settings**.
2. In the left sidebar under **Account**, select **API Keys**. This will show your key generation portal and a table of your keys.

**2. Configure and create your key**

1. In the **Create API Key** banner, enter a **key name** and select the **Expiration** period.
2. Select **Create API Key** to generate the key.

> [!NOTE]
>
> For added security, you will not be able to retrieve your key again. Copy and store your API key securely before proceeding.

Your API keys are displayed in a table on the **API Keys** page. Return to this page at any time to see your keys, check expiry dates, or revoke an existing key.

## Authentication Methods

You can authenticate your requests using either method below:

### Header Authentication (Recommended)

Include your API key in the `X-API-Key` header:

```http Header Method theme={null}
POST /v1/reports/visibility HTTP/1.1
Host: api.tryprofound.com
X-API-Key: your_api_key_here
Content-Type: application/json
```

```bash cURL theme={null}
curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: your_api_key_here" \
  -H "Content-Type: application/json"
```

## API Key Scope & Permissions

Your API key provides access to:

* **Organization Data**: Data in the organizations you belong to
* **Categories**: Available data categories and their metadata
* **Regional Data**: Location-based information and filters
* **Reports**: Generated reports and analytics for your categories
* **Raw Data**: Unprocessed data within your accessible categories

### Key ownership

* **Keys belong to you**: Each API key is tied to the user who created it. Every request made with the key is attributed to that user, even when someone else sends it.
* **Keys are private to their owner**: Profound stores only a hash of each key, so no one, including organization admins, can view a key after it's created.
* **Access follows membership**: If you're removed from an organization, your key stays in place but no longer reaches data in that organization. To stop a key from working entirely, revoke it on the **API Keys** page.

## Security Best Practices

* **Keep keys private**: Your API key is sensitive information. Never share it publicly or commit it to version control. Store it securely and treat it like a password.
* **Use environment variables**: Store keys in environment variables or secure configuration files.
* **Rotate regularly**: Consider rotating your API keys periodically for enhanced security.
* **Monitor usage**: Check your API usage regularly to detect any unauthorized access.

## Authentication Errors

Common authentication errors and their meanings:

| Status Code | Error | Description |
| - | - | - |
| `401 Unauthorized` | Invalid API key | The provided API key is incorrect or expired |
| `403 Forbidden` | Insufficient permissions | Your API key doesn't have access to the requested resource |
| `429 Too Many Requests` | Rate limit exceeded | You've exceeded your API key rate limit |

## Need Help?

If you encounter authentication issues, [contact support](mailto:support@tryprofound.com) for assistance.


# Changelog
Source: https://docs.tryprofound.com/rest-api/changelog

New endpoints, changes, deprecations, and fixes in the Profound REST API

## Overview

This changelog tracks all changes, updates, and improvements to the Profound REST API. We document new features, bug fixes, breaking changes, deprecations, and security updates to help you stay informed and plan your integrations accordingly.

> [!NOTE]
>
> V2 covers the Answer Engine Insights and Agent Analytics reports under `/v2/reports/` (visibility, citations, sentiment, query-fanouts, factcheck, and factcheck/claims) and `/v2/prompts/answers`. Each endpoint also has a `/stream` variant that returns Server-Sent Events (SSE).
>
> V1 endpoints remain active, including V1 report endpoints. For more information about
> API versioning, see the [Introduction](/rest-api/introduction#api-versioning).

## Change categories

Changes are organized into the following categories:

* **Added** - New endpoints, features, or capabilities
* **Changed** - Modifications to existing endpoints or behavior
* **Deprecated** - Features that will be removed in a future version
* **Removed** - Features that have been removed
* **Fixed** - Bug fixes and corrections
* **Security** - Security improvements and vulnerability patches
* **Breaking** - Changes that require action to maintain compatibility

**September 2026**

Added

### Added

* Added page-level citation sentiment to [`POST /v2/reports/sentiment`](/rest-api/reports/query-sentiment-v2) with `source: "citation"`. Citation rows include positive sentiment, negative sentiment, model-balanced citation share, and rank; they can be sorted by citation share or positive sentiment.
* Added `citation_category` (`is`/`in`) and normalized-page search (`page` with `contains_case_insensitive`) filters for citation sentiment. Citation mode is page-only and supports both cursor pagination and the `/stream` SSE variant; it does not support `group_by`.
* New Prompt Volumes endpoints: [Keyword Volume](/rest-api/prompt-volumes/query-volume-v2) (`POST /v2/prompt-volumes/volume/on-the-fly`) for weekly and monthly volume projections for one keyword, and [Keyword Intent Shares](/rest-api/prompt-volumes/query-intents-v2) (`POST /v2/prompt-volumes/intents/on-the-fly`) for the intent mix behind it.

**August 2026**

Added

### Added

* New beta YouTube social reports: [Query YouTube Channels](/rest-api/reports/query-youtube-channels-v2) and [Query YouTube Videos](/rest-api/reports/query-youtube-videos-v2) for channel rollups, cited video rankings, and attribution modes.
* New endpoint: [Get Citation Tags](/api-reference/organization/get-category-citation-tags) - The custom citation tags defined for a category, as `{ value, name }` per tag. Pass any of these values to the new `citation_tag` filter below.
* New `citation_tag` filter on [`POST /v2/reports/citations`](/rest-api/reports/query-citations-v2) - Narrow a citations report to URLs carrying one or more of your custom tags (`is`/`in`; values in one `in` leaf are OR'd).

**July 2026**

Added

### Added

* New endpoint: [Get Citation Categories](/api-reference/organization/get-category-citation-categories) - The citation-category filter options for a category: the built-in buckets (Owned, Competition, Earned Media, Institution, PR Wire, Social, Other) plus any custom categories.

**June 2026**

Beta

### Added

* New **beta** `/v2/reports` endpoints — a unified `{ info, data }` request/response shape, accepting names **or** UUIDs in filters, with cursor pagination and a `/stream` (SSE) variant:
  * `POST /v2/reports/visibility` - Asset visibility, share of voice, and average position
  * `POST /v2/reports/citations` - Domain and page citation counts, share, and rank
  * `POST /v2/reports/sentiment` - Per-brand positive/negative sentiment with optional comparison periods
  * `POST /v2/reports/query-fanouts` - Search queries an AI generates behind the scenes when answering a prompt
  * `POST /v2/reports/factcheck` - FactCheck scores, with a `/stream` (SSE) variant
  * `POST /v2/reports/factcheck/claims` - FactCheck inaccurate claims, with a `/stream` (SSE) variant
* New **beta** endpoint: `POST /v2/prompts/answers` - Raw per-execution rows (mentions, citations, search queries), with a `/stream` (SSE) variant

### Deprecated

* The V1 report endpoints (`POST /v1/reports/visibility`, `/citations`, `/sentiment`, `/sentiment-v2`, `/query-fanouts`) are being phased out in favor of the V2 equivalents above. They remain active; a sunset date will be announced before removal.

**April 2026**

v1.5.0

### Added

* New endpoint: [List Agents](/api-reference/agents/list-agents) - List agents available to your organization
* New endpoint: [Get an Agent](/api-reference/agents/get-an-agent) - Retrieve an agent and its schema details
* New endpoint: [Run an Agent](/api-reference/agents/run-an-agent) - Start a new run for an agent
* New endpoint: [Get an Agent Run](/api-reference/agents/get-an-agent-run) - Retrieve the status and result details for an agent run

**April 2026**

Deprecation Notice

### Deprecated

* Deprecated raw Agent Analytics endpoints:
  * `POST /v1/logs/raw` - Get Logs
  * `POST /v1/logs/raw/bots` - Get Bots
* Deprecation started on `2026-04-10`
* API keys created on or after `2026-04-10` can no longer access these endpoints
* Existing API keys created before `2026-04-10` can continue using these endpoints until the sunset date of `2026-06-10`
* Migrate to the V2 aggregated report endpoints instead:
  * `POST /v2/reports/bots` for bot traffic reporting
  * `POST /v2/reports/referrals` for referral traffic reporting
* For examples using the replacement endpoints, see [Agent Analytics V2](/rest-api/examples/agent-analytics)

**March 2026**

v2.0.0

### Added

* New `/v2/reports` endpoints with **hourly granularity**, bucketed by UTC hour (see [Agent Analytics V2](/rest-api/examples/agent-analytics))
  * `POST /v2/reports/bots` - Bot traffic data at hourly resolution
  * `POST /v2/reports/referrals` - Referral traffic data at hourly resolution, including UTM and referrer distinction
* Because V2 uses UTC hour buckets rather than EST day buckets, users in any timezone can now align queries to their own local calendar days

**December 2025**

v1.4.0

### Added

* New endpoint: Get [Referrals Report](/api-reference/reports/get-referrals-report-v1) - Retrieve referral traffic data from daily aggregated reports
* New endpoint: Get [Bots Report](/api-reference/reports/get-bots-report-v1) - Retrieve bot traffic analytics with metrics for citations, indexing, training, and visit counts

### Changed

* Reorganized example documentation structure under new [Example Requests](/rest-api/examples/example) tab

### Removed

* Legacy examples page

**December 2025**

v1.3.0

### Added

* New endpoint: [Get Personas](/api-reference/organization/get-personas) - Retrieve all personas associated with your organization
* New endpoint: [Get Category Personas](/api-reference/organization/get-category-personas) - Retrieve all personas associated to the given category
* `persona` dimension to citations, visibility and sentiment reports
* `persona` to the prompt answers
* `model_id` to the prompt answers
* `sentiment_themes` to the prompt answers, deprecating `themes`. `sentiment_themes` returns the theme and its sentiment

### Changed

* Require a non-empty list value when using the `in` or `not_in` filter operator
* `created_at` field is not allowed to be disabled

### Fixed

* Avoid exporting duplicate prompts in the prompt answers.

**November 2025**

v1.2.0

### Added

* New endpoint: [Get Category Assets](/api-reference/organization/get-category-assets) - Retrieve assets associated with a specific category
* New endpoint: [Get Assets](/api-reference/organization/get-assets) - Retrieve all assets associated with your organization
* `asset_id` dimension in [Query Visibility](/api-reference/reports/query-visibility) and [Query Sentiment](/api-reference/reports/query-sentiment)

### Changed

* Require `hostname`, `path`, `root_domain` and `url` to be included as dimensions when used as filters

### Fixed

* `contains` filter not being applied correctly
* Typo in sentiment metric field name
  * **Before:** `ocurrences`
  * **After:** `occurrences`
* Filters containing UUIDs not being applied correctly
* Incorrect handling of mixed timezone and non-timezone values in `start_date` and `end_date` filters


# Date ranges and timezones
Source: https://docs.tryprofound.com/rest-api/date-ranges

How to work with dates and timezones in the Profound API

## Overview

All data in our system is stored in UTC, but our processing runs on an EST (Eastern Standard Time) schedule. Understanding how date ranges work is critical to querying the correct data for your use case.

> [!WARNING]
>
> Incorrect timezone handling is the most common cause of missing or unexpected data in API responses. Please read this guide carefully.

## System Behavior

### Data Processing Schedule

Our system processes data according to the EST timezone:

* The "day" of January 1st EST starts at **00:00 EST** (05:00 UTC)
* The "day" of January 1st EST ends at **23:59 EST** (04:59 UTC next day)

All data is stored with UTC timestamps in our database.

### Date Format Support

The API accepts dates in multiple formats:

1. **Date only** (recommended for daily queries): `2025-01-01`
2. **Date and time without timezone**: `2025-01-01T00:00:00`
3. **Date and time with UTC timezone**: `2025-01-01T05:00:00Z`

## How Dates Are Interpreted

### Without Timezone Specification

When you provide a date **without a timezone** (no `Z` suffix), the API interprets it as **EST**:

```json theme={null}
{
  "start_date": "2025-01-01",
  "end_date": "2025-01-31"
}
```

**What happens:**

* `2025-01-01` → Interpreted as `2025-01-01T00:00:00 EST` → Converted to `2025-01-01T05:00:00Z` UTC
* `2025-01-31` → Interpreted as `2025-01-31T00:00:00 EST` → Converted to `2025-01-31T05:00:00Z` UTC

This matches how our system defines "days" and is the **recommended approach** for most use cases.

### With UTC Timezone (Z suffix)

When you provide a date **with the Z suffix**, the API interprets it as **literal UTC**:

```json theme={null}
{
  "start_date": "2025-01-01T00:00:00Z",
  "end_date": "2025-02-01T00:00:00Z"
}
```

**What happens:**

* `2025-01-01T00:00:00Z` → Used as-is in UTC
* `2025-02-01T00:00:00Z` → Used as-is in UTC

> [!WARNING]
>
> Using UTC timestamps (with Z) requires you to manually convert from EST to UTC. This approach exists for backwards compatibility but is **not recommended** for new integrations.

## Examples

### Correct: Query January 2025 data (EST days)

```json theme={null}
{
  "start_date": "2025-01-01",
  "end_date": "2025-02-01"
}
```

This returns all data from:

* Start: January 1st, 00:00 EST (05:00 UTC on Jan 1st)
* End: February 1st, 00:00 EST (05:00 UTC on Feb 1st)

The range is inclusive (`<=`), but since data typically doesn't have timestamps exactly at 00:00, this effectively captures all of January.

### Incorrect: Query January 2025 with UTC midnight

```json theme={null}
{
  "start_date": "2025-01-01T00:00:00Z",
  "end_date": "2025-02-01T00:00:00Z"
}
```

This returns data from:

* Start: December 31st, 19:00 EST (00:00 UTC on Jan 1st)
* End: January 31st, 19:00 EST (00:00 UTC on Feb 1st)

**Problem:** You're missing 5 hours at the start and including 5 hours you don't want at the end.

### Correct: Query January 2025 with UTC (manual conversion)

If you must use UTC timestamps, you need to account for the EST offset:

```json theme={null}
{
  "start_date": "2025-01-01T05:00:00Z",
  "end_date": "2025-02-01T05:00:00Z"
}
```

This correctly captures all of January (EST days), but requires manual timezone math.

## Time Ranges

### Date-only format defaults to midnight

When using date-only format, times default to `00:00:00`:

```json theme={null}
{
  "start_date": "2025-01-01"  // Equivalent to 2025-01-01T00:00:00 EST
}
```

### Specifying exact times

You can specify exact times for more granular queries:

```json theme={null}
{
  "start_date": "2025-01-01T08:00:00",  // 8 AM EST
  "end_date": "2025-01-01T17:00:00"     // 5 PM EST
}
```

This queries data from 8 AM to 5 PM EST on January 1st.

### Date range behavior

Date ranges are **inclusive** on both ends `(start <= x <= end)`:

```json theme={null}
{
  "start_date": "2025-01-01",
  "end_date": "2025-01-02"
}
```

This includes:

* All of January 1st (from 00:00 EST onwards)
* January 2nd at exactly 00:00 EST
* Since data is timestamped throughout the day, there typically isn't data exactly at 00:00, so in practice this captures all of January 1st

To query a single day, set `end_date` to the next day at 00:00.

## Best Practices

> [!TIP]
>
> **Recommended:** Use date-only format without timezone for daily queries

```json theme={null}
{
  "start_date": "2025-01-01",
  "end_date": "2025-02-01"
}
```

> [!TIP]
>
> **Recommended:** Use datetime without timezone for intraday queries

```json theme={null}
{
  "start_date": "2025-01-15T09:00:00",
  "end_date": "2025-01-15T17:00:00"
}
```

> [!TIP]
>
> **Recommended:** Use datetime without timezone for hourly report queries (via [Agent Analytics](/rest-api/examples/agent-analytics) **V2 (hourly)** tab)

```json theme={null}
{
  "start_date": "2025-01-15T14:00:00",
  "end_date": "2025-01-15T14:59:59"
}
```

> [!WARNING]
>
> **Fractional-offset timezone users:** When querying at hour-level granularity with `/v2/reports/*`, query 2 consecutive hours to get complete data for a single local hour. See [Fractional-Offset Timezones](#fractional-offset-timezones) for details.

> [!WARNING]
>
> **Avoid:** Using Z suffix unless you understand UTC offset implications

```json theme={null}
{
  "start_date": "2025-01-01T00:00:00Z",  // Probably not what you want
  "end_date": "2025-02-01T00:00:00Z"
}
```

## Daylight Saving Time

EST observes Daylight Saving Time (DST):

* **Standard Time (EST):** UTC-5 (typically November - March)
* **Daylight Time (EDT):** UTC-4 (typically March - November)

The API automatically handles DST transitions. When you specify dates without timezone, the conversion to UTC accounts for whether DST was active on that date.

**Example during DST:**

```json theme={null}
{
  "start_date": "2025-07-01"  // Interpreted as EDT (UTC-4)
}
```

→ Converted to `2025-07-01T04:00:00Z` UTC (not 05:00 because of DST)

## Fractional-Offset Timezones

Some timezones use fractional UTC offsets rather than whole-hour offsets:

| Timezone | UTC Offset |
| - | - |
| India Standard Time (IST) | +5:30 |
| Afghanistan Time (AFT) | +4:30 |
| Myanmar Time (MMT) | +6:30 |
| Nepal Time (NPT) | +5:45 |
| Chatham Islands (CHAST) | +12:45 |
| Marquesas Islands (MART) | -9:30 |

Because hourly buckets are aligned to whole-hour UTC boundaries, a single local hour in these timezones will always straddle two hourly buckets. To get complete data for any given local hour, you must query **2 consecutive hours**.

### Example: Querying for 2:00 PM IST

IST is UTC+5:30, so 2:00 PM IST = 08:30 UTC. The hour from 2:00 PM to 3:00 PM IST (08:30 - 09:30 UTC) spans two UTC hour buckets:

* **08:00 - 08:59 UTC** (contains the first 30 minutes: 2:00 PM - 2:30 PM IST)
* **09:00 - 09:59 UTC** (contains the last 30 minutes: 2:30 PM - 3:00 PM IST)

To retrieve the full hour, query both buckets:

```json theme={null}
{
  "start_date": "2025-01-15T08:00:00Z",
  "end_date": "2025-01-15T09:59:59Z"
}
```

> [!WARNING]
>
> This applies when querying at **hour-level granularity** via `/v2/reports/*`. If you are in a fractional-offset timezone and request data by hour, always request 2 consecutive hours to avoid missing data that falls in the adjacent bucket.

## Troubleshooting

### Missing data at day boundaries

**Problem:** You're querying January 1st but missing the first or last few hours.

**Solution:** Make sure you're not using the `Z` suffix. Use `"start_date": "2025-01-01"` instead of `"start_date": "2025-01-01T00:00:00Z"`.

### Data from previous/next day appearing

**Problem:** Your daily query includes data from adjacent days.

**Solution:** You're likely using `Z` suffix with midnight UTC, which doesn't align with EST days. Remove the timezone suffix.

### Unexpected DST behavior

**Problem:** Your queries have a 1-hour difference during DST transitions.

**Solution:** Let the API handle DST automatically by using dates without timezone. If you must use UTC, remember the offset changes between EST (-5) and EDT (-4).

## Future Changes

> [!NOTE]
>
> In a future version of the API, we plan to support full ISO 8601 timezone offsets (for example, `-05:00`, `+09:00`) to enable multi-timezone queries. Dates without timezone specification will be deprecated at that time.

For now, the recommended approach is to use dates without timezone and let the API handle the EST conversion for you.


# Agent Analytics
Source: https://docs.tryprofound.com/rest-api/examples/agent-analytics

Sample requests for Agent Analytics endpoints.

This guide provides practical examples of common API requests. All examples use JSON format and require authentication via API key.

# Reports

**Daily Reports (V1)**

Examples for **V1** report endpoints (`/v1/reports/*`): daily pre-aggregated data in EST.
For hourly granularity, use the **Hourly Reports (V2)** tab above.

> [!NOTE]
>
> Raw Agent Analytics log endpoints are retired. For raw per-execution
> prompt-answer rows, use `POST /v2/prompts/answers`. For bot and referral
> traffic, use the aggregated `/v2/reports/bots` and `/v2/reports/referrals`
> endpoints.

* **Capacity:** Full date ranges in one request using aggregated report endpoints.
* **Granularity:** Data is bucketed by day. Timestamps are normalized to the start of each day in EST.

## Bot Reports

**Bot Highlights**

Get a count of visits by bot type:

![Dashboard Metrics](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-bot-visit-highlights.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=f61cba46eb807a2712365c4ad71b8e1e)

```http theme={null}
POST /v1/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "metrics": ["citations", "training", "indexing"]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [
        20640,
        8450,
        3646
      ],
      "dimensions": []
    }
  ]
}
```

**Citations by Day**

Get a count of ChatGPT citations by day:

![By Day Breakdown](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-citations-by-bot-by-day.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=42042eb802d987cc0f20ede22bedfe1a)

> [!TIP]
>
> Note you can easily retrieve the full daily list broken down by bot by substituting:
>
> ```
> "dimensions": ["date", "bot_name"],
> "filters": [],
> ```

```http theme={null}
POST /v1/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["date"],
  "filters": [
      {"field": "bot_name", "operator": "is", "value": "ChatGPT-User"}
  ],
  "order_by": {
      "date": "desc"
  },
  "metrics": ["citations"]
}
```

```json theme={null}
{
    "info": {},  // removed for brevity
    "data": [
        {
            "metrics": [
                2698
            ],
            "dimensions": [
                "2025-12-18"
            ]
        },
        {
            "metrics": [
                2847
            ],
            "dimensions": [
                "2025-12-17"
            ]
        },
        {
            "metrics": [
                2986
            ],
            "dimensions": [
                "2025-12-16"
            ]
        },
        {
            "metrics": [
                3029
            ],
            "dimensions": [
                "2025-12-15"
            ]
        },
        {
            "metrics": [
                2075
            ],
            "dimensions": [
                "2025-12-14"
            ]
        },
        {
            "metrics": [
                1881
            ],
            "dimensions": [
                "2025-12-13"
            ]
        },
        {
            "metrics": [
                2717
            ],
            "dimensions": [
                "2025-12-12"
            ]
        }
    ]
}
```

**Page Visits by Provider**

Get a count of page visits by bot type, scoped to a provider:

> [!NOTE]
>
> Note the OpenAI **Platforms** selection in the top right.

![Page Visits by Provider](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-page-visits-by-provider.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=2234a9d846acd32bf2379d7ceca88089)

> [!TIP]
>
> See metrics for all providers by removing the filter.

```http theme={null}
POST /v1/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["path"],
  "filters": [
      {"field": "bot_provider", "operator": "is", "value": "openai"}
  ],
  "order_by": {"citations": "desc"},
  "metrics": ["citations", "indexing", "training"],
  "pagination": {
      "limit": 10,
      "offset": 0
  }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
      {
        "metrics": [
          1255,
          1614,
          38
        ],
        "dimensions": [
          "/"
        ]
      },
      {
        "metrics": [
          643,
          5,
          1
        ],
        "dimensions": [
          "/guides/generative-engine-optimization-geo-guide-2025"
        ]
      },
      {
        "metrics": [
          603,
          18,
          3
        ],
        "dimensions": [
          "/blog/now-tracking-gpt-5-2-in-chatgpt"
        ]
      },
      {
        "metrics": [
          584,
          7,
          3
        ],
        "dimensions": [
          "/blog/ai-platform-citation-patterns"
        ]
      },
      {
        "metrics": [
          445,
          1,
          0
        ],
        "dimensions": [
          "/blog/best-generative-engine-optimization-tools"
        ]
      },
      {
        "metrics": [
          396,
          14,
          1
        ],
        "dimensions": [
          "/profound-index"
        ]
      },
      {
      "metrics": [
          385,
          6,
          1
      ],
      "dimensions": [
          "/guides/what-is-answer-engine-optimization"
      ]
      },
      {
        "metrics": [
          376,
          2,
          0
        ],
        "dimensions": [
          "/blog/semrush-ai-visibility-toolkit-review"
        ]
      },
      {
        "metrics": [
          372,
          5,
          2
        ],
        "dimensions": [
          "/blog/understanding-grok-a-comprehensive-guide-to-grok-websearch-grok-deepsearch"
        ]
      },
      {
        "metrics": [
          371,
          7,
          1
        ],
        "dimensions": [
          "/blog/choosing-ai-visibility-provider"
        ]
      }
  ]
}
```

## Referral Reports

**Referrals By Platform**

Get a count of referrals broken down by platform:

> [!NOTE]
>
> Note:
>
> `other` represents referrals from **Traditional Sources**, for example LinkedIn, Facebook, etc.
>
> `internal` represents referrals from within your website, for example a blog post linking to another blog post.

![Referrals By Platform](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-platform.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=d4ef5e6ecb21e0f05cd03cbcc07fa22b)

```http theme={null}
POST /v1/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["referral_source"],
  "filters": [
    {"field": "referral_source", "operator": "not_in", "value": ["none", "internal"]}
  ],
  "metrics": ["visits"]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [
        222022
      ],
      "dimensions": [
        "other"
      ]
    },
    {
      "metrics": [
        5171
      ],
      "dimensions": [
        "openai"
      ]
    },
    {
      "metrics": [
        102
      ],
      "dimensions": [
        "perplexity"
      ]
    },
    {
      "metrics": [
        39
      ],
      "dimensions": [
        "gemini"
      ]
    },
    {
      "metrics": [
        22
      ],
      "dimensions": [
        "anthropic"
      ]
    },
    {
      "metrics": [
        2
      ],
      "dimensions": [
        "deepseek"
      ]
    }
  ]
}
```

**Bot Referrals By Day**

Get a count of referrals from AI bots broken down by day:

![Bot Referrals By Day](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-day.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=4af067d50a0ec61cba78d505098338a8)

```http theme={null}
POST /v1/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["date"],
  "filters": [
    {"field": "referral_source", "operator": "not_in", "value": ["none", "other", "internal"]}
  ],
  "order_by": {"date": "desc"},
  "metrics": ["visits"]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [
        977
      ],
      "dimensions": [
        "2025-12-18"
      ]
    },
    {
      "metrics": [
        906
      ],
      "dimensions": [
        "2025-12-17"
      ]
    },
    {
      "metrics": [
        719
      ],
      "dimensions": [
        "2025-12-16"
      ]
    },
    {
      "metrics": [
        571
      ],
      "dimensions": [
        "2025-12-15"
      ]
    },
    {
      "metrics": [
        665
      ],
      "dimensions": [
        "2025-12-14"
      ]
    },
    {
      "metrics": [
        655
      ],
      "dimensions": [
        "2025-12-13"
      ]
    },
    {
      "metrics": [
        844
      ],
      "dimensions": [
        "2025-12-12"
      ]
    }
  ]
}
```

**Page Level Referrals**

Get a count of referrals broken down by a given page:

> [!NOTE]
>
> Note `/blog` in the search bar.

![Page Level Referrals](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-page.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=bcd6917a05026d5aae2d3d95391101c9)

> [!TIP]
>
> **% of Bot Referrals** can be calculated by summing the total number of referrals, and dividing each bot's referrals by the total.

```http theme={null}
POST /v1/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["path"],
  "filters": [
    {"field": "referral_source", "operator": "not_in", "value": ["none", "other", "internal"]},
    {"field": "path", "operator": "contains", "value": "/blog"}
  ],
  "metrics": ["visits"],
  "order_by": {"visits": "desc"},
  "pagination": {"limit": 10, "offset": 0}
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [
        915
      ],
      "dimensions": [
        "/blog/ai-platform-citation-patterns"
      ]
    },
    {
      "metrics": [
        350
      ],
      "dimensions": [
        "/blog/bring-profound-data-directly-into-your-ai-workflow-with-mcp"
      ]
    },
    {
      "metrics": [
        99
      ],
      "dimensions": [
        "/blog/citation-overlap-strategy"
      ]
    },
    {
      "metrics": [
        87
      ],
      "dimensions": [
        "/blog/the-data-on-reddit-and-ai-search"
      ]
    },
    {
      "metrics": [
        65
      ],
      "dimensions": [
        "/blog/best-generative-engine-optimization-tools"
      ]
    },
    {
      "metrics": [
        58
      ],
      "dimensions": [
        "/blog/chatgpt-entity-update"
      ]
    },
    {
      "metrics": [
        56
      ],
      "dimensions": [
        "/blog/prompt-volumes-the-new-way-to-see-what-customers-ask-answer-engines"
      ]
    },
    {
      "metrics": [
        50
      ],
      "dimensions": [
        "/blog/seeing-what-customers-see-direct-ai-search-engine-monitoring-vs-api-limitations"
      ]
    },
    {
      "metrics": [
        40
      ],
      "dimensions": [
        "/blog/choosing-ai-visibility-provider"
      ]
    },
    {
      "metrics": [
        33
      ],
      "dimensions": [
        "/blog/ai-search-volatility"
      ]
    }
  ]
}
```

**Referrals By Bot**

Get a count of referrals broken down by bot:

![Referrals By Bot](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-bot.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=a50dbee002c88fe8a851dc3c3de364fa)

> [!TIP]
>
> **% of Referrals** here means the proportion of referred visits from non-internal sources that came from this bot.
>
> To calculate this, sum the total number of referrals in the response:
>
> `total_referrals = 222126 + 5172 + 102 + 39 + 22 + 2 = 227481`
>
> Then, for each bot, divide its referrals by the total:
>
> `ChatGPT % = (5172 / 227481) * 100 = 2.27%`
>
> ...

```http theme={null}
POST /v1/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["referral_source"],
  "filters": [
    {"field": "referral_source", "operator": "not_in", "value": ["none", "internal"]}
  ],
  "metrics": ["visits"],
  "order_by": {"visits": "desc"}
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [
        222126
      ],
      "dimensions": [
        "other"
      ]
    },
    {
      "metrics": [
        5172
      ],
      "dimensions": [
        "openai"
      ]
    },
    {
      "metrics": [
        102
      ],
      "dimensions": [
        "perplexity"
      ]
    },
    {
      "metrics": [
        39
      ],
      "dimensions": [
        "gemini"
      ]
    },
    {
      "metrics": [
        22
      ],
      "dimensions": [
        "anthropic"
      ]
    },
    {
      "metrics": [
        2
      ],
      "dimensions": [
        "deepseek"
      ]
    }
  ]
}
```

**Hourly Reports (V2)**

V2 endpoints provide reports at **hourly granularity**, with data bucketed by UTC hour. They support all the same dimensions, metrics and filters as V1, plus new fields unique to the hourly tables - **UTM** and **referral** traffic distinction.

V1 daily aggregates are fixed to ET, which means users in other timezones had no way to get data bucketed into their own local days. V2 solves this: because data is bucketed by UTC hour, you can convert your local day boundaries to UTC and query data that lines up with your own calendar days.

> [!NOTE]
>
> **V1 vs V2:** Use the **V1 (daily)** tab for daily pre-aggregated reports
> (EST-based). V2 is for hour-level granularity (UTC). Raw per-execution
> prompt-answer rows are available through `POST /v2/prompts/answers`.

> [!WARNING]
>
> **Fractional-offset timezones:** If your local timezone uses a fractional UTC offset (for example, IST +5:30, Nepal +5:45), a single local hour spans two UTC hour buckets. When querying at hour-level granularity, request **2 consecutive hours** to get complete data for that local hour. See [Date Ranges & Timezones](/rest-api/date-ranges#fractional-offset-timezones) for details.

## Bot Reports

Returns bot traffic data aggregated from the hourly bot breakdown table.

**Bot Highlights**

Get a count of visits by bot type:

![Dashboard Metrics](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-bot-visit-highlights.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=f61cba46eb807a2712365c4ad71b8e1e)

```http theme={null}
POST /v2/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "metrics": ["citations", "training", "indexing"]
}
```

**Citations by Day**

Get a count of ChatGPT citations by day:

![By Day Breakdown](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-citations-by-bot-by-day.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=42042eb802d987cc0f20ede22bedfe1a)

```http theme={null}
POST /v2/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["date"],
  "filters": [
    { "field": "bot_name", "operator": "is", "value": "ChatGPT-User" }
  ],
  "order_by": { "date": "desc" },
  "metrics": ["citations"]
}
```

**Page Visits by Provider**

Get top paths by citations for a specific provider:

![Page Visits by Provider](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-page-visits-by-provider.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=2234a9d846acd32bf2379d7ceca88089)

```http theme={null}
POST /v2/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["path"],
  "filters": [
    { "field": "bot_provider", "operator": "is", "value": "openai" }
  ],
  "order_by": { "citations": "desc" },
  "metrics": ["citations", "indexing", "training"],
  "pagination": { "limit": 10, "offset": 0 }
}
```

**Hourly Bot Traffic**

Get hourly bot traffic breakdown for a UTC day. Use UTC timestamps to align the query to your local calendar day:

```http theme={null}
POST /v2/reports/bots HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-16T00:00:00Z",
  "end_date": "2025-12-16T23:59:59Z",
  "dimensions": ["hour"],
  "metrics": ["count", "citations"],
  "order_by": { "hour": "asc" }
}
```

### Request body

<table>
  <colgroup>
    <col />

    <col />

    <col />

    <col />

    <col />
  </colgroup>

  <thead>
    <tr>
      <th>Field</th>
      <th>Type</th>
      <th>Required</th>
      <th>Default</th>
      <th>Description</th>
    </tr>
  </thead>

  <tbody>
    <tr>
      <td>`domain`</td>
      <td>string</td>
      <td>yes</td>
      <td>-</td>
      <td>Domain to query (e.g. `www.tryprofound.com`)</td>
    </tr>

    <tr>
      <td>`start_date`</td>
      <td>string</td>
      <td>yes</td>
      <td>-</td>
      <td>Start date in UTC. Accepts `YYYY-MM-DD`, `YYYY-MM-DD HH:MM`, `YYYY-MM-DD HH:MM:SS`, or full ISO timestamp</td>
    </tr>

    <tr>
      <td>`end_date`</td>
      <td>string</td>
      <td>no</td>
      <td>now (UTC)</td>
      <td>End date in UTC. Same formats as start\_date</td>
    </tr>

    <tr>
      <td>`metrics`</td>
      <td>string\[]</td>
      <td>yes</td>
      <td>-</td>
      <td>One or more metrics to return (see below)</td>
    </tr>

    <tr>
      <td>`dimensions`</td>
      <td>string\[]</td>
      <td>no</td>
      <td>`[]`</td>
      <td>Fields to group results by (see below)</td>
    </tr>

    <tr>
      <td>`filters`</td>
      <td>filter\[]</td>
      <td>no</td>
      <td>`[]`</td>
      <td>Filters to narrow results (see below)</td>
    </tr>

    <tr>
      <td>`date_interval`</td>
      <td>string</td>
      <td>no</td>
      <td>`"day"`</td>
      <td>Bucket size for `date` dimension: `"hour"` or `"day"`</td>
    </tr>

    <tr>
      <td>`order_by`</td>
      <td>object</td>
      <td>no</td>
      <td>first metric desc</td>
      <td>Map of field -> `"asc"` or `"desc"`</td>
    </tr>

    <tr>
      <td>`pagination`</td>
      <td>object</td>
      <td>no</td>
      <td>`{ limit: 10000, offset: 0 }`</td>
      <td>Pagination settings</td>
    </tr>
  </tbody>
</table>

### Metrics

| Metric | Description |
| - | - |
| `count` | Total unique bot visits (all bot types) |
| `citations` | Unique visits from AI assistant bots (e.g. ChatGPT-User, Perplexity-User) |
| `indexing` | Unique visits from web indexing bots (e.g. Googlebot, bingbot) |
| `training` | Unique visits from AI training bots (e.g. GPTBot, ClaudeBot) |
| `last_visit` | Most recent visit timestamp |

### Dimensions

| Dimension | Description |
| - | - |
| `date` | Bucketed by `date_interval` (day or hour). UTC-based |
| `hour` | Raw UTC hour bucket — equivalent to `date` + `date_interval: "hour"` with no extra aggregation |
| `path` | URL path |
| `bot_name` | Bot name (e.g. `ChatGPT-User`, `GPTBot`) |
| `bot_provider` | Bot provider (e.g. `openai`, `anthropic`) |
| `bot_type` | Bot category: `ai_assistant`, `index`, `ai_training`, `ai_agent` |

### Filters

| Field | Operators | Values |
| - | - | - |
| `path` | `is`, `not_is`, `contains`, `not_contains`, `in`, `not_in` | URL path string |
| `bot_name` | `is`, `not_is`, `contains`, `not_contains`, `in`, `not_in` | Bot name string |
| `bot_provider` | `is`, `not_is`, `contains`, `not_contains`, `in`, `not_in` | Provider string |
| `bot_type` | `is`, `not_is`, `in`, `not_in` | `ai_assistant`, `index`, `ai_training`, `ai_agent` |

***

## Referral Reports

Returns referral traffic data aggregated from the hourly referral breakdown table.

**Referrals By Platform**

Get a count of referrals broken down by platform (excluding internal traffic):

![Referrals By Platform](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-platform.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=d4ef5e6ecb21e0f05cd03cbcc07fa22b)

```http theme={null}
POST /v2/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["referral_source"],
  "filters": [
    { "field": "referral_source", "operator": "not_in", "value": ["none", "internal"] }
  ],
  "metrics": ["visits"]
}
```

**Bot Referrals By Day**

Get a count of referrals from AI bots broken down by day:

![Bot Referrals By Day](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-day.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=4af067d50a0ec61cba78d505098338a8)

```http theme={null}
POST /v2/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["date"],
  "filters": [
    { "field": "referral_source", "operator": "not_in", "value": ["none", "other", "internal"] }
  ],
  "order_by": { "date": "desc" },
  "metrics": ["visits"]
}
```

**Page Level Referrals**

Get top blog paths by external referral visits:

![Page Level Referrals](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-page.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=bcd6917a05026d5aae2d3d95391101c9)

```http theme={null}
POST /v2/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["path"],
  "filters": [
    { "field": "referral_source", "operator": "not_in", "value": ["none", "other", "internal"] },
    { "field": "path", "operator": "contains", "value": "/blog" }
  ],
  "metrics": ["visits"],
  "order_by": { "visits": "desc" },
  "pagination": { "limit": 10, "offset": 0 }
}
```

**Referrals By Bot**

Get top referral sources ranked by visits:

![Referrals By Bot](https://mintcdn.com/profound-37face47/_84snNj2y4O4wlvX/images/rest-api/aa-examples/get-referrals-by-bot.png?fit=max\&auto=format\&n=_84snNj2y4O4wlvX\&q=85\&s=a50dbee002c88fe8a851dc3c3de364fa)

```http theme={null}
POST /v2/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-12",
  "end_date": "2025-12-18",
  "dimensions": ["referral_source"],
  "filters": [
    { "field": "referral_source", "operator": "not_in", "value": ["none", "internal"] }
  ],
  "metrics": ["visits"],
  "order_by": { "visits": "desc" }
}
```

**UTM Traffic**

Get UTM referral traffic by Hour:

![Hourly UTM Traffic](https://mintcdn.com/profound-37face47/gM-2pDTh8T5hVvln/images/rest-api/aa-examples/get-utm-traffic-hourly.png?fit=max\&auto=format\&n=gM-2pDTh8T5hVvln\&q=85\&s=ef4d4e3f68a345f9a9cd96705066c04e)

```http theme={null}
POST /v2/reports/referrals HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "domain": "www.tryprofound.com",
  "start_date": "2025-12-16",
  "end_date": "2025-12-17",
  "dimensions": ["hour"],
  "filters": [
    { "field": "referral_type", "operator": "is", "value": "utm" }
  ],
  "metrics": ["visits"],
  "order_by": { "hour": "asc" }
}
```

### Request body

<table>
  <colgroup>
    <col />

    <col />

    <col />

    <col />

    <col />
  </colgroup>

  <thead>
    <tr>
      <th>Field</th>
      <th>Type</th>
      <th>Required</th>
      <th>Default</th>
      <th>Description</th>
    </tr>
  </thead>

  <tbody>
    <tr>
      <td>`domain`</td>
      <td>string</td>
      <td>yes</td>
      <td>-</td>
      <td>Domain to query</td>
    </tr>

    <tr>
      <td>`start_date`</td>
      <td>string</td>
      <td>yes</td>
      <td>-</td>
      <td>Start date in UTC. Accepts `YYYY-MM-DD`, `YYYY-MM-DD HH:MM`, `YYYY-MM-DD HH:MM:SS`, or full ISO timestamp</td>
    </tr>

    <tr>
      <td>`end_date`</td>
      <td>string</td>
      <td>no</td>
      <td>now (UTC)</td>
      <td>End date in UTC. Same formats as start\_date</td>
    </tr>

    <tr>
      <td>`metrics`</td>
      <td>string\[]</td>
      <td>yes</td>
      <td>-</td>
      <td>One or more metrics to return (see below)</td>
    </tr>

    <tr>
      <td>`dimensions`</td>
      <td>string\[]</td>
      <td>no</td>
      <td>`[]`</td>
      <td>Fields to group results by (see below)</td>
    </tr>

    <tr>
      <td>`filters`</td>
      <td>filter\[]</td>
      <td>no</td>
      <td>`[]`</td>
      <td>Filters to narrow results (see below)</td>
    </tr>

    <tr>
      <td>`date_interval`</td>
      <td>string</td>
      <td>no</td>
      <td>`"day"`</td>
      <td>Bucket size for `date` dimension: `"hour"` or `"day"`</td>
    </tr>

    <tr>
      <td>`order_by`</td>
      <td>object</td>
      <td>no</td>
      <td>first metric desc</td>
      <td>Map of field -> `"asc"` or `"desc"`</td>
    </tr>

    <tr>
      <td>`pagination`</td>
      <td>object</td>
      <td>no</td>
      <td>`{ limit: 10000, offset: 0 }`</td>
      <td>Pagination settings</td>
    </tr>
  </tbody>
</table>

### Metrics

| Metric | Description |
| - | - |
| `visits` | Unique page visits |
| `last_visit` | Most recent visit timestamp |

### Dimensions

| Dimension | Description |
| - | - |
| `date` | Bucketed by `date_interval` (day or hour). UTC-based |
| `hour` | Raw UTC hour bucket — equivalent to `date` + `date_interval: "hour"` with no extra aggregation |
| `path` | URL path |
| `referral_source` | Referral source (e.g. `openai`, `perplexity`, `anthropic`, `other`, `internal`, `none`) |
| `referral_type` | Traffic category: `internal`, `referer`, `utm`, `none` |

### Filters

| Field | Operators | Values |
| - | - | - |
| `path` | `is`, `not_is`, `contains`, `not_contains`, `in`, `not_in` | URL path string |
| `referral_source` | `is`, `not_is`, `contains`, `not_contains`, `in`, `not_in` | Referral source string |
| `referral_type` | `is`, `not_is`, `in`, `not_in` | `internal`, `referer`, `utm`, `none` |

***

## Raw Data

> [!WARNING]
>
> The raw Agent Analytics endpoints `POST /v1/logs/raw` and
> `POST /v1/logs/raw/bots` were retired on 2026-06-10. Use the [V2 bot and
> referral report examples](/rest-api/examples/agent-analytics) for aggregated traffic data,
> or [V2 prompt answers](/rest-api/reports/query-answers-v2) for raw
> per-execution rows.


# Answer Engine Insights
Source: https://docs.tryprofound.com/rest-api/examples/answer-engine-insights

Sample requests for Answer Engine Insights endpoints.

This guide provides practical examples of common API requests. All examples use JSON format and require authentication via API key.

> [!NOTE]
>
> Replace `your_api_key`, `your_category_id`, and `your_company_name` with your
> actual values. Category IDs can be obtained from `/v1/org/categories`.

> [!WARNING]
>
> All examples use simplified date formats. For production use, please read [Date Ranges & Timezones](/rest-api/date-ranges) to understand timezone handling and avoid common pitfalls with date parameters.

## Visibility Reports

Retrieve visibility metrics and performance data for companies within specific categories. See the full [endpoint reference](/api-reference/reports/query-visibility) for all available metrics, dimensions, and filters.

**Visibility Score**

![Visibility Score](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/visibility-score.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=7f23b1edd6d2bf9566e23b404cc9a118)

**Company Visibility Trends Over Time**

Track a specific company's visibility performance with daily granularity:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "date_interval": "day",
  "metrics": ["visibility_score"],
  "dimensions": ["date"],
  "filters": [
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.45], "dimensions": ["2026-01-01"] },
    { "metrics": [0.48], "dimensions": ["2026-01-02"] },
    { "metrics": [0.52], "dimensions": ["2026-01-03"] },
    { "metrics": [0.51], "dimensions": ["2026-01-04"] },
    { "metrics": [0.55], "dimensions": ["2026-01-05"] },
    { "metrics": [0.58], "dimensions": ["2026-01-06"] },
    { "metrics": [0.60], "dimensions": ["2026-01-07"] }
  ]
}
```

**Use case:** Monitor daily performance changes and identify trends.

**Top 5 Companies by Visibility Score**

Get the top 5 companies with the highest visibility score in a category:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["visibility_score"],
  "dimensions": ["asset_name"],
  "pagination": {
    "limit": 5
  }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.583], "dimensions": ["Company A"] },
    { "metrics": [0.343], "dimensions": ["Company B"] },
    { "metrics": [0.231], "dimensions": ["Company C"] },
    { "metrics": [0.216], "dimensions": ["Company D"] },
    { "metrics": [0.173], "dimensions": ["Company E"] }
  ]
}
```

**Use case:** Identify market leaders and benchmark against top performers.

**Share of Voice**

![Share of Voice](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/share-of-voice.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=bc57db58e7b6e8f8ff994eb14427b37b)

**Share of Voice by Platform**

Analyze visibility score and share of voice across different platforms for a specific company:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["visibility_score", "share_of_voice"],
  "dimensions": ["model"],
  "filters": [
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.72, 0.15], "dimensions": ["ChatGPT"] },
    { "metrics": [0.65, 0.12], "dimensions": ["Perplexity"] },
    { "metrics": [0.58, 0.09], "dimensions": ["Gemini"] },
    { "metrics": [0.41, 0.06], "dimensions": ["Claude"] }
  ]
}
```

**Use case:** Understand platform-specific performance and opportunities.

**Share of Voice by Region**

Analyze visibility score and share of voice across different regions for a specific company:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["visibility_score", "share_of_voice"],
  "dimensions": ["region"],
  "filters": [
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [0.85, 0.32],
      "dimensions": ["North America"]
    },
    {
      "metrics": [0.72, 0.28],
      "dimensions": ["Europe"]
    }
  ]
}
```

**Use case:** Identify regional strengths and weaknesses in visibility score and share of voice.

**Average Position**

![Average Position](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/average-position.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=bcc94198e117996a2a638aad4e273d5e)

**Top Companies by Average Position**

To replicate the Average Position view in the platform, two API calls are needed. The platform shows average position for the top assets ranked by visibility, not all assets sorted by position.

**Step 1:** Get the top assets by visibility score:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["visibility_score"],
  "dimensions": ["asset_name"],
  "order_by": { "visibility_score": "desc" },
  "pagination": { "limit": 50 }
}
```

**Step 2:** Use those asset names to get their average position:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["average_position"],
  "dimensions": ["asset_name"],
  "filters": [
    {
      "field": "asset_name",
      "operator": "in",
      "value": ["Company A", "Company B", "Company C", "Company D", "Company E"]
    }
  ],
  "order_by": { "average_position": "asc" }
}
```

> [!NOTE]
>
> `"asc"` ordering: lower position is better, so #1 rank has the lowest value. The `"in"` filter uses the asset names returned from Step 1.

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [1.6], "dimensions": ["Company A"] },
    { "metrics": [2.0], "dimensions": ["Company B"] },
    { "metrics": [2.2], "dimensions": ["Company C"] },
    { "metrics": [3.2], "dimensions": ["Company D"] },
    { "metrics": [3.5], "dimensions": ["Company E"] }
  ]
}
```

**Use case:** Identify top-ranked companies and benchmark against competitors.

**Average Position Trends Over Time**

Track a specific company's average position with daily granularity:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "date_interval": "day",
  "metrics": ["average_position"],
  "dimensions": ["date"],
  "filters": [
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [2.5], "dimensions": ["2026-01-01"] },
    { "metrics": [2.3], "dimensions": ["2026-01-02"] },
    { "metrics": [2.1], "dimensions": ["2026-01-03"] },
    { "metrics": [2.0], "dimensions": ["2026-01-04"] },
    { "metrics": [2.2], "dimensions": ["2026-01-05"] },
    { "metrics": [2.4], "dimensions": ["2026-01-06"] },
    { "metrics": [2.2], "dimensions": ["2026-01-07"] }
  ]
}
```

**Use case:** Monitor daily ranking changes and identify trends.

**Ranking By Topic**

![Visibility Ranking by Topic](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/visibility-ranking-by-topic.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=f128d36afa172443c788c669de31a9fb)

**Visibility by Topic and Prompt**

Analyze visibility score and share of voice across different topics and prompt types for a specific company:

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["visibility_score", "share_of_voice"],
  "dimensions": ["prompt", "topic"],
  "filters": [
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [0.45, 0.88],
      "dimensions": ["What is the best CRM software?", "CRM Software"]
    },
    {
      "metrics": [0.38, 0.75],
      "dimensions": ["Top project management tools for teams", "Project Management"]
    },
    {
      "metrics": [0.30, 0.62],
      "dimensions": ["Best enterprise collaboration platforms", "Collaboration"]
    }
  ]
}
```

**Use case:** Analyze visibility score and share of voice across different topics and prompt types for a specific company.

**Mixed AND/OR Tag Filtering**

Combine multiple tag groups to slice a category by intersecting independent dimensions.

Each filter in the `filters` array narrows the result. To express **OR within a group**, list values in a single `in` filter. To express **AND across groups**, send multiple `in` filters on the same field.

```http theme={null}
POST /v1/reports/visibility HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["share_of_voice"],
  "dimensions": ["asset_name"],
  "filters": [
    {
      "field": "tag_id",
      "operator": "in",
      "value": [
        "tag_id_product_a",
        "tag_id_product_b",
        "tag_id_product_c"
      ]
    },
    {
      "field": "tag_id",
      "operator": "in",
      "value": ["tag_id_product_d"]
    }
  ]
}
```

This matches prompts tagged with **at least one of** `product_a`, `product_b`, or `product_c` **and also** tagged `tag_id_product_d`. Equivalent boolean form: `(product_a OR product_b OR product_c) AND tag_id_product_d`.

> [!NOTE]
>
> A single `in` filter listing all four tag IDs would mean "any of these four":
> `(product_a OR product_b OR product_c OR tag_id_product_d)`

## Sentiment Analysis

Analyze sentiment data and emotional responses across companies and topics. See the full [endpoint reference](/api-reference/reports/query-sentiment) for all available metrics, dimensions, and filters.

> [!TIP]
>
> The same filter composition shown in the **Mixed AND/OR Tag Filtering** tab under [Visibility Reports](#visibility-reports) works on this endpoint.

**Top 5 Companies by Positive Sentiment**

![Sentiment Analysis](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/sentiment-analysis.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=fd208d6c433caf243a98c81ec05de1e3)

Identify companies with the most positive sentiment:

```http theme={null}
POST /v1/reports/sentiment HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["positive"],
  "dimensions": ["asset_name"],
  "pagination": { "limit": 5 }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [2413], "dimensions": ["Company A"] },
    { "metrics": [1856], "dimensions": ["Company B"] },
    { "metrics": [1204], "dimensions": ["Company C"] },
    { "metrics": [987], "dimensions": ["Company D"] },
    { "metrics": [714], "dimensions": ["Company E"] }
  ]
}
```

> [!NOTE]
>
> Sentiment metrics are integer counts, not percentages like visibility. `positive` = count of positive sentiment instances for that asset.
>
> Available sentiment metrics: `positive`, `negative`, `occurrences` (total count).

**Use case:** Understand which companies have the best reputation and sentiment.

**Company Sentiment by Theme**

![Sentiment Themes](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/sentiment-themes.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=c0036f42a019b145ab31c3235dbeab83)

Analyze positive and negative sentiment across different themes for a specific company:

```http theme={null}
POST /v1/reports/sentiment HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "metrics": ["positive", "negative"],
  "dimensions": ["theme"],
  "pagination": {
    "limit": 5
  },
  "filters": [
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "metrics": [12, 45],
      "dimensions": ["Customer Support"]
    },
    {
      "metrics": [8, 32],
      "dimensions": ["Pricing"]
    },
    {
      "metrics": [6, 28],
      "dimensions": ["Ease of Use"]
    },
    {
      "metrics": [4, 50],
      "dimensions": ["Product Quality"]
    },
    {
      "metrics": [2, 18],
      "dimensions": ["Integration"]
    }
  ]
}
```

**Use case:** Identify themes driving positive or negative sentiment for targeted improvements.

## Citation Reports

Track mentions, references, and citation patterns across domains and pages. See the full [endpoint reference](/api-reference/reports/query-citations) for all available metrics, dimensions, and filters.

> [!TIP]
>
> The same filter composition shown in the **Mixed AND/OR Tag Filtering** tab under [Visibility Reports](#visibility-reports) works on this endpoint.

**Citation Share**

![Citation Share](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/citation-share.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=2989691c8660880e1a8470e05f1bb9e6)

**Citation Share Over Time**

Track citation share by domain with daily granularity:

```http theme={null}
POST /v1/reports/citations HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-04-08",
  "end_date": "2026-04-15",
  "metrics": ["citation_share"],
  "dimensions": ["date", "hostname"],
  "date_interval": "day",
  "order_by": { "date": "asc" }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.092], "dimensions": ["2026-04-08 00:00:00", "example-a.com"] },
    { "metrics": [0.028], "dimensions": ["2026-04-08 00:00:00", "example-b.com"] },
    { "metrics": [0.021], "dimensions": ["2026-04-08 00:00:00", "example-c.com"] },
    { "metrics": [0.018], "dimensions": ["2026-04-08 00:00:00", "example-d.com"] },
    { "metrics": [0.015], "dimensions": ["2026-04-08 00:00:00", "example-e.com"] },
    { "metrics": [0.105], "dimensions": ["2026-04-09 00:00:00", "example-a.com"] },
    { "metrics": [0.030], "dimensions": ["2026-04-09 00:00:00", "example-b.com"] },
    { "metrics": [0.024], "dimensions": ["2026-04-09 00:00:00", "example-c.com"] },
    { "metrics": [0.019], "dimensions": ["2026-04-09 00:00:00", "example-d.com"] },
    { "metrics": [0.013], "dimensions": ["2026-04-09 00:00:00", "example-e.com"] },
    { "metrics": [0.088], "dimensions": ["2026-04-10 00:00:00", "example-a.com"] },
    { "metrics": [0.026], "dimensions": ["2026-04-10 00:00:00", "example-b.com"] }
  ]
}
```

**Use case:** Monitor how citation share trends over time across competing domains.

**Top 5 Domains by Citation Share**

Get the top 5 domains ranked by citation share with total citation count:

```http theme={null}
POST /v1/reports/citations HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-04-08",
  "end_date": "2026-04-15",
  "metrics": ["citation_share", "count"],
  "dimensions": ["hostname"],
  "order_by": { "citation_share": "desc" },
  "pagination": { "limit": 5 }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.087, 358], "dimensions": ["example-a.com"] },
    { "metrics": [0.025, 102], "dimensions": ["example-b.com"] },
    { "metrics": [0.023, 95], "dimensions": ["example-c.com"] },
    { "metrics": [0.019, 78], "dimensions": ["example-d.com"] },
    { "metrics": [0.015, 61], "dimensions": ["example-e.com"] }
  ]
}
```

**Use case:** Identify the most cited domains in your category and their relative share.

**Citation Categories & Domains**

![Citation Categories & Domains](https://mintcdn.com/profound-37face47/ho6CxSyI1KFXwLNh/images/rest-api/aei-examples/citation-categories-domains.png?fit=max\&auto=format\&n=ho6CxSyI1KFXwLNh\&q=85\&s=d48e23a78ab7411c7e5e1c10552be081)

**Citation Categories**

Get citation share and count broken down by citation category:

```http theme={null}
POST /v1/reports/citations HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-04-08",
  "end_date": "2026-04-15",
  "metrics": ["citation_share", "count"],
  "dimensions": ["citation_category"],
  "order_by": { "citation_share": "desc" }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.136, 542], "dimensions": ["Earned Media"] },
    { "metrics": [0.115, 458], "dimensions": ["Social"] },
    { "metrics": [0.079, 315], "dimensions": ["Owned"] },
    { "metrics": [0.053, 211], "dimensions": ["Competition"] },
    { "metrics": [0.034, 136], "dimensions": ["Blog"] },
    { "metrics": [0.028, 112], "dimensions": ["Partner"] },
    { "metrics": [0.024, 96], "dimensions": ["Distribution"] },
    { "metrics": [0.020, 80], "dimensions": ["Documentation"] },
    { "metrics": [0.016, 64], "dimensions": ["Press"] },
    { "metrics": [0.012, 48], "dimensions": ["Video"] },
    { "metrics": [0.010, 40], "dimensions": ["other"] }
  ]
}
```

**Use case:** Understand which citation categories drive the most mentions in your category.

**Top 10 Citation Domains**

Get the top 10 cited domains by citation share and count, grouped by citation category:

```http theme={null}
POST /v1/reports/citations HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-04-08",
  "end_date": "2026-04-15",
  "metrics": ["citation_share", "count"],
  "dimensions": ["citation_category", "hostname"],
  "order_by": { "citation_share": "desc" },
  "pagination": { "limit": 10 }
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    { "metrics": [0.038, 152], "dimensions": ["Earned Media", "example-a.com"] },
    { "metrics": [0.035, 140], "dimensions": ["Social", "example-f.com"] },
    { "metrics": [0.028, 112], "dimensions": ["Earned Media", "example-b.com"] },
    { "metrics": [0.025, 100], "dimensions": ["Earned Media", "example-c.com"] },
    { "metrics": [0.024, 96], "dimensions": ["Social", "example-g.com"] },
    { "metrics": [0.021, 84], "dimensions": ["other", "example-d.com"] },
    { "metrics": [0.018, 72], "dimensions": ["Social", "example-h.com"] },
    { "metrics": [0.015, 60], "dimensions": ["Earned Media", "example-e.com"] }
  ]
}
```

**Use case:** Identify the top 10 most cited domains and which citation category they belong to.

## Raw Data Access

Access unprocessed prompt and answer data for custom analysis. See the full [endpoint reference](/api-reference/prompts/get-answers) for all available parameters and filters.

**Visibility Answers**

Get all visibility answers for a date range:

```http theme={null}
POST /v1/prompts/answers HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "filters": [
    {
      "field": "prompt_type",
      "operator": "is",
      "value": "visibility"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "created_at": "2026-01-01T08:15:00Z",
      "prompt": "What are the best project management tools?",
      "mentions": ["Company A", "Company B", "Company C"],
      "prompt_type": "open-ended",
      "response": "There are several leading project management tools available today. Company A offers comprehensive project tracking with built-in collaboration features...",
      "citations": ["https://www.example.com/reviews", "https://www.example.com/best-tools"],
      "themes": ["feature comparison", "pricing"],
      "topic": "Project Management",
      "region": "North America",
      "model": "ChatGPT",
      "model_id": "3f8a1b2c-1234-5678-abcd-ef0123456789",
      "asset": "Company A"
    },
    ...
  ]
}
```

**Use case:** Build custom dashboards and perform specialized analysis with raw answer data.

**Sentiment Answers**

Get sentiment prompt answers filtered by a specific asset:

```http theme={null}
POST /v1/prompts/answers HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "category_id": "your_category_id",
  "start_date": "2026-01-01",
  "end_date": "2026-01-07",
  "filters": [
    {
      "field": "prompt_type",
      "operator": "is",
      "value": "sentiment"
    },
    {
      "field": "asset_name",
      "operator": "is",
      "value": "your_company_name"
    }
  ]
}
```

```json theme={null}
{
  "info": {},  // removed for brevity
  "data": [
    {
      "created_at": "2026-01-02T11:30:00Z",
      "prompt": "What do users think about Company A's analytics features?",
      "mentions": ["Company A"],
      "prompt_type": "brand-direct",
      "response": "Users generally have positive feedback about Company A's analytics capabilities. The platform is praised for its comprehensive dashboard and intuitive reporting features...",
      "citations": ["https://www.example.com/review/company-a"],
      "themes": ["analytics quality", "ease of use"],
      "topic": "Product Feedback",
      "region": "Europe",
      "model": "Perplexity",
      "model_id": "c4d5e6f7-3456-7890-cdef-012345678901",
      "asset": "Company A"
    },
    ...
  ]
}
```

**Use case:** Analyze raw sentiment responses for a specific brand across models and regions.


# Example requests
Source: https://docs.tryprofound.com/rest-api/examples/example

Sample API requests and responses to help you get started

> **[Answer engine insights](/rest-api/examples/answer-engine-insights)**
>
> Sample requests for the answer engine insights endpoints.

> **[Agent Analytics](/rest-api/examples/agent-analytics)**
>
> Sample requests for the Agent Analytics endpoints.

> **[Workflows](/rest-api/examples/workflows)**
>
> Sample requests for the Workflows endpoints.

> **[Knowledge Bases](/rest-api/examples/knowledge-bases)**
>
> Sample requests for the Knowledge Bases endpoints.

> **[OpenAI ads](/rest-api/examples/openai-ads)**
>
> Sample requests for the OpenAI ads endpoints.


# Knowledge Bases
Source: https://docs.tryprofound.com/rest-api/examples/knowledge-bases

Sync documents from an external system into a Profound Knowledge Base and search it through the API

This guide shows how to sync documents from an external system such as Guru into a Profound knowledge base. All examples require authentication with an API key.

> [!NOTE]
>
> Knowledge bases must be created in the Profound app. The API cannot create a knowledge base. Replace `your_api_key`, `your_knowledge_base_id`, and `your_organization_id` with your actual values.

## Syncing documents walkthrough

Use this flow to mirror documents and folders from an external system:

1. Use [**List Knowledge Bases**](/api-reference/knowledge-bases/list-knowledge-bases) to find the knowledge base ID.
2. Use [**Add Folder**](/api-reference/knowledge-bases/add-folder) to mirror source folders when needed.
3. Use [**Add Document**](/api-reference/knowledge-bases/add-document) to upload each source document.
4. Use [**Update Document**](/api-reference/knowledge-bases/update-document) for idempotent re-syncs.
5. Use [**Delete Document**](/api-reference/knowledge-bases/delete-document) or [**Delete Folder**](/api-reference/knowledge-bases/delete-folder) when content is removed from the source system.

### 1. List Knowledge Bases

First, get the ID of the knowledge base you want to sync. The `id` returned here is required as the `knowledge_base_id` path parameter for every other Knowledge Base request.

```http theme={null}
GET /v1/knowledge-bases HTTP/1.1
X-API-Key: your_api_key
```

```json theme={null}
{
  "data": [
    {
      "id": "your_knowledge_base_id",
      "name": "Product Documentation",
      "slug": "product-documentation",
      "description": "Product guides and reference material",
      "created_at": "2026-04-24T19:03:15.459951Z"
    }
  ],
  "pagination": {}
}
```

If your API key can access multiple organizations, pass `organization_id` on the request:

```http theme={null}
GET /v1/knowledge-bases?organization_id=your_organization_id HTTP/1.1
X-API-Key: your_api_key
```

Use the same query parameter on subsequent requests when the API key spans multiple organizations.

### 2. Add a Folder

Folders are empty when created. A document's folder must already exist, so create the source hierarchy before uploading documents, and use folder paths such as `Engineering/API` to mirror the hierarchy in Guru. Create parent folders before their nested folders. Creating a folder that already exists returns `409 Conflict`.

```http theme={null}
POST /v1/knowledge-bases/your_knowledge_base_id/folders HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "path": "Engineering"
}
```

```json theme={null}
{
  "message": "Folder added.",
  "path": "Engineering"
}
```

### 3. Add a Document

Add a document as JSON by sending its text in the `text` field. `folder` is optional; when provided, it is the folder path containing the document.

```http theme={null}
POST /v1/knowledge-bases/your_knowledge_base_id/documents HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "name": "authentication.md",
  "folder": "Engineering",
  "text": "# Authentication\n\nUse an API key to authenticate requests."
}
```

You can also upload a file as multipart form data. This is useful when your source system provides document files directly:

```bash theme={null}
curl -X POST "https://api.tryprofound.com/v1/knowledge-bases/your_knowledge_base_id/documents" \
  -H "X-API-Key: your_api_key" \
  -F "name=authentication.md" \
  -F "folder=Engineering" \
  -F "file=@./authentication.md"
```

Both forms return the document name, path, and folder:

```json theme={null}
{
  "message": "Document added.",
  "name": "authentication.md",
  "path": "Engineering/authentication.md",
  "folder": "Engineering"
}
```

Adding a document does not overwrite an existing document: if the document path already exists, the request returns `409 Conflict`. Use update instead when the path already exists.

### 4. Update a Document

Update overwrites an existing document at the target path. The JSON request has the same `name`, `text`, and optional `folder` fields as add:

```http theme={null}
PUT /v1/knowledge-bases/your_knowledge_base_id/documents HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "name": "authentication.md",
  "folder": "Engineering",
  "text": "# Authentication\n\nUse the latest API key instructions."
}
```

Multipart uploads use the same `name` and optional `folder` fields, with `file` replacing `text`:

```bash theme={null}
curl -X PUT "https://api.tryprofound.com/v1/knowledge-bases/your_knowledge_base_id/documents" \
  -H "X-API-Key: your_api_key" \
  -F "name=authentication.md" \
  -F "folder=Engineering" \
  -F "file=@./authentication.md"
```

Update targets an existing document path in an existing folder; a missing folder returns `404 Not Found`.

### 5. Delete a Document

Delete a document by sending its name in a JSON request body. Include `folder` in the name when deleting a nested document:

```http theme={null}
DELETE /v1/knowledge-bases/your_knowledge_base_id/documents HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "name": "Engineering/authentication.md"
}
```

### 6. Delete a Folder

Delete an empty folder by setting `recursive` to `false` (the default):

```http theme={null}
DELETE /v1/knowledge-bases/your_knowledge_base_id/folders HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "path": "Engineering",
  "recursive": false
}
```

If the folder is not empty, the request returns `409 Conflict` and nothing is deleted. To delete the folder and all of its contents, set `recursive` to `true`:

```json theme={null}
{
  "path": "Engineering",
  "recursive": true
}
```

### 7. Search a Knowledge Base

Search with a required `query` and `top_k` between 1 and 100. Set `return_full_page` to `true` to request full page content instead of snippets.

```http theme={null}
POST /v1/knowledge-bases/your_knowledge_base_id/search HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "query": "How do I authenticate?",
  "top_k": 5,
  "return_full_page": false,
  "filters": {
    "tags": ["api"],
    "folders": ["Engineering"]
  }
}
```

`filters.tags` matches documents with any of the supplied tags. `filters.folders` limits the search to folder paths and currently accepts exactly one folder.

Each result includes an `id`, relevance `score`, `metadata`, and matched `content`:

```json theme={null}
{
  "data": [
    {
      "id": "Engineering/authentication.md",
      "score": 0.94,
      "metadata": {
        "folder_path": "Engineering",
        "source_filename": "authentication.md"
      },
      "content": "Use an API key to authenticate requests."
    }
  ],
  "pagination": {}
}
```

> [!TIP]
>
> For an external-system sync, loop over the source documents, use their folder paths to mirror the source hierarchy, use `PUT` for idempotent re-syncs, and use `DELETE` for documents removed from the source.


# OpenAI ads
Source: https://docs.tryprofound.com/rest-api/examples/openai-ads

Sample requests for OpenAI Ads Partner API endpoints

This guide provides examples of common API requests. All examples require authentication via API key.

> [!NOTE]
>
> Replace `<API_KEY>` with your actual Profound API key value. Learn how to generate it in [Authentication](/rest-api/authentication).

## Examples

### Get campaign insights for a date range

```bash theme={null}
curl -G "https://api.tryprofound.com/v1/ads/openai-ads/ad-account/insights" \
  -H "X-API-Key: your_api_key" \
  -d "aggregation_level=campaign" \
  --data-urlencode 'time_ranges[]={"type":"date_range","since":"2026-07-01","until":"2026-07-18"}'
```

### Get daily campaign insights

```bash theme={null}
curl -G "https://api.tryprofound.com/v1/ads/openai-ads/ad-account/insights" \
  -H "X-API-Key: your_api_key" \
  -d "aggregation_level=campaign" \
  -d "time_granularity=daily" \
  --data-urlencode 'time_ranges[]={"type":"date_range","since":"2026-07-01","until":"2026-07-18"}'
```


# Workflows
Source: https://docs.tryprofound.com/rest-api/examples/workflows

Sample requests for the Workflow endpoints

This guide provides practical examples of common Workflow API requests. All examples require authentication via API key.

> [!NOTE]
>
> The endpoints are named agents, but they list and run [Workflows](/agents/workflows/about-workflows), the node-based automations from the Workflow Builder. Agents built with AI Marketer (Aim) aren't available through the API yet.

> [!NOTE]
>
> Replace `your_api_key`, `your_agent_id`, and `your_run_id` with your actual values. Use the list endpoint first to discover the Workflows available to your organization.

## Starter Template walkthrough

Suppose you created a Workflow using **Profound's Starter Template**. This Workflow has one required text input labeled **Any Text (This is an example input)**.

This is the typical flow:

1. Use [**List Agents**](/api-reference/agents/list-agents) to find the Workflow ID.
2. Use [**Get an Agent**](/api-reference/agents/get-an-agent) to inspect the input schema.
3. Use [**Run an Agent**](/api-reference/agents/run-an-agent) with an `inputs` object keyed by the schema's property names.
4. Use [**Get an Agent Run**](/api-reference/agents/get-an-agent-run) to check whether the run succeeded or failed.

### 1. List Agents

First, identify the Workflow you want to run:

```http theme={null}
GET /v1/agents?statuses=published&limit=100 HTTP/1.1
X-API-Key: your_api_key
```

```json theme={null}
{
  "data": [
    {
      "id": "your_agent_id",
      "organization_id": "your_organization_id",
      "name": "Starter Template",
      "status": "published",
      "created_at": "2026-04-24T19:03:15.459951Z",
      "description": "This template is a simple starter agent that helps you get familiar with how the tool works. Try it out, customize it, and then build your own!"
    }
  ]
}
```

### 2. Get an Agent

Next, retrieve the Workflow and inspect its schema. This response gives you the Workflow's input schema and output schema as JSON Schema. You use `schema.input` to determine which fields are required and which variable IDs to send in the `inputs` object when you run the Workflow. You can also use `schema.output` to understand the shape of the result you'll get back from the run.

```http theme={null}
GET /v1/agents/your_agent_id?version=published HTTP/1.1
X-API-Key: your_api_key
```

```json theme={null}
{
  "id": "your_agent_id",
  "organization_id": "your_organization_id",
  "name": "Starter Template",
  "status": "published",
  "created_at": "2026-04-24T19:03:15.459951Z",
  "description": "This template is a simple starter agent that helps you get familiar with how the tool works. Try it out, customize it, and then build your own!",
  "schema": {
    "input": {
      "type": "object",
      "properties": {
        "your_input_variable_id": {
          "type": "string",
          "title": "Any Text (This is an example input)"
        }
      },
      "additionalProperties": false,
      "description": "Provide inputs as an object keyed by variable ID.",
      "required": ["your_input_variable_id"]
    },
    "output": {
      "type": "object",
      "properties": {
        "your_output_variable_id": {
          "type": "string",
          "title": "Example LLM Output"
        }
      },
      "additionalProperties": false,
      "description": "Agent outputs are returned as an object keyed by variable ID."
    }
  }
}
```

> [!TIP]
>
> The `inputs` object must use the schema property key as the field name. In this case, the required input key is `your_input_variable_id`, not the human-readable label.

### 3. Run an Agent

Now start a run using that input key:

```http theme={null}
POST /v1/agents/your_agent_id/runs HTTP/1.1
Content-Type: application/json
X-API-Key: your_api_key

{
  "inputs": {
    "your_input_variable_id": "<your_input_value>"
  }
}
```

```json theme={null}
{
  "id": "your_run_id",
  "agent_id": "your_agent_id",
  "status": "queued",
  "started_at": "2026-04-24T19:08:34.495737Z"
}
```

> [!NOTE]
>
> This endpoint returns `202 Accepted`. A successful submission only means the run was accepted; you still need to fetch the run to see whether it ultimately succeeded or failed.

### 4. Get an Agent Run

Finally, fetch the run result:

```http theme={null}
GET /v1/agents/your_agent_id/runs/your_run_id HTTP/1.1
X-API-Key: your_api_key
```

```json theme={null}
{
  "id": "your_run_id",
  "agent_id": "your_agent_id",
  "status": "succeeded",
  "started_at": "2026-04-24T20:53:08.241633Z",
  "finished_at": "2026-04-24T20:53:09.314671Z",
  "outputs": {
    "your_output_variable_id": "<your_output_value>"
  }
}
```


# Releases
Source: https://docs.tryprofound.com/rest-api/integrations/tableau/releases

Release notes and download links for Profound Tableau connector (WDC 3.0)

**v1.2.0**

2026-09-14

* Sentiment report temporarily unavailable while it moves to our latest reporting API
* Clearer error messages when a request fails
* Security and dependency updates

[Download](https://static.tryprofound.com/integrations/tableau/profound-connector-1.2.0.taco)

**v1.1.0**

2026-04-24

* Fixes for large responses

[Download](https://static.tryprofound.com/integrations/tableau/profound-connector-1.1.0.taco)

**v1.0.0**

2025-08-13

* Initial release of Tableau (WDC 3.0) for data import

[Download](https://static.tryprofound.com/integrations/tableau/profound-connector-1.0.0.taco)


# Usage
Source: https://docs.tryprofound.com/rest-api/integrations/tableau/usage

Import data into Tableau using the Profound Tableau connector (WDC 3.0)

## System Requirements

> **Minimum required version**
>
> * Tableau Desktop 2025.2 or higher

### Compatible Operating Systems

* Windows 10 or higher
* macOS 12 (Monterey) or higher

## Download

The connector is available on our [releases page](/rest-api/integrations/tableau/releases). Download the latest version of `profound-connector.taco` from there.

## Installation

**Windows**

### Windows Installation

1. **Locate the Tableau connectors directory**

   Open Windows Explorer and navigate to:

   ```
   C:\Users\[YourUsername]\Documents\My Tableau Repository\Connectors
   ```

   > [!NOTE]
   >
   > If the `Connectors` folder doesn't exist, create it manually

2. **Copy the .taco file**
   * Copy the downloaded `profound-connector.taco` file
   * Paste it into the Connectors folder
   * **Important**: Do not unzip the .taco file

3. **Restart Tableau Desktop**
   * Completely close Tableau Desktop if it's open
   * Reopen the application

4. **Verify the installation**
   * The connector will appear as "AEO Connector by Profound" in the connectors list

**macOS**

### macOS Installation

1. **Locate the Tableau connectors directory**

   * Open Finder
   * Press `Cmd + Shift + G`
   * Enter the following path:

   ```
   ~/Documents/My Tableau Repository/Connectors
   ```

   > [!NOTE]
   >
   > If the `Connectors` folder doesn't exist, create it

2. **Copy the .taco file**
   * Drag the `profound-connector.taco` file to the Connectors folder
   * **Important**: The file must keep the .taco extension

3. **Restart Tableau Desktop**
   * Completely close Tableau Desktop if it's open
   * Reopen the application

4. **Verify the installation**
   * The connector will appear as "AEO Connector by Profound" in the connectors list

## First Connection

**1. Open Tableau Desktop**

Launch Tableau Desktop 2025.2 or higher

**2. Select the Connector**

1. On the start page, click **"More..."** in the left panel under "To a Server"
2. Look for **"AEO Connector by Profound"** in the list
3. Click to select it

   ![More Connectors](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-3/more_connectors.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=c46b742b1e73b211630e0f136e042974)

**3. Authenticate**

The connector UI will open with an API Key field:

* Enter your **API Key** generated from the platform
* Click **"Access"** to proceed

  ![Enter Key](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-3/enter_key.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=4c9c7a7791c64214855b594f591c0ff4)

**4. Select Data Source**

Once authenticated, you can choose between:

* **Raw Answer**: Access raw data with custom filtering options
* **Report Types**: Select from available pre-configured aggregated reports

Select your preferred data source and apply any filters as needed

![Select Data Source](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-3/select_data.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=c66bd0cb54dcc2ad7fe2bf9ca821b7a6)

**5. Import to Tableau**

Click **"Import to Tableau"** to load the selected data into Tableau

![Imported Table](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-3/imported_table.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=148b4e98607f599581ac501c93ce5b86)

## Troubleshooting

### The connector doesn't appear in Tableau

> **Check Tableau version**
>
> * Go to **Help → About Tableau**
> * Confirm you have version 2025.2 or higher

> **Verify file location**
>
> **Windows:**
>
> * Confirm the file is in `C:\Users\[YourUsername]\Documents\My Tableau Repository\Connectors`
>
> **macOS:**
>
> * Confirm the file is in `~/Documents/My Tableau Repository/Connectors`

> **Fully restart Tableau**
>
> Make sure to close all Tableau Desktop windows before reopening

### Connection Errors

> **Invalid API Key**
>
> * Verify you've copied the complete API Key without spaces
> * Check that your API Key is active
> * Generate a new API Key if necessary

> **No data appears**
>
> * Verify the selected endpoint contains data for your filters
> * Check date ranges aren’t too restrictive
> * Confirm API key has access to the requested data

## Updating the Connector

When a new version is available:

1. Download the new version of the .taco file from the releases page
2. Delete the old file from the Connectors folder
3. Copy the new .taco file
4. Restart Tableau Desktop

## Frequently Asked Questions

> **Where do I generate my API Key?**
>
> [Here you can refer to the API Key generation documentation](/rest-api/authentication)

> **How do I know which version of the connector I have?**
>
> The version is included in the .taco filename you downloaded.


# WDC 2.0
Source: https://docs.tryprofound.com/rest-api/integrations/tableau/wdc-2

Import data into Tableau using Web Data Connector 2.0

> [!WARNING]
>
> WDC 2.0 is deprecated. Use the [Profound Tableau connector (WDC 3.0)](/rest-api/integrations/tableau/usage) instead.

> **[Access Tool](https://tableau.tryprofound.com/)**
>
> [https://tableau.tryprofound.com/](https://tableau.tryprofound.com/)

## Requirements

> **API Access**
>
> Valid API key with permissions to access desired endpoints

> **Internet Connection**
>
> Stable connection to access the export tool

## How to Use

**1. Open Tableau Desktop**

Launch Tableau Desktop (version 10.4 or higher required)

**2. Connect via WDC**

Select **More** on the left side of the screen and search for **Web Data Connector**:

![More Connectors](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-2/more_connectors.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=99dc16da2e321bdaf3bb3a24841134b8)

**3. Connect via WDC**

Enter the tool URL: `https://tableau.tryprofound.com/`

![Add Connector](https://mintcdn.com/profound-37face47/9rAfUZyB5pu4Tp2f/images/rest-api/tableau/wdc-2/add_connector.png?fit=max\&auto=format\&n=9rAfUZyB5pu4Tp2f\&q=85\&s=4b1c697e3f09b9f23f7d8ee428a13ac9)

**4. Authorize Connection**

\[!TIP] Looking for your API key? [Click here](/rest-api/authentication#getting-your-api-key)
Enter your API key when prompted. Select **Access** when ready:

![Enter Key](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-2/enter_key.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=f6f22be20abd5747c24610b57c6a7c91)

**5. Select Data Source**

Select the desired data and configure any additional filters:

![Select Data Source](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-2/select_data.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=704f0fc0635f53d5a3849459584bb7b3)

**6. Import to Tableau**

Click **Import to Tableau** to import data directly into your Tableau workspace

![Waiting to Populate](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-2/waiting_to_populate.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=f7dcc82b16d2b694bc0a09db71c5e279)

> [!TIP]
>
> It may take a few minutes to populate the data into Tableau.

![Populated Data](https://mintcdn.com/profound-37face47/Gnsk1t-l2OsDCIqT/images/rest-api/tableau/wdc-2/populated_data.png?fit=max\&auto=format\&n=Gnsk1t-l2OsDCIqT\&q=85\&s=9226ba0852d05253ae102846fcd5cbb4)

## Available Data Sources

> [!TIP]
>
> The tool supports both processed reports and raw data endpoints

* **Report Endpoints**: Pre-processed analytics and summary data
* **Raw Data**: Unprocessed data directly from database

## Troubleshooting

> **Invalid API Key**
>
> > [!WARNING]
> >
> > Ensure your API key has the correct permissions for the selected endpoint
>
> **Solutions:**
>
> * Verify API key is correct and active

> **No Data Available**
>
> **Solutions:**
>
> * Verify the selected endpoint contains data for your filters
> * Check date ranges aren't too restrictive
> * Confirm API key has access to the requested data

> **Export Timeout**
>
> **Solutions:**
>
> * Try reducing the data range
> * Use more specific filters to limit data volume
> * Contact support if large exports are needed regularly

> **Tableau Connection Issues**
>
> > [!NOTE]
> >
> > Tableau integration uses WDC 2.0, which works with most Tableau Desktop versions
>
> **Solutions:**
>
> * Ensure Tableau Desktop 10.4 or higher
> * Check firewall settings allow WDC connections
> * Verify the tool URL is correctly entered in Tableau

## Technical Notes

> [!WARNING]
>
> Tableau integration uses Web Data Connector 2.0, which is a legacy but stable technology supported by Tableau

* **Data Types**: Automatic detection and formatting for dates, numbers, and text
* **File Size**: Large exports may take additional time to process
* **Refresh**: Tableau connections can be refreshed to get updated data

## Need Help?

If you encounter issues, contact our [support team](mailto:support@tryprofound.com) for assistance.


# Introduction
Source: https://docs.tryprofound.com/rest-api/introduction

An overview of the REST API and its capabilities

> [!NOTE]
>
> **Available on**: the Enterprise plan

## Overview

Our API allows developers to interact with the data displayed on our platform programmatically. All API responses are returned in JSON format, making it easy to integrate with modern applications and tools.

### Base URL

All API requests should be made to:

```
https://api.tryprofound.com
```

**Key capabilities:**

* Automate data retrieval and processing tasks
* Create integrations with third-party applications
* Generate custom reports and analytics
* Access processed reports and raw per-execution prompt-answer data

## API Features

### Report Generation

Create comprehensive reports with processed data from our prompt system. Compare performance metrics across different companies, time periods, or evaluation criteria to gain insights into your data.

### Raw Per-Execution Data

Access raw per-execution prompt-answer rows through
`POST /v2/prompts/answers`, including model responses, mentions, citations,
and search queries.

### User-scoped access

Each API key is a personal access token that belongs to the user who created it. Requests made with the key are attributed to that user and reach data only in the organizations the user belongs to. For details, see [Key ownership](/rest-api/authentication#key-ownership).

## Getting Started

To begin using the API, you'll need an API key from your user profile on our platform. This key authenticates your requests as you and grants access to data in the organizations you belong to.

For authentication details and request examples, see:

* [Authentication](/rest-api/authentication) - How to authenticate your requests
* [Date Ranges & Timezones](/rest-api/date-ranges) - Understanding date parameters and timezone handling
* [Example Requests](/rest-api/examples/example) - Sample API calls and responses
  * [Answer Engine Insights](/rest-api/examples/answer-engine-insights)
  * [Agent Analytics](/rest-api/examples/agent-analytics)
  * [Workflows](/rest-api/examples/workflows)

## API Fundamentals

### Data Organization

Each organization has access to specific categories and datasets. Use the `/v1/org` endpoints to:

* Discover available data categories
* Obtain identifiers for filtering
* Understand your data scope and permissions

### Required Parameters

Answer Engine Insights reports require a **category identifier** and
**date range parameters** (`start_date` and `end_date`). Traffic reports
for bots and referrals are domain-scoped instead. See [Date Ranges &
Timezones](/rest-api/date-ranges) for proper formatting.

### API Versioning

The API currently has both `v1` and `v2` endpoints:

* `v1` endpoints remain active, including the existing V1 report endpoints
* `v2` now spans Answer Engine Insights reports under
  `/v2/reports/visibility`, `/v2/reports/citations`,
  `/v2/reports/sentiment`, `/v2/reports/query-fanouts`,
  `/v2/reports/factcheck`, and `/v2/reports/factcheck/claims`, as well as
  `/v2/prompts/answers`. Each has a `/stream` SSE variant.

Versioned paths ensure compatibility as the API evolves.

## Rate Limits

**Default limits:** 600 requests per hour per API key.

### Rate Limit Headers

Every API response includes these headers:

* `X-RateLimit-Limit`: Your total request limit
* `X-RateLimit-Remaining`: Remaining requests in current period
* `X-RateLimit-Reset`: When the limit resets (Unix timestamp)

### Handling Rate Limits

When you exceed the limit, you'll receive a `429 Too Many Requests` error. The response includes a `Retry-After` header indicating how long to wait before retrying.

Need higher limits? Contact our support team to discuss your requirements.

## Response Format

All API responses use JSON format with consistent structure for errors and successful responses.


# Keyword Intent Shares
Source: https://docs.tryprofound.com/rest-api/prompt-volumes/query-intents-v2

POST /v2/prompt-volumes/intents/on-the-fly
Intent shares for one keyword across the requested cohort weeks.

Shares are fractions from 0 to 1. Cohorts with two or fewer matching users
are omitted for privacy. Doesn't use the daily keyword allowance.

The mix of user intents behind conversations that match one keyword,
averaged across the weeks in your date range. Takes the same `keyword`,
`matching_type`, date range, `regions`, and `platforms` as
[Keyword Volume](/rest-api/prompt-volumes/query-volume-v2).

* **Rows:** one per intent, with `main_classification`, an optional `sub_category_classification`, and `pct_conversation`.
* **`pct_conversation`** is a fraction from 0 to 1 of the classified matching conversations. Multiply by 100 for a percentage.
* **Privacy:** weeks with two or fewer matching users are left out.

> [!NOTE]
>
> This endpoint doesn't use your daily keyword allowance. It shares a burst
> rate limit with [Keyword Volume](/rest-api/prompt-volumes/query-volume-v2).

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/prompt-volumes/intents/on-the-fly \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "keyword": "ai search optimization",
    "matching_type": "phrase_match",
    "start_date": "2026-08-01",
    "end_date": "2026-08-31"
  }'
```


# Keyword Volume
Source: https://docs.tryprofound.com/rest-api/prompt-volumes/query-volume-v2

POST /v2/prompt-volumes/volume/on-the-fly
Weekly and monthly volume projections for one keyword.

Each organization can look up 1,000 distinct keywords per UTC day; over
the cap returns 429 with Retry-After. Slices with two or fewer users are
omitted for privacy.

Weekly and monthly volume projections for one keyword, looked up on demand.
Pass a `keyword`, a `matching_type`, and a date range; narrow with `regions`
and `platforms`.

* **`matching_type`:** `exact_match` or `phrase_match`.
* **Rows:** one per country, platform, and period. The response has both weekly and monthly rows; `frequency` tells them apart (`Week - Monday Start` or `Month`).
* **Privacy:** slices with two or fewer users are left out of the response.
* **`organization_id`:** required only when your API key has access to more than one organization. It picks the organization whose daily allowance the lookup uses.

> [!NOTE]
>
> **Daily keyword allowance.** Each organization can look up 1,000 distinct
> keywords per UTC day. Repeating a keyword you already looked up that day
> is free. A new keyword over the cap returns `429` with `X-KeywordQuota-*`
> and `Retry-After` headers. A lookup that returns no rows still counts
> toward the allowance.

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/prompt-volumes/volume/on-the-fly \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "keyword": "ai search optimization",
    "matching_type": "phrase_match",
    "start_date": "2026-08-01",
    "end_date": "2026-08-31"
  }'
```


# Get answers
Source: https://docs.tryprofound.com/rest-api/reports/query-answers-v2

POST /v2/prompts/answers

The raw per-execution rows behind every report: one row per model response,
with its mentions, citations, and search queries. No aggregation.

* **No `group_by` / `sort`:** rows are newest-first by date.
* **`include`** picks which fields to return (`response`, `mentions`, `citations`, `search_queries`, `sentiment_claims`, …); omit it for every field except `citation_details`, which must be requested explicitly.
* `mentions` is a flat, deduped list in the order entities appear in the response.
* `sentiment_claims` lists the claims extracted from the response, each with `asset`, `theme`, `claim`, and `sentiment` (`positive` or `negative`).
* **Filters:** prompt-level fields plus `analysis_type` (prompt-level: `visibility`·`sentiment`·`factcheck`), and top-level `domain`/`page` leaves: `is` one value or `in` a list (exact cited-URL match).

> [!NOTE]
>
> New to the v2 reports? See [Filtering & concepts](/rest-api/reports/reports-v2-overview) for the shared request shape, filter tree, grouping, and pagination.

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/prompts/answers/stream` takes the **same request body** and
> returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per row.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per row, same shape as data[] above
> data: {"date": "2026-06-19", "model": {"name": "ChatGPT"}, "prompt": "<prompt text>", "mentions": ["Profound", "<string>"], "citations": ["https://example.com/article"]}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/prompts/answers \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-18",
    "end_date": "2026-06-19",
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 27690,
    "count": 10,
    "next_cursor": "Z0FBQUFBQnFQcjNsbDZZaWlzZjR2cXJt...",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "include": [
      "run_id", "date", "model", "topic", "topic_id", "persona", "region",
      "tags", "prompt", "prompt_id", "response", "mentions", "citations",
      "search_queries", "analysis_types", "sentiment_claims"
    ],
    "start_date": "2026-06-18",
    "end_date": "2026-06-19",
    "filter": null
  },
  "data": [
    {
      "run_id": "<uuid>",
      "date": "2026-06-19",
      "model": { "id": "<uuid>", "name": "ChatGPT" },
      "topic": "<string>",
      "topic_id": "<uuid>",
      "region": "United States",
      "persona": null,
      "tags": [],
      "prompt": "<prompt text>",
      "prompt_id": "<uuid>",
      "response": "<full model response text>",
      "mentions": ["Profound", "<string>"],
      "citations": ["https://example.com/article"],
      "search_queries": ["<search query>"],
      "analysis_types": ["visibility"],
      "sentiment_claims": [
        { "asset": "Profound", "theme": "<string>", "claim": "<string>", "sentiment": "positive" }
      ]
    }
    // ...9 more rows (count: 10, total_results: 27690); paginate with next_cursor
  ]
}
```


# Citations
Source: https://docs.tryprofound.com/rest-api/reports/query-citations-v2

POST /v2/reports/citations

Which domains and pages AI answers cite, ranked most-cited first. Group by
`page` for URL-level rows; use `scope: "owned"` to see only domains you own.

* **Metrics:** `count` (raw citations), `citation_share` (per-model average share), `rank`, `first_cited_at` (pages only).
* **`group_by`:** `page`, `date`, `model`, `topic`, `region`, `persona`, `prompt`.
* **No `sort`:** rows are always ranked most-cited first.
* **Citation-layer filters:** `domain` (subdomain-aware), `page`, `analysis_type` (`visibility`·`sentiment`·`factcheck`·`all`), `citation_category` (`owned`·`competition`·`social`·`earned_media`·`earned_institutions`·`pr_wire`·`other`·custom), `citation_tag` (your custom tags: list them with [Get Citation Tags](/api-reference/organization/get-category-citation-tags)).

> [!NOTE]
>
> `citation_category` and `citation_tag` are both top-level `and` leaves
> accepting `is` / `in`; values in one `in` are OR'd, so
> `{"field": "citation_tag", "op": "in", "value": ["Editorial", "Docs"]}`
> matches URLs carrying either tag.

> [!NOTE]
>
> `count` and `citation_share` measure different things: `citation_share` is
> averaged per AI model, so it won't sort in lockstep with raw `count`.

> [!NOTE]
>
> New to the v2 reports? See [Filtering & concepts](/rest-api/reports/reports-v2-overview) for the shared request shape, filter tree, grouping, and pagination.

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/citations/stream` takes the **same request body** and
> returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per row.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per row, same shape as data[] above
> data: {"domain": "reddit.com", "rank": 1, "count": 13940, "citation_share": 0.040}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/citations \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15"
  }'
```

```bash cURL (filter by citation tag) theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/citations \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": {
      "field": "citation_tag",
      "op": "in",
      "value": ["Editorial", "Docs"]
    }
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 11662,
    "count": 10,
    "next_cursor": "Z0FBQUFBQnFQcWdrbDRiSHpGaktVNm5N...",
    "scope": "all",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "metrics": ["count", "citation_share", "rank"],
    "analysis_types": ["visibility"],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": null
  },
  "data": [
    { "domain": "reddit.com", "rank": 1, "count": 13940, "citation_share": 0.040 },
    { "domain": "youtube.com", "rank": 2, "count": 16287, "citation_share": 0.037 },
    { "domain": "tryprofound.com", "rank": 3, "count": 10414, "citation_share": 0.029 }
    // ...7 more rows in this page (count: 10, total_results: 11662); paginate with next_cursor
  ]
}
```


# FactCheck claims
Source: https://docs.tryprofound.com/rest-api/reports/query-factcheck-claims-v2

POST /v2/reports/factcheck/claims

The inaccurate claims behind your accuracy score: the false statements AI
answers made about your category, from the platform's FactCheck tab. Returns a
flat, paginated list, or `group_by` one dimension to section them. A claim is
`{ cluster_id, claim, occurrence }` plus whatever `include` adds.

* **`occurrence`:** the % of scoped responses that carry the claim.
* **`include`:** any of `theme`, `reasoning`, `models`, `evidence`, `citation_sources`.
  * `models[]` — which models make the claim: `{ id, name, occurrence }`.
  * `evidence[]` — refuting knowledge-base snippets: `{ id, kb_path, kb_snippet, source_updated_at }`.
  * `citation_sources[]` — the cited pages driving the claim: `{ href, hostname, citation_category, domain_category, snippet, citation_share }`. `citation_share` is a percentage, model-balanced, matching the Citations tab.
* **`group_by`:** 0 or 1 of `model`, `region`, `persona`, `prompt`, `topic`, `tag`, `theme` (no `citation` or `date`: those are scores dimensions). Empty → one flat paginated list; a dimension → one section per value: `{ <dim>: { id, name }, accuracy, accurate, inaccurate, total_claims, claims: […] }`, each carrying **all** its claims.
* **`filter`:** the same constrained tree as [scores](/rest-api/reports/query-factcheck-v2), a top-level `and` of single-field leaves over `model`/`topic`/`region`/`persona`/`prompt`/`tag` (only `topic` negatable).

> [!NOTE]
>
> A few guardrails return `422` (with a message): `citation_sources` isn't available with `group_by` (its share is scope-sensitive) — request it on the flat list; `group_by: ["tag"]` can't be combined with a `tag` filter (tags are multi-valued, so co-tag sections would escape the filter); and sectioning by `model` over an unusually large scope (>\~2000 inaccurate clusters) — narrow the range or `filter`.

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/factcheck/claims/stream` takes the **same request body**
> and returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per claim.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap.
>
> The stream is the **flat list only**: `group_by` and the per-claim
> `evidence`/`citation_sources` lookups return `422`; use the non-stream
> endpoint for sectioned or enriched claims.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per claim, same shape as data[] above
> data: {"cluster_id": "7a5…", "claim": "Profound covers more than 10 AI engines.", "occurrence": 6.6}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/factcheck/claims \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "include": ["reasoning", "citation_sources"]
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 41,
    "count": 25,
    "next_cursor": "eyJ…",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "group_by": [],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": null,
    "include": ["reasoning", "citation_sources"]
  },
  "data": [
    {
      "cluster_id": "7a576…",
      "claim": "Profound covers more than 10 AI engines.",
      "occurrence": 6.6,
      "reasoning": "Conflates GEO tooling with the engines Profound actually tracks.",
      "citation_sources": [
        {
          "href": "tryprofound.com/blog/best-geo-tools",
          "hostname": "tryprofound.com",
          "citation_category": "owned",
          "domain_category": "owned",
          "snippet": "Profound tracks ten answer engines…",
          "citation_share": 14.2
        }
      ]
    }
  ]
}
```


# FactCheck
Source: https://docs.tryprofound.com/rest-api/reports/query-factcheck-v2

POST /v2/reports/factcheck

Accuracy scores for your category's fact-checked claims: the numbers behind
the platform's FactCheck (Accuracy) tab. `group_by` picks the slice; `accuracy`
is a ratio `0`–`1` (`accurate / (accurate + inaccurate)`). Per-category, so there
is no `asset` or `scope`.

* **Metrics (every row):** `accuracy` (`0`–`1`), `accurate`, `inaccurate`.
* **`group_by`:** 0–2 of `date`, `model`, `region`, `persona`, `prompt`, `topic`, `tag`, `theme` — **or one `citation`** (citation can't be combined with another dimension). Empty → one headline score; `["date"]` → the daily accuracy series; two dims (for example, `["model", "date"]`) → one row per combination, in a single query.
* **Row shape follows `group_by`:** a value dimension → `{ <dim>: { id, name }, … }`; `["citation"]` → `{ citation: { url, citation_category }, … }` (a citation is a URL, not an id/name); `["date"]` → `{ date, … }`.
* **`filter`:** scopes which responses are counted: a top-level `and` of single-field leaves over `model`, `topic`, `region`, `persona`, `prompt`, `tag`. One leaf per field (use `in` for OR within a field); only `topic` may be negated. This is a **narrower** grammar than the other v2 reports (no `or`/`not` nesting).

> [!NOTE]
>
> New to the v2 reports? See [Filtering & concepts](/rest-api/reports/reports-v2-overview) for the shared `{ info, data }` shape, grouping, and pagination. FactCheck uses that envelope but has no `scope`/`assets`/`metrics` params and a constrained `filter` (above). For the inaccurate claims themselves, see [FactCheck Claims](/rest-api/reports/query-factcheck-claims-v2).

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/factcheck/stream` takes the **same request body** and
> returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per row.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per row, same shape as data[] above
> data: {"model": {"id": "a1c9…", "name": "ChatGPT"}, "accuracy": 0.986, "accurate": 1980, "inaccurate": 28}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/factcheck \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "group_by": ["model"]
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 9,
    "count": 9,
    "next_cursor": null,
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "group_by": ["model"],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": null
  },
  "data": [
    { "model": { "id": "a1c9…", "name": "ChatGPT" }, "accuracy": 0.986, "accurate": 1980, "inaccurate": 28 },
    { "model": { "id": "b7f2…", "name": "Microsoft Copilot" }, "accuracy": 0.831, "accurate": 1360, "inaccurate": 276 }
  ]
}
```


# Query Fanouts
Source: https://docs.tryprofound.com/rest-api/reports/query-fanouts-v2

POST /v2/reports/query-fanouts

The search queries an AI generates behind the scenes when answering a prompt.
**Coverage is partial:** metrics cover only runs that emitted fanout data (see
`info.coverage_note`).

* **Metrics:** `total_fanouts` (query count), `fanouts_per_execution` (avg queries per execution), `query_variations` (distinct fanout queries), `share`.
* **`group_by`:** `prompt` (default), `query`, `model`, `region`, `date`.
* `query` requires `prompt` in `group_by`; `share` requires `query`; `query_variations` can't be combined with `query`.
* **`analysis_type`** filter defaults to `visibility`.

> [!NOTE]
>
> New to the v2 reports? See [Filtering & concepts](/rest-api/reports/reports-v2-overview) for the shared request shape, filter tree, grouping, and pagination.

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/query-fanouts/stream` takes the **same request body** and
> returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per row.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per row, same shape as data[] above
> data: {"prompt": "AI search optimization startups...", "rank": 1, "total_fanouts": 190, "fanouts_per_execution": 3.0, "query_variations": 143}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/query-fanouts \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15"
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 322,
    "count": 10,
    "next_cursor": "Z0FBQUFBQnFQcVJ3R1RlcGtXc0g5eTZT...",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "metrics": ["total_fanouts", "fanouts_per_execution", "query_variations"],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "coverage_note": "Fanout metrics cover only runs that emitted query-fanout data.",
    "filter": null
  },
  "data": [
    { "prompt": "AI search optimization startups...", "rank": 1, "total_fanouts": 190, "fanouts_per_execution": 3.0, "query_variations": 143 },
    { "prompt": "Best AI visibility products...", "rank": 2, "total_fanouts": 168, "fanouts_per_execution": 2.7, "query_variations": 143 }
    // ...8 more rows in this page (count: 10, total_results: 322); paginate with next_cursor
  ]
}
```


# Sentiment
Source: https://docs.tryprofound.com/rest-api/reports/query-sentiment-v2

POST /v2/reports/sentiment

Per-brand sentiment as percentages from either AI responses or the pages those
responses cite. `asset` is **required** because sentiment is calculated for one
brand at a time.

* **`source: "response"`** (default): sentiment in AI responses. Metrics are
  `positive_sentiment`, `negative_sentiment`, and opt-in `occurrence`.
* **`source: "citation"`**: one row per cited page, with
  `positive_sentiment`, `negative_sentiment`, and model-balanced
  `citation_share`. It defaults to sorting by `citation_share`; explicitly set
  `sort.field` to `positive_sentiment` to sort by page sentiment instead.
* **Comparison dates:** add `comparison_start_date` and `comparison_end_date`.
  Response rows include previous sentiment metrics. Citation rows include the
  page's previous `citation_share`.

Response sentiment supports up to two `group_by` dimensions from `topic`,
`region`, `model`, `prompt`, `persona`, `tag`, `theme`, `claim`, `run`, and
`competitor`, plus `date`. Citation sentiment is page-only and rejects
`group_by`; use filters to narrow the page population.

Both sources accept prompt-layer filters such as `model`, `topic`, `region`,
`persona`, `prompt`, and `tag`, plus top-level `theme` / `claim` filters. Citation
sentiment additionally accepts these top-level filters:

* `citation_category`: `is` or `in`. Values can be built-in categories
  (`owned`, `competition`, `social`, `earned_media`, `earned_institutions`,
  `pr_wire`, `other`) or a custom category value.
* `page`: `contains_case_insensitive` with one non-empty value. This searches
  the normalized page URL.

`include_cited_websites: true` applies only to response sentiment grouped by
`theme` and/or `claim`.

> [!NOTE]
>
> Citation sentiment describes the cited page. A model filter changes which
> models' citations contribute to page eligibility and citation share; it does
> not turn page sentiment into a model-specific sentiment score.

> [!NOTE]
>
> New to the v2 reports? See [Filtering & concepts](/rest-api/reports/reports-v2-overview) for the shared request shape, filter tree, grouping, and pagination.

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/sentiment/stream` supports both response and citation
> sentiment and returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per row.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap. With `source: "citation"`, each `result` is the same
> page-level shape returned in `data` by the paginated endpoint.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per row, same shape as data[] above
> data: {"theme": {"id": "a14e0c2d", "name": "Analytics"}, "rank": 1, "positive_sentiment": 95.4, "negative_sentiment": 4.6}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/sentiment \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "asset": "Profound",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15"
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 1,
    "count": 1,
    "next_cursor": null,
    "asset": "Profound",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "metrics": ["positive_sentiment", "negative_sentiment"],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": null
  },
  "data": [
    {
      "positive_sentiment": 79.1,
      "negative_sentiment": 20.9
    }
  ]
}
```

## Citation-page example

This request returns owned pages containing `tryprofound.com`, ordered by their
current citation share. Combine citation-layer leaves with other filters under a
top-level `and`.

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/sentiment \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "asset": "Profound",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "source": "citation",
    "metrics": ["positive_sentiment", "negative_sentiment", "citation_share"],
    "sort": {"field": "citation_share", "dir": "desc"},
    "filter": {
      "and": [
        {"field": "citation_category", "op": "is", "value": "owned"},
        {"field": "page", "op": "contains_case_insensitive", "value": "tryprofound.com"}
      ]
    },
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 1,
    "count": 1,
    "next_cursor": null,
    "asset": "Profound",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "..."],
    "metrics": ["positive_sentiment", "negative_sentiment", "citation_share"],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": {
      "and": [
        {"field": "citation_category", "op": "is", "value": "owned"},
        {"field": "page", "op": "contains_case_insensitive", "value": "tryprofound.com"}
      ]
    }
  },
  "data": [
    {
      "page": "tryprofound.com/",
      "rank": 1,
      "positive_sentiment": 100.0,
      "negative_sentiment": 0.0,
      "citation_share": 1.42
    }
  ]
}
```


# Brands
Source: https://docs.tryprofound.com/rest-api/reports/query-shopping-brands-v2

POST /v2/reports/shopping/brands

This report shows brand visibility in ChatGPT shopping results. It gives how
often each brand appears, its average position, and its rank. Select the brands
with `assets`. Divide the results with `group_by`. Each metric is a named field
on the row.

* **Metrics:** `visibility_score` (the share of shopping results that include the brand); `average_position` (the average rank when the brand appears; a **lower** number is **better**); `visibility_rank`.
* **`group_by`:** `date`, `topic`, `region`, or `prompt`.
* **Brand selection:** Select the brands with the `assets` parameter. Give one name or a list. `scope: "all"` ranks all brands. `scope: "owned"` (the default) returns only your brands. An `assets` selection returns one page and ignores `limit`.
* **Filter fields:** `topic`, `region`, `persona`, `prompt`, `tag`.

> [!NOTE]
>
> Shopping data comes from ChatGPT only. No shopping report has a `model`
> group\_by or a `model` filter.

> [!NOTE]
>
> The v2 reports share one request shape. For the filter tree, grouping, and
> pagination, see [Filtering & concepts](/rest-api/reports/reports-v2-overview).

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/shopping/brands/stream` uses the **same request body**. It
> returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event for each row.
> The stream rejects `limit` and `cursor`. It returns all rows by default. To
> set a maximum, use `max_results` (maximum 50000). Use the stream to get a
> large result in one request.

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/shopping/brands \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "group_by": ["date"],
    "metrics": ["visibility_score", "average_position", "visibility_rank"],
    "scope": "owned",
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 1,
    "count": 1,
    "next_cursor": null,
    "scope": "owned",
    "assets": null,
    "models": ["ChatGPT"],
    "metrics": ["visibility_score", "average_position", "visibility_rank"],
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": null
  },
  "data": [
    {
      "asset": { "name": "Example Brand", "owned": true },
      "date": "2026-06-01",
      "visibility_score": 0.42,
      "average_position": 3.2,
      "visibility_rank": 1
    }
  ]
}
```


# Merchants
Source: https://docs.tryprofound.com/rest-api/reports/query-shopping-merchants-v2

POST /v2/reports/shopping/merchants

This report gives merchant data for ChatGPT shopping results. `group_by`
selects one of three views. `merchant_share` is the merchant's share of offers.
`merchant_visibility` is the merchant's share of the runs it appears in. These
are two different measures.

### Views

* **Distribution** (omit `group_by`) — one row for each merchant. Metrics: `merchant_share`, `merchant_share_rank`, `merchant_visibility`, `merchant_visibility_rank`. To get a time series, add `date`.
* **Brand share** (`group_by: ["brand"]`) — one row for each merchant and brand. Metrics: `merchant_share`, `brand_share` (the brand's share in the merchant), `visibility_rank`.
* **Top products** (`group_by: ["product"]`) — one row for each merchant and product. Metrics: `merchant_visibility`, `product_visibility`, `product_rank`.

**Filter fields:** `topic`, `region`, `persona`, `prompt`, `tag`. The
distribution view also accepts a `brand` filter.

### Rules

* Use the metrics that are valid for the view. Other metrics return `422`.
* Group by `brand` **or** `product`. Do not use both.
* `date` is available in the distribution view only. `group_by: ["brand", "date"]` or `["product", "date"]` returns `422`.
* The `brand` filter is available in the distribution view only.

> [!NOTE]
>
> In the distribution view, `merchant_visibility` and `merchant_visibility_rank`
> are run-appearance metrics. The report gets them from a second query and adds
> them to each merchant. Request them only when you need them.

> [!NOTE]
>
> Shopping data comes from ChatGPT only. No shopping report has a `model`
> group\_by or a `model` filter.

> [!NOTE]
>
> The v2 reports share one request shape. For the filter tree, grouping, and
> pagination, see [Filtering & concepts](/rest-api/reports/reports-v2-overview).

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/shopping/merchants/stream` uses the **same request body**.
> It returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event for each row.
> The stream rejects `limit` and `cursor`. It returns all rows by default. To
> set a maximum, use `max_results` (maximum 50000).

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/shopping/merchants \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "metrics": ["merchant_share", "merchant_share_rank", "merchant_visibility", "merchant_visibility_rank"],
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 24,
    "count": 10,
    "next_cursor": "eyJvZmZzZXQiOjEwfQ==",
    "view": "distribution",
    "models": ["ChatGPT"],
    "metrics": ["merchant_share", "merchant_share_rank", "merchant_visibility", "merchant_visibility_rank"],
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": null
  },
  "data": [
    {
      "merchant_name": "Example Retailer",
      "merchant_share": 0.25,
      "merchant_share_rank": 1,
      "merchant_visibility": 0.40,
      "merchant_visibility_rank": 1
    }
  ]
}
```


# Products
Source: https://docs.tryprofound.com/rest-api/reports/query-shopping-products-v2

POST /v2/reports/shopping/products

This report shows product visibility in ChatGPT shopping results. It can also
give the merchant offers for each product. The report returns each product as
`{ name, brand }` with its metrics as named fields.

* **Metrics:** `visibility_score`; `average_position` (a **lower** number is **better**); `visibility_rank`; the position-frequency metrics `position1_percentage`, `position2_percentage`, `position3_percentage`, `position_above3_percentage`; `product_rating`; `product_num_reviews`.
* **Position-frequency metrics:** Each one is the share of the product's appearances at that slot (`position1` = top; `position_above3` = below rank 3). The values are **0–1 fractions**, not 0–100. The four values sum to about 1.
* **`group_by`:** `date`, `topic`, or `prompt`.
* **Filter fields:** `topic`, `region`, `persona`, `prompt`, `tag`, `brand`, `merchant`.

### Merchant offers

To attach the merchant offers to each product, set `include_merchants: true`.
The report then adds `merchants` (a list of `{ name, price }`), `product_url`,
and `product_image_urls`. This mode does not accept `group_by` or
`target_product`. The position-frequency metrics are not available in this mode.

### Competitor mode

To return one product and its top competitors, set `target_product` to a
product name. To set the number of competitors, use `competitor_limit` (the
default is 5). Competitor mode is available in the item view only. Do not use it
with `include_merchants`.

> [!NOTE]
>
> Shopping data comes from ChatGPT only. No shopping report has a `model`
> group\_by or a `model` filter.

> [!NOTE]
>
> The v2 reports share one request shape. For the filter tree, grouping, and
> pagination, see [Filtering & concepts](/rest-api/reports/reports-v2-overview).

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/shopping/products/stream` uses the **same request body**.
> It returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event for each row.
> The stream rejects `limit` and `cursor`. It returns all rows by default. To
> set a maximum, use `max_results` (maximum 50000).

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/shopping/products \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "include_merchants": true,
    "metrics": ["visibility_score", "product_rating", "product_num_reviews"],
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 128,
    "count": 10,
    "next_cursor": "eyJvZmZzZXQiOjEwfQ==",
    "models": ["ChatGPT"],
    "include_merchants": true,
    "metrics": ["visibility_score", "product_rating", "product_num_reviews"],
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": null
  },
  "data": [
    {
      "product": { "name": "Example Running Shoe", "brand": "Example Brand" },
      "merchants": [
        { "name": "Example Retailer", "price": "$120.00" },
        { "name": "Example Brand", "price": "$110.00" }
      ],
      "product_url": "https://www.example.com/p/example-running-shoe",
      "product_image_urls": ["https://images.example.com/products/example-running-shoe.jpg"],
      "product_rating": 4.5,
      "product_num_reviews": 1200,
      "visibility_score": 0.30
    }
  ]
}
```


# Trigger rate
Source: https://docs.tryprofound.com/rest-api/reports/query-shopping-trigger-rate-v2

POST /v2/reports/shopping/trigger-rate

This report shows how often ChatGPT returns shopping results for your prompts.
It gives the total runs, the shopping-triggered runs, and the trigger rate.

* **Metrics:** `total_runs`, `shopping_triggered_runs`, `trigger_rate_percentage`. `trigger_rate_percentage` is `shopping_triggered_runs / total_runs`. It is a **0–1 fraction**, not 0–100. To get a percent, multiply the value by 100.
* **`group_by`:** `date`, `topic`, `region`, `persona`, `prompt`.
* **Filter fields:** `topic`, `region`, `persona`, `prompt`, `tag`.

To get the rate for each prompt, add `prompt` to `group_by`. To get the rate for
each topic, add `topic`. To get a time series, add `date` and set `interval`.

> [!NOTE]
>
> Shopping data comes from ChatGPT only. No shopping report has a `model`
> group\_by or a `model` filter.

> [!NOTE]
>
> The v2 reports share one request shape. For the filter tree, grouping, and
> pagination, see [Filtering & concepts](/rest-api/reports/reports-v2-overview).

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/shopping/trigger-rate/stream` uses the **same request
> body**. It returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event for each row.
> The stream rejects `limit` and `cursor`. It returns all rows by default. To
> set a maximum, use `max_results` (maximum 50000).

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/shopping/trigger-rate \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "group_by": ["date"],
    "metrics": ["total_runs", "shopping_triggered_runs", "trigger_rate_percentage"],
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 30,
    "count": 10,
    "next_cursor": "eyJvZmZzZXQiOjEwfQ==",
    "models": ["ChatGPT"],
    "metrics": ["total_runs", "shopping_triggered_runs", "trigger_rate_percentage"],
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": null
  },
  "data": [
    {
      "date": "2026-06-01",
      "total_runs": 1240,
      "shopping_triggered_runs": 388,
      "trigger_rate_percentage": 0.313
    }
  ]
}
```


# Visibility
Source: https://docs.tryprofound.com/rest-api/reports/query-visibility-v2

POST /v2/reports/visibility

How often your asset appears in AI answers, plus share of voice and average
position. Select the asset(s) with `assets`, break results down with
`group_by`, and read each metric as a named field on the row.

* **Metrics:** `visibility_score` (share of answers the asset appeared in), `share_of_voice` (share of all mentions), `average_position` (average rank when mentioned; **lower is better**).
* **`group_by`:** `date`, `model`, `topic`, `region`, `prompt`, `persona`.
* **Select assets** with the `assets` param (a name, a list, or `{ op, value }`). `scope: "all"` ranks every asset; `"owned"` returns only yours.
* **`sort`** by any requested metric, for example `{ "field": "visibility_score" }`.

> [!NOTE]
>
> New to the v2 reports? See [Filtering & concepts](/rest-api/reports/reports-v2-overview) for the shared request shape, filter tree, grouping, and pagination.

> **Streaming (SSE) variant (same body, /stream)**
>
> `POST /v2/reports/visibility/stream` takes the **same request body** and
> returns [Server-Sent Events](https://developer.mozilla.org/docs/Web/API/Server-sent_events):
> one `summary` event (the `info` block), then one `result` event per row.
> `limit`/`cursor` are ignored; it returns everything by default. Pass
> `max_results` to cap.
>
> ```text Response (text/event-stream) theme={null}
> event: summary
> data: { ...the info block... }
>
> event: result    ← one per row, same shape as data[] above
> data: {"asset": {"name": "Profound", "owned": true}, "rank": 1, "visibility_score": 0.48}
> ```

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/visibility \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-09",
    "end_date": "2026-06-15"
  }'
```

```json 200 theme={null}
{
  "info": {
    "total_results": 1,
    "count": 1,
    "next_cursor": null,
    "scope": "owned",
    "models": ["ChatGPT", "Google Gemini", "Perplexity", "Claude", "..."],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "asset_filter": null,
    "filter": null
  },
  "data": [
    {
      "asset": { "name": "Profound", "owned": true },
      "rank": 1,
      "visibility_score": 0.48,
      "share_of_voice": 0.077,
      "average_position": 2.5
    }
  ]
}
```


# Channels
Source: https://docs.tryprofound.com/rest-api/reports/query-youtube-channels-v2

POST /v2/reports/social/youtube/channels
Rank the YouTube channels cited in a category, or the video categories they publish in.

Rank the YouTube channels cited in a category, or the video categories they
publish in. Calculated from Profound's citation data, not YouTube analytics.

* **Source types:** `video`, `short`, `channel`, `playlist`, or `other`.
  Omit `source_types` to include `video`, `short`, `channel`, and `playlist`.
  Provide a non-empty list; duplicate values are de-duplicated. Rollups
  cannot include `other`, because those citations have no channel.
* **`group_by`:** `channel` (the default), `video_category`, or `source_type`.
  Supported cross-tabs are `["channel", "video_category"]`,
  `["channel", "source_type"]`, and `["channel", "model"]`. Duplicate
  dimensions are rejected.
* **Time series:** Set `interval` to `day`, `week`, or `month` to return one
  row per entity per period. `date` is the bucket start in ET, so a weekly or
  monthly bucket can start before `start_date`. Omit it for window totals.
* **Rows:** Fields are conditional: `date`, `model`, `source_type`, and
  `video_category` appear only with the matching `interval` or `group_by`.
  With `group_by: ["video_category"]`, the category is returned in `name`, not
  `video_category`. An unresolved video category is returned as `""`, not
  `null`. `rank` is the leading channel's rank in the full ranked set and is
  repeated across that channel's cross-tab rows.
* **Pagination:** `limit` defaults to 10 and accepts 1–50. Pass the opaque
  `info.next_cursor` back as the request `cursor` to get the next page;
  `info` echoes your `cursor` on pages after the first. The response `info`
  echoes `category_id`, the effective `limit`, and `interval`.
* **Totals:** `total_results` counts distinct channels in the window, while
  `count` is the number of rows returned; they use different units.
* **Filters:** Prompt-level fields are `model`, `topic`, `region`, `prompt`,
  `persona`, `tag`, and `analysis_type`. The `channel` filter accepts `is`,
  `in`, `contains`, `not_contains`, `contains_case_insensitive`, and
  `not_contains_case_insensitive`: `is`/`in` select exact handles, while the
  other operators match channel titles or handles.
  `domain` and `page` are rejected. A `channel` leaf cannot share an `or` or
  `not` with a prompt-level leaf; put separate layers in `and` clauses.

> [!WARNING]
>
> For cross-tabs, `limit` counts leading channels, not returned rows. For
> example, `limit: 25` with `group_by: ["channel", "video_category"]` returns
> 25 channels but can return more than 25 rows.

> [!NOTE]
>
> The v2 reports share a request shape and filter tree. See
> [Filtering & concepts](/rest-api/reports/reports-v2-overview) for filter
> operators, grouping, date ranges, pagination, and filter depth.

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/social/youtube/channels \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "group_by": ["channel", "source_type"],
    "source_types": ["video", "short"],
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "total_results": 137,
    "count": 2,
    "next_cursor": "eyJvIjoxMH0",
    "limit": 10,
    "interval": null,
    "group_by": ["channel", "source_type"],
    "models": ["ChatGPT", "Google Gemini"],
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": null,
    "source_types": ["video", "short"]
  },
  "data": [
    {
      "name": "Example Channel",
      "handle": "examplechannel",
      "url": "https://www.youtube.com/@examplechannel",
      "rank": 1,
      "source_type": "video",
      "count": 128,
      "videos": 14,
      "citation_share": 0.37
    },
    {
      "name": "Example Channel",
      "handle": "examplechannel",
      "url": "https://www.youtube.com/@examplechannel",
      "rank": 1,
      "source_type": "short",
      "count": 42,
      "videos": 8,
      "citation_share": 0.12
    }
  ]
}
```


# Videos
Source: https://docs.tryprofound.com/rest-api/reports/query-youtube-videos-v2

POST /v2/reports/social/youtube/videos
Rank cited YouTube videos, for one channel or across all of them.

Rank cited YouTube videos, for one channel or across all of them. Calculated
from Profound's citation data, not YouTube analytics.

* **Source types:** `video`, `short`, `channel`, `playlist`, or `other`.
  `other` is available with `unattributed` or `all`; with the default
  `attributed` mode, requests containing `other` are rejected. Omit
  `source_types` to return `video` and `short` with the default `attributed`
  mode; `unattributed` and `all` widen the default to all five source types.
  Provide a non-empty list; duplicate values are de-duplicated.
* **Attribution:** `attributed` (the default), `unattributed`, or `all`.
  An unattributed row has no channel: `source_type` is `other` for a search or
  feed URL that names no source, and any other type is a source we have no
  channel for.
* **Grouping and time series:** Not supported. `/videos` accepts no `group_by`
  or `interval`; unknown fields are rejected.
* **Pagination:** `limit` defaults to 10 and accepts 1–50. Pass the opaque
  `info.next_cursor` back as the request `cursor` to get the next page;
  `info` echoes your `cursor` on pages after the first. The response `info`
  echoes `category_id`, the effective `limit`, and `attribution`.
* **Filters:** Prompt-level fields are `model`, `topic`, `region`, `prompt`,
  `persona`, `tag`, and `analysis_type`. The `channel` filter accepts `is`,
  `in`, `contains`, `not_contains`, `contains_case_insensitive`, and
  `not_contains_case_insensitive`: `is`/`in` select exact handles, while the
  other operators match channel titles or handles.
  `domain` and `page` are rejected. A `channel` leaf cannot share an `or` or
  `not` with a prompt-level leaf; put separate layers in `and` clauses.

> [!NOTE]
>
> The v2 reports share a request shape and filter tree. See
> [Filtering & concepts](/rest-api/reports/reports-v2-overview) for filter
> operators, date ranges, pagination, and filter depth. Grouping and time
> series guidance does not apply to this endpoint.

```bash cURL theme={null}
curl -X POST https://api.tryprofound.com/v2/reports/social/youtube/videos \
  -H "X-API-Key: <your-api-key>" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": {
      "field": "channel",
      "op": "in",
      "value": ["examplechannel"]
    },
    "attribution": "attributed",
    "limit": 10
  }'
```

```json 200 theme={null}
{
  "info": {
    "category_id": "182bd5e5-6e1a-4fe4-a799-aa6d9a6ab26e",
    "total_results": 1,
    "count": 1,
    "next_cursor": null,
    "limit": 10,
    "attribution": "attributed",
    "models": ["ChatGPT", "Google Gemini"],
    "start_date": "2026-06-01",
    "end_date": "2026-06-30",
    "filter": {
      "field": "channel",
      "op": "in",
      "value": ["examplechannel"]
    },
    "source_types": ["video", "short"]
  },
  "data": [
    {
      "video_id": "dQw4w9WgXcQ",
      "source_type": "video",
      "title": "Example Video",
      "channel_title": "Example Channel",
      "channel_handle": "examplechannel",
      "url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
      "channel_url": "https://www.youtube.com/@examplechannel",
      "rank": 1,
      "published_at": "2026-06-12",
      "duration_seconds": 245,
      "video_category": "Education",
      "count": 73,
      "citation_share": 0.21
    }
  ]
}
```


# Filtering & concepts
Source: https://docs.tryprofound.com/rest-api/reports/reports-v2-overview

How the v2 report endpoints work: the shared request shape, the filter tree, grouping, scope, pagination, and streaming

The v2 report endpoints share **one request and response shape**:
[Visibility](/rest-api/reports/query-visibility-v2),
[Citations](/rest-api/reports/query-citations-v2),
[Sentiment](/rest-api/reports/query-sentiment-v2),
[Query Fanouts](/rest-api/reports/query-fanouts-v2),
[Answers](/rest-api/reports/query-answers-v2), and
[FactCheck](/rest-api/reports/query-factcheck-v2) (accuracy scores +
[claims](/rest-api/reports/query-factcheck-claims-v2)). Learn it once and every
report works the same way. YouTube social reports are also v2 reports, with
channel and video-specific grouping and attribution fields; see
[YouTube Channels](/rest-api/reports/query-youtube-channels-v2) and
[YouTube Videos](/rest-api/reports/query-youtube-videos-v2) for their
endpoint-specific request and response fields.

> [!NOTE]
>
> FactCheck uses the same `{ info, data }` envelope but is **per-category** (no
> `scope`/`assets`/`metrics` params) and takes a **narrower `filter`**: a
> top-level `and` of single-field leaves, only `topic` negatable. See its pages
> for specifics.

> [!NOTE]
>
> All v2 endpoints accept **names or UUIDs** anywhere a filter takes a value.
> Get UUIDs from `GET /v1/org/models`, `/v1/org/regions`, `/v1/org/personas`,
> `/v1/org/assets`, `/v1/org/categories`, and the per-category
> `…/topics`, `…/tags`, `…/prompts` endpoints.

## Common request fields

| Field | Type | Notes |
| - | - | - |
| `category_id` | UUID (required) | The category to query. |
| `start_date` / `end_date` | date (required) | `YYYY-MM-DD`, Eastern Time, **inclusive** on both ends. |
| `scope` | `owned` · `all` | Restrict to your owned assets/domains, or rank everything. Defaults vary per report. |
| `group_by` | string\[] | Break results into rows by dimension (see [Grouping](#grouping)). |
| `metrics` | string\[] | Which metrics to compute. Returned as **named fields on each row**. |
| `interval` | `day` · `week` · `month` | Bucket size when grouping by `date`. Default `day`. |
| `filter` | tree | `and`/`or`/`not`/leaf tree (see [Filters](#filters)). |
| `sort` | `{ field, dir }` | Order rows (where supported). |
| `limit` | `1`–`50` | Rows per page. Default `10`. |
| `cursor` | string | Page token from `info.next_cursor`. |

> [!WARNING]
>
> Unlike the v1 reports (where `end_date` is exclusive), **v2 `end_date` is
> inclusive**. To get June 9–15, send `start_date: "2026-06-09"`,
> `end_date: "2026-06-15"`.

## Grouping

`group_by` turns one aggregate row into one row per value. Each grouped field
is echoed back on the row:

```json theme={null}
// group_by: ["model"]  →  each row carries the model it's for
{ "asset": { "name": "Profound", "owned": true }, "model": { "id": "…", "name": "ChatGPT" }, "visibility_score": 0.52 }
```

* Group by `date` (with `interval`) for a time series.
* Rows carry a `rank` when grouped by a **non-date** dimension.
* Available dimensions differ per report; see each endpoint's reference.

## Metrics

Request the metrics you want; they come back as **named fields** on each row
(no positional arrays, no `info.query` lookup):

```json theme={null}
{ "rank": 1, "visibility_score": 0.48, "share_of_voice": 0.077, "average_position": 2.5 }
```

## Scope and asset selection

* **`scope`:** `owned` limits to assets/domains you own; `all` ranks across everything.
* **Visibility only:** pick the asset(s) with the separate **`assets`** param: a
  name (`is`), a list (`in`, which overrides `scope`), or `{ op, value }`. A
  selection returns all matches and ignores `limit`.
* **Sentiment** requires an **`asset`** (sentiment is per-brand).

## Filters

`filter` is a recursive tree, **max depth 3**, for every report, including
YouTube social reports:

```json theme={null}
{
  "and": [
    { "or":  [ { "field": "model", "op": "is", "value": "ChatGPT" },
               { "field": "model", "op": "is", "value": "Perplexity" } ] },
    { "not": { "field": "region", "op": "is", "value": "United States" } }
  ]
}
```

Node types: `{ "and": [ … ] }`, `{ "or": [ … ] }`, `{ "not": <node> }`, and
leaves `{ "field", "op", "value" }`.

### Operators

| Operator | Meaning |
| - | - |
| `is` / `not_is` | Exact match / negated |
| `in` / `not_in` | Match any value in a list (non-empty) / negated |
| `contains` / `not_contains` | Substring (case-sensitive) |
| `contains_case_insensitive` / `not_contains_case_insensitive` | Substring (case-insensitive) |
| `matches` | Regex (pattern ≥ 3 chars) |
| `exists` | Has any value; only on `tag` / `persona` (wrap in `not` for "none") |

`value` is a single value, or a list for `in` / `not_in`. Names **or** UUIDs;
`contains` / `matches` match on names.

### Two filter layers

Fields fall into two layers. They combine with `and`; **`or`/`not` can't mix
layers** (doing so returns `422`).

**Prompt layer** (full tree, full operator set):
`model`, `topic`, `region`, `persona`, `prompt`, `tag`.

**Entity / citation layer** (top-level `and` leaves only, varies per report):

| Report | Entity-layer fields |
| - | - |
| Visibility | The entity (`asset`) uses the **`assets`** param, *not* `filter`. |
| Citations | `domain` (full ops, subdomain-aware), `page` (full ops), `analysis_type` (`visibility`·`sentiment`·`factcheck`·`all`), `citation_category` (`owned`·`competition`·`social`·`earned_media`·`earned_institutions`·`pr_wire`·`other`·custom), `citation_tag` (your custom tags; `is`/`in`) |
| Sentiment | `theme` / `claim`: `is`/`in`, single value, name or id. With `source: "citation"`: `citation_category` (`is`/`in`) and `page` (`contains_case_insensitive`, one value) |
| Query Fanouts | `analysis_type` (`visibility`·`sentiment`·`factcheck`·`all`), `is`/`in` |
| Answers | `analysis_type` is **prompt-level** (`visibility`·`sentiment`·`factcheck`; `is`/`in`/`not_in`; omit = all). `domain`/`page` are top-level `and` leaves: `is` one value or `in` a list (exact cited-URL match) |
| YouTube social | `channel` is the entity-layer filter. `domain` and `page` do not apply because YouTube rows are already scoped to YouTube sources. |

> [!NOTE]
>
> For citations, filter `domain` in the domains report and `page` when you
> `group_by: ["page"]`; each filters its own report's entity.

> [!NOTE]
>
> Sentiment with `source: "citation"` always returns page rows and does not
> support `group_by`. Use prompt-layer filters to narrow which citations count,
> and use `citation_category` or `page` to narrow the returned pages.

> [!NOTE]
>
> YouTube reports use the `channel` entity filter. It can be combined with
> prompt-level filters using `and`, but not with them under `or` or `not`.
> YouTube `domain` and `page` filters are rejected rather than approximated.

> [!NOTE]
>
> List a category's citation tags with [Get Citation Tags](/api-reference/organization/get-category-citation-tags).
> Put every tag you want in a single `in` leaf (values are OR'd). AND-ing two
> separate `citation_tag` leaves, or passing an empty list, returns `422`.

## Sorting

Where supported, `sort` is `{ "field": "<metric>", "dir": "asc" | "desc" }`.
The field must be a requested, sortable metric (or `date` when grouped by
date). **Citations has no `sort`:** it's always ranked most-cited first.

## Pagination

Responses return `limit` rows plus `info.next_cursor`. Pass that token back as
`cursor` to get the next page; `next_cursor` is `null` on the last page.

## Streaming

Every report endpoint except YouTube social reports has a **`/stream`** variant
(Server-Sent Events): a `summary` event (the `info` block), then one `result`
event per row. `limit`/`cursor` are ignored; it returns everything by default.
Pass `max_results` to cap. Sentiment streaming accepts both `source: "response"`
and `source: "citation"` using the same source-specific row shapes as the
paginated endpoint.

## Response shape

Every report returns `{ info, data }`:

```json theme={null}
{
  "info": {
    "total_results": 8427,
    "count": 10,
    "next_cursor": "…",
    "models": ["ChatGPT", "Google Gemini", "..."],
    "start_date": "2026-06-09",
    "end_date": "2026-06-15",
    "filter": null
  },
  "data": [
    { "rank": 1, "visibility_score": 0.48 }
  ]
}
```

`info` echoes the resolved query (models in scope, the applied filter, dates,
pagination); `data` is the rows, with metrics as named fields and any
`group_by` dimensions attached.


# Response format
Source: https://docs.tryprofound.com/rest-api/response-format

Understanding report response structures and data interpretation

## Overview

Report endpoints (`/v1/reports/*`) use a specialized array-based response format to optimize performance and reduce payload size. This guide explains how to interpret these responses.

> [!NOTE]
>
> Other endpoints like `/v1/org/*` and `/v1/prompts/*` use standard JSON object
> structures that are documented in the OpenAPI reference.

## Report Response Structure

All report endpoints return data in this optimized format:

```json theme={null}
{
  "data": [
    {
      "dimensions": ["example.com", "/some/path", "2023-10-01"],
      "metrics": [10, 0.05]
    },
    {
      "dimensions": ["another-site.com", "/different/path", "2023-10-02"],
      "metrics": [15, 0.08]
    }
  ],
  "info": {
    "query": {
      "date_interval": "day",
      "dimensions": ["hostname", "path", "date"],
      "filters": [
        {
          "field": "hostname",
          "operator": "in",
          "value": ["example.com", "another-site.com"]
        }
      ],
      "metrics": ["count", "share_of_voice"]
    },
    "total_rows": 200
  }
}
```

## Interpreting Report Data

### Array Position Mapping

The key to understanding report responses is that array positions correspond exactly to your request parameters:

#### Dimensions Array

Values map to the order of dimensions in your request:

* `dimensions[0]` → First dimension requested (`hostname`)
* `dimensions[1]` → Second dimension requested (`path`)
* `dimensions[2]` → Third dimension requested (`date`)

#### Metrics Array

Values map to the order of metrics in your request:

* `metrics[0]` → First metric requested (`count`)
* `metrics[1]` → Second metric requested (`share_of_voice`)

#### Example Mapping

For the first result object above:

* **Hostname**: `dimensions[0]` = "example.com"
* **Path**: `dimensions[1]` = "/some/path"
* **Date**: `dimensions[2]` = "2023-10-01"
* **Count**: `metrics[0]` = 10
* **Share of Voice**: `metrics[1]` = 0.05

### Info Object

The `info` object provides metadata about your query:

| Field | Description |
| - | - |
| `query` | Your exact request parameters echoed back |
| `total_rows` | Total number of results available (useful for pagination) |

## Working with Report Responses

### Processing Data

Example of converting the array format to a more traditional object structure:

```javascript JavaScript theme={null}
// Process report response
const processReportData = (response) => {
  const { dimensions: dimNames, metrics: metricNames } = response.info.query;
  
  return response.data.map(row => {
    const result = {};
    
    // Map dimensions
    dimNames.forEach((name, index) => {
      result[name] = row.dimensions[index];
    });
    
    // Map metrics
    metricNames.forEach((name, index) => {
      result[name] = row.metrics[index];
    });
    
    return result;
  });
};

// Usage
const processedData = processReportData(apiResponse);
// Result: [{ hostname: "example.com", path: "/some/path", date: "2023-10-01", count: 10, share_of_voice: 0.05 }]

```

```python Python theme={null}
def process_report_data(response):
    """Convert array-based response to dict format"""
    query_info = response['info']['query']
    dim_names = query_info['dimensions']
    metric_names = query_info['metrics']

    results = []
    for row in response['data']:
        result = {}

        # Map dimensions
        for i, name in enumerate(dim_names):
            result[name] = row['dimensions'][i]

        # Map metrics
        for i, name in enumerate(metric_names):
            result[name] = row['metrics'][i]

        results.append(result)

    return results

# Usage
processed_data = process_report_data(api_response)
# Result: [{"hostname": "example.com", "path": "/some/path", "date": "2023-10-01", "count": 10, "share_of_voice": 0.05}]
```

```php PHP theme={null}
function processReportData($response) {
    $queryInfo = $response['info']['query'];
    $dimNames = $queryInfo['dimensions'];
    $metricNames = $queryInfo['metrics'];

    $results = [];
    foreach ($response['data'] as $row) {
        $result = [];

        // Map dimensions
        foreach ($dimNames as $index => $name) {
            $result[$name] = $row['dimensions'][$index];
        }

        // Map metrics
        foreach ($metricNames as $index => $name) {
            $result[$name] = $row['metrics'][$index];
        }

        $results[] = $result;
    }

    return $results;
}
```

## Best Practices

1. **Always use the `info.query` object** to understand the structure of your response data
2. **Process responses programmatically** using the mapping examples above rather than hardcoding array positions
3. **Handle pagination** using the `total_rows` field to determine if more data is available
4. **Monitor rate limit headers** to avoid hitting API limits
5. **Validate your request parameters** match the expected dimensions and metrics for your use case

## Need Help?

If you encounter unexpected response formats or need clarification on specific report responses, contact our support team for assistance.


# JavaScript SDK
Source: https://docs.tryprofound.com/sdks/javascript-package

Install and use the official JavaScript and TypeScript SDK for the Profound API

The JavaScript SDK is published on npm as [@profoundai/client](https://www.npmjs.com/package/@profoundai/client). It ships with TypeScript types.

## Installation

```bash npm wrap theme={null}
npm install @profoundai/client
```

```bash yarn wrap theme={null}
yarn add @profoundai/client
```

```bash pnpm wrap theme={null}
pnpm add @profoundai/client
```

```bash bun wrap theme={null}
bun add @profoundai/client
```

## Quick start

```typescript Create a client and run a report expandable wrap theme={null}
import Profound from '@profoundai/client';

const client = new Profound({
  apiKey: "your-api-key-here" // or set PROFOUND_API_KEY env var
});

// Get organization categories
const categories = await client.organizations.categories.list();
console.log(categories);

// Generate a report
const report = await client.reports.visibility({
  category_id: categories[0].id,
  start_date: "2024-01-01",
  end_date: "2024-01-31",
  dimensions: ["asset_name"],
  metrics: ["visibility_score"]
});
console.log(report);
```

## Authentication

The SDK reads your API key from either of these sources:

1. **Environment variable** (recommended): set `PROFOUND_API_KEY` in your environment.
2. **Constructor parameter**: pass `apiKey` when initializing the client.

```bash Set the environment variable wrap theme={null}
export PROFOUND_API_KEY="your-api-key-here"
```

For details on obtaining an API key, see [Authentication](/rest-api/authentication).

## Error handling

```typescript Catch API errors expandable wrap theme={null}
import Profound from '@profoundai/client';

const client = new Profound();

try {
  const report = await client.reports.visibility({
    category_id: 'invalid-id',
    start_date: '2024-01-01',
    end_date: '2024-01-31',
    dimensions: ['asset_name'],
    metrics: ['visibility_score']
  });
} catch (error) {
  if (error instanceof Profound.APIError) {
    console.error(`API Error: ${error.status} - ${error.message}`);
  }
}
```

## Rate limiting

The API has a default limit of 600 requests per hour per API key. When you exceed this limit, you receive a `429 Too Many Requests` error. The SDK throws an error that you can catch and handle.

```typescript Handle rate limits expandable wrap theme={null}
import Profound from '@profoundai/client';

try {
  const report = await client.reports.visibility(...);
} catch (error) {
  if (error instanceof Profound.RateLimitError) {
    console.log('Rate limit exceeded. Please wait before retrying.');
  }
}
```

## Async support

The SDK is fully async/await based. All methods return promises.

```typescript Await any method wrap theme={null}
import Profound from '@profoundai/client';

const client = new Profound();

const categories = await client.organizations.categories.list();
```

## Code examples

Every endpoint in the API documentation includes working JavaScript samples. Browse the **Endpoints** section in the REST API & SDKs tab for examples covering:

* Organization management
* Category discovery
* Report generation
* Raw data access

> [!TIP]
>
> SDK interfaces may change as we improve the API. Pin to a specific version in production environments.


# Get started with SDKs
Source: https://docs.tryprofound.com/sdks/overview

Pick the Python or JavaScript SDK and find its setup guide

Profound provides official SDKs for Python and JavaScript/TypeScript, making it easier to integrate our API into your applications.

**Key benefits:**

* Type-safe interfaces
* Automatic authentication handling
* Built-in retry logic and error handling
* Comprehensive code examples for every endpoint
* Async support for modern applications

## Choose your SDK

> **[Python SDK](/sdks/python-package)**

> **[JavaScript SDK](/sdks/javascript-package)**


# Python SDK
Source: https://docs.tryprofound.com/sdks/python-package

Install and use the official Python SDK for the Profound API

The Python SDK is published on PyPI as [profound](https://pypi.org/project/profound/).

## Installation

```bash Install the package wrap theme={null}
pip install profound
```

## Quick start

```python Create a client and run a report expandable wrap theme={null}
from profound import Profound

client = Profound(
    api_key="your-api-key-here"  # or set PROFOUND_API_KEY env var
)

# Get organization categories
categories = client.organizations.categories.list()
print(categories)

# Generate a report
report = client.reports.visibility(
    category_id=categories[0]['id'],
    start_date="2024-01-01",
    end_date="2024-01-31",
    dimensions=["asset_name"],
    metrics=["visibility_score"]
)
print(report)
```

## Authentication

The SDK reads your API key from either of these sources:

1. **Environment variable** (recommended): set `PROFOUND_API_KEY` in your environment.
2. **Constructor parameter**: pass `api_key` when initializing the client.

```bash Set the environment variable wrap theme={null}
export PROFOUND_API_KEY="your-api-key-here"
```

For details on obtaining an API key, see [Authentication](/rest-api/authentication).

## Error handling

```python Catch API errors expandable wrap theme={null}
from profound import Profound, APIError

client = Profound()

try:
    report = client.reports.visibility(
        category_id="invalid-id",
        start_date="2024-01-01",
        end_date="2024-01-31",
        dimensions=["asset_name"],
        metrics=["visibility_score"]
    )
except APIError as e:
    print(f"API Error: {e.status_code} - {e.message}")
```

## Rate limiting

The API has a default limit of 600 requests per hour per API key. When you exceed this limit, you receive a `429 Too Many Requests` error. The SDK raises an error that you can catch and handle.

```python Handle rate limits wrap theme={null}
from profound import RateLimitError

try:
    report = client.reports.visibility(...)
except RateLimitError as e:
    print("Rate limit exceeded. Please wait before retrying.")
```

## Async support

The SDK supports both synchronous and asynchronous usage:

```python Synchronous and asynchronous clients expandable wrap theme={null}
# Synchronous
from profound import Profound
client = Profound()
categories = client.organizations.categories.list()

# Asynchronous
from profound import AsyncProfound
import asyncio

async def main():
    client = AsyncProfound()
    categories = await client.organizations.categories.list()

asyncio.run(main())
```

## Code examples

Every endpoint in the API documentation includes working Python samples. Browse the **Endpoints** section in the REST API & SDKs tab for examples covering:

* Organization management
* Category discovery
* Report generation
* Raw data access

> [!TIP]
>
> SDK interfaces may change as we improve the API. Pin to a specific version in production environments.


# Profound Knowledge Base Fetcher
Source: https://docs.tryprofound.com/bots/knowledge-base-fetcher

What the bot fetches, when it runs, and how site operators can identify or block it

The Profound Knowledge Base Fetcher is an HTTP agent that reads a website's sitemap when you import that site into a [Knowledge Base](/context-manager/manage-knowledge-bases). It also re-downloads PDF files already in a Knowledge Base to check whether they have changed.

## Summary

| | |
| - | - |
| **User-Agent string** | `ProfoundKnowledgeBase/1.0 (+https://docs.tryprofound.com/bots)` |
| **Trigger** | User-initiated sitemap imports, and auto-sync when enabled for an imported folder or page |
| **Schedule** | Only on the auto-sync cadence you choose for an imported folder or page: daily, weekly, or monthly. Off by default |
| **Resources fetched** | Sitemap files and child sitemaps they reference, and PDF files already imported into a Knowledge Base |
| **Network origin** | AWS, no fixed IP allow-list available |

## What it does

When you import a website with the **Sitemap** tab of **Crawl Website**, the fetcher requests the sitemap URL you enter and any child sitemaps listed in a sitemap index. Profound queues the page URLs it finds for import.

The fetcher only reads sitemap files, so it runs when the **Sitemap URL** ends in `.xml` or `.xml.gz`. If you enter a plain domain instead, Profound finds the site's pages another way and this fetcher isn't involved.

You can leave out `https://`, and the fetcher adds it. If the first attempt finds no pages and the address has no `www.`, the fetcher tries again with `www.` added.

With auto-sync on, the fetcher reads the sitemap again on every sync run to pick up new or removed pages. It also downloads each PDF that was imported from the web on that cadence and compares a fingerprint of its bytes with the previous version. Profound processes only the files that changed.

## What it doesn't do

* **Fetch HTML pages**, images, scripts, or stylesheets. Page content is fetched separately
* **Read `robots.txt`**
* **Submit forms**, follow login flows, or access authenticated areas
* **Run JavaScript** or use a headless browser
* **Reach private or internal network targets**

## Request behavior

| | |
| - | - |
| **Per request timeout** | 20 seconds |
| **Redirects** | Followed manually, up to 5 hops |
| **Concurrency** | Sitemap files are fetched one at a time. During auto-sync, up to 10 documents sync at once, so up to 10 PDF downloads can run in parallel |
| **Sitemap limits** | Up to 50 sitemap files and 3 levels of nesting per import, 20 MB per file |
| **PDF limit** | 200 MB per file |
| **Caching** | None. Each import or sync issues fresh requests |

### Headers sent

```txt wrap theme={null}
User-Agent: ProfoundKnowledgeBase/1.0 (+https://docs.tryprofound.com/bots)
Accept-Encoding: gzip, deflate
```

PDF fingerprint requests send `Accept-Encoding: identity` instead.

## How to identify the bot

Use the exact User-Agent header:

```txt wrap theme={null}
User-Agent: ProfoundKnowledgeBase/1.0 (+https://docs.tryprofound.com/bots)
```

> [!TIP]
>
> All Profound bots use User-Agent strings that start with `Profound`, so a prefix match on `Profound` is a good way to identify all Profound traffic, current and future.

## How to block

Block the fetcher using any of the following methods:

* A web application firewall (WAF) or content delivery network (CDN) rule matching User-Agent containing `ProfoundKnowledgeBase`
* A WAF or CDN rule matching User-Agent containing the prefix `Profound`

## Reporting abuse

If you observe behavior that doesn't match this documentation, such as requests for HTML pages, request rates that look like a crawl, or activity from an unrecognized `Profound*` User-Agent, report it to [security@tryprofound.com](mailto:security@tryprofound.com) with example log lines.

Profound treats these reports as security issues and responds to all of them.


# ProfoundBot
Source: https://docs.tryprofound.com/bots/profoundbot



ProfoundBot is a user-initiated HTTP agent. When you supply a URL through the Profound platform or API, it fetches that public web page and returns its content as Markdown and/or HTML.

## Summary

| | |
| - | - |
| **Operator** | Profound ([tryprofound.com](https://tryprofound.com)) |
| **User-Agent** | `ProfoundBot/1.0 (+https://docs.tryprofound.com/bots)` |
| **Trigger** | User-initiated |
| **Schedule** | None |
| **Resources fetched** | The single URL the customer supplies |
| **Network origin** | AWS by default; enrolled domains are fetched from a dedicated egress IP range (see [Network origin and dedicated IPs](#network-origin-and-dedicated-ips) below) |

## What it does

When you supply a public webpage URL to read the content from, ProfoundBot makes a bounded set of HTTP requests to that URL's origin to fetch the page, then returns the content as Markdown, HTML or both.

When a page is rendered with a headless browser, that page's own subresources (scripts, stylesheets, images) load as they would in any browser. ProfoundBot doesn't follow links to other pages.

## What it doesn't do

* **Crawl.** Each customer action fetches exactly one URL. The bot does not recursively follow links or build an index of your site
* **Submit forms**, follow login flows, or access authenticated areas
* **Persist** a long-running index of your site. Fetched content is delivered to you and is not republished
* **Reach private or internal network targets**

## Request behavior

Each user action triggers a small, bounded interaction scoped to the single requested URL.

| | |
| - | - |
| **Redirects** | Followed |
| **Caching** | None: each customer invocation issues fresh requests |
| **Headers** | `User-Agent: ProfoundBot/1.0 (+https://docs.tryprofound.com/bots)` |

## Network origin and dedicated IPs

By default, ProfoundBot egresses from a dynamic IP range. Identify it by its User-Agent.

You can enroll a domain (a per-domain setting, off by default) so that ProfoundBot fetches it from a dedicated egress IP range that Profound owns. This is useful if you prefer to allowlist Profound by IP rather than by User-Agent. When a domain is enrolled, requests egress from this range:

* `134.49.168.0/24`

Domains enrolled before Profound moved to this range keep egressing from the previous dedicated IPs until you move them:

* `54.71.251.60`
* `54.185.59.110`
* `100.22.234.65`

These dedicated IPs are also available in a [machine-readable JSON format](/bots/profoundbot.json).

> [!TIP]
>
> To enroll a domain for dedicated-IP fetching, go to your Profound account **Settings → Web Scrape** and toggle the **Primary Domain Scraping via Static IPs** setting to **On**.

> [!NOTE]
>
> If some of your domains still use the previous IPs, **Settings → Web Scrape** lists them with a **Move to the new IP range** button. Add `134.49.168.0/24` to your allowlist before you move them, because after the move ProfoundBot fetches those domains only from the new range.

## `robots.txt` handling

ProfoundBot is user-initiated and fetches a single page per user action rather than crawling. Support for honoring `Disallow` directives in `robots.txt` is planned for a future release.

To prevent ProfoundBot from accessing your site in the meantime, see [How to block](#how-to-block) section.

## How to identify the bot

Use the exact User-Agent header:

```
User-Agent: ProfoundBot/1.0 (+https://docs.tryprofound.com/bots)
```

> [!TIP]
>
> All Profound bots use User-Agent strings that start with `Profound`, so a prefix match is a good way to identify all Profound traffic, current and future. For enrolled domains, you can also match on the dedicated egress IPs listed in the [Network origin and dedicated IPs](#network-origin-and-dedicated-ips) section.

## How to block

Block ProfoundBot using any of the following methods:

* A WAF or CDN rule matching User-Agent containing `ProfoundBot`
* A WAF or CDN rule matching User-Agent containing the prefix `Profound`
* For enrolled domains, a firewall or WAF rule blocking the dedicated egress range (`134.49.168.0/24`) and, for domains that still use them, the previous dedicated IPs (`54.71.251.60`, `54.185.59.110`, `100.22.234.65`)

## Reporting abuse

If you observe behavior that doesn't match this documentation, such as recursive crawling, request rates inconsistent with user-initiated single-page fetches, or activity from an unrecognized `Profound*` User-Agent, report it to [security@tryprofound.com](mailto:security@tryprofound.com) with example log lines.

Profound treats these reports as security issues and responds to all of them.


# Profound Sheets Sitemap Importer
Source: https://docs.tryprofound.com/bots/sitemap-importer



The Profound Sheets Sitemap Importer is a user-initiated HTTP agent that fetches a website's sitemap for its URLs to be imported into a Profound Sheet. This bot only runs when an authenticated Profound user clicks **+ Create new → Sitemap** inside the Sheets section of the Profound platform and supplies a site URL.

## Summary

| | |
| - | - |
| **User-Agent string** | `ProfoundSheetsSitemapImporter/1.0` |
| **Trigger** | User-initiated |
| **Schedule** | None |
| **Resources fetched** | `sitemap.xml`, sitemap-index variants, `robots.txt`, and any sitemap files those documents reference |
| **Network origin** | AWS, no fixed IP allow-list available |

## What it does

When initiated with a website origin URL, the Sitemap Importer makes a small, bounded set of HTTP GET requests to locate and read a site's sitemap. It tries the following URLs in sequence:

* `https://{origin}/sitemap.xml`
* `https://{origin}/sitemap_index.xml`
* `https://{origin}/sitemap-index.xml`
* `https://{origin}/robots.txt` (parsed only for `Sitemap:` directives, which are added to the candidate list)
* any child sitemaps referenced by a sitemap index

If you enter a URL ending in `.xml`, the importer also tries that URL directly.

The first response that parses as a valid XML sitemap is used. The URLs it contains are returned to you for import. The importer does not store, cache, redistribute, or republish sitemap content.

## What it doesn't do

* **Fetch HTML pages**, images, scripts, stylesheets, or any other assets
* **Submit forms**, follow login flows, or access authenticated areas
* **Run JavaScript** or use a headless browser
* **Crawl recursively** beyond sitemap-index to child-sitemap links
* **Persist** a long-running index of your site

## Request behavior

Each import action triggers a small burst of requests scoped to a single origin.

| | |
| - | - |
| **Per request timeout** | 10 seconds |
| **Redirects** | Followed manually, up to 5 hops |
| **Concurrency** | Sequential (not in parallel) |
| **Caching** | None. Each user invocation issues fresh requests |

### Headers sent

```txt theme={null}
User-Agent: ProfoundSheetsSitemapImporter/1.0
Accept: application/xml,text/xml,application/xhtml+xml,text/plain;q=0.9,*/*;q=0.8
```

### Volume and frequency

Traffic from the Importer is driven entirely by customer activity. There's no continuous crawl. A typical interaction involves a handful of requests, after which the Importer doesn't contact an origin again until another import action targets the same site.

## `robots.txt` handling

The importer reads `robots.txt` to discover sitemap URLs declared via `Sitemap:` directives. It doesn't evaluate `Disallow` rules against sitemap files, because the sitemap protocol treats those files as publicly discoverable.

To prevent the importer from accessing your site, see [How to block](#how-to-block) section.

## How to identify the bot

Use the exact User-Agent header:

```txt theme={null}
User-Agent: ProfoundSheetsSitemapImporter/1.0
```

> [!TIP]
>
> All Profound bots use User-Agent strings that start with `Profound`, so a prefix match on `Profound` is a good way to identify all Profound traffic, current and future.

## How to block

Block the importer using any of the following methods:

* A WAF or CDN rule matching User-Agent equal to `ProfoundSheetsSitemapImporter/1.0`
* A WAF or CDN rule matching User-Agent containing the prefix `Profound`
* Returning a 4xx response to the bot's User-Agent on `/sitemap.xml`, `/sitemap_index.xml`, `/sitemap-index.xml`, and `/robots.txt`

## Reporting abuse

If you observe behavior that doesn't match this documentation, such as requests for resources other than sitemap and `robots.txt` files, request rates that look like a crawl, or activity from an unrecognized `Profound*` User-Agent, report it to [security@tryprofound.com](mailto:security@tryprofound.com) with example log lines.

Profound treats these reports as security issues and responds to all of them.


# Citation Share for your owned domains
Source: https://docs.tryprofound.com/cookbook/citations/citation-share

Compute the headline owned-domain citation share plus the period-over-period delta: what the Profound Citation Share tile shows

The "Citation Share" tile in the Profound app shows the share of all
citations pointing at domains you own, plus the change vs the previous
window of equal length (the `+1.3% / -0.5%` delta badge). Same two-call
pattern as Visibility's period-over-period: run the citations aggregate
once for the current window and once for the prior window, diff the two
shares.

## How this example works

1. **Look up owned domains** via `/categories/{id}/assets`. Keep rows
   flagged `is_owned`, collect their `website` + `alternate_domains`.
2. **Citations call** with `dimensions: ["root_domain"]` returns one row
   per cited domain. Sum `citation_share` across rows whose `root_domain`
   matches one of yours → headline value.
3. **Run the call twice** (current and prior window) and diff the two
   shares for the period-over-period delta.

> [!NOTE]
>
> Most Profound categories scope citations to a `prompt_type` (the toggle
> in the UI). To match the default UI value, send
> `filters=[{"field": "prompt_type", "operator": "is", "value": "visibility"}]`.

```python Python theme={null}
import os
from datetime import date, timedelta
from urllib.parse import urlparse
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
DAYS          = 7
INCLUSIVE_END = date(2026, 5, 11)    # the last day of your current window


def normalize(host_or_url):
    """Strip scheme, www, and trailing slash so domains compare cleanly."""
    p = urlparse(host_or_url if "://" in host_or_url else f"https://{host_or_url}")
    return (p.hostname or "").replace("www.", "").rstrip("/")


def get_owned_domains(category_id):
    """All domains across assets flagged is_owned in the category."""
    out = set()
    for a in client.organizations.categories.assets(category_id):
        if not a.is_owned:
            continue
        if a.website:
            out.add(normalize(a.website))
        for d in a.alternate_domains or []:
            out.add(normalize(d))
    return out


def get_owned_citation_share(category_id, owned_domains, start, end):
    """Sum citation_share across all rows whose root_domain is one of yours."""
    res = client.reports.citations(
        category_id=category_id,
        start_date=start.isoformat(),
        end_date=end.isoformat(),
        metrics=["count", "citation_share"],
        dimensions=["root_domain"],
        filters=[{"field": "prompt_type", "operator": "is", "value": "visibility"}],
    )
    m = res.info.query["metrics"]
    d = res.info.query["dimensions"]
    i_share  = m.index("citation_share")
    i_domain = d.index("root_domain")
    return sum(
        row.metrics[i_share]
        for row in res.data
        if normalize(row.dimensions[i_domain]) in owned_domains
    )


def current_and_prior_windows(inclusive_end, days):
    """Two (start, end_exclusive) pairs of equal length, back-to-back."""
    current = (
        inclusive_end - timedelta(days=days - 1),
        inclusive_end + timedelta(days=1),     # +1 day → exclusive end
    )
    prior = (current[0] - timedelta(days=days), current[0])
    return current, prior


# Helper — translate the category name to the UUID the report API expects.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


# Resolve name → ID, then run both windows.
category_id    = find_category_id(CATEGORY_NAME)
owned          = get_owned_domains(category_id)
current, prior = current_and_prior_windows(INCLUSIVE_END, DAYS)

current_share = get_owned_citation_share(category_id, owned, *current)
prior_share   = get_owned_citation_share(category_id, owned, *prior)
delta_pp      = (current_share - prior_share) * 100

print(f"Citation Share: {current_share:.1%}  ({delta_pp:+.1f} pp vs prev period)")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here
export CATEGORY_ID=your-category-uuid

# 1) Owned domains
curl "https://api.tryprofound.com/v1/org/categories/$CATEGORY_ID/assets" \
  -H "X-API-Key: $PROFOUND_API_KEY"

# 2a) Current window
curl -X POST "https://api.tryprofound.com/v1/reports/citations" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["count", "citation_share"],
    "dimensions": ["root_domain"],
    "filters": [{"field": "prompt_type", "operator": "is", "value": "visibility"}]
  }'

# 2b) Previous window (same body, prior dates)
curl -X POST "https://api.tryprofound.com/v1/reports/citations" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-04-28",
    "end_date": "2026-05-05",
    "metrics": ["count", "citation_share"],
    "dimensions": ["root_domain"],
    "filters": [{"field": "prompt_type", "operator": "is", "value": "visibility"}]
  }'
```


# Citation Share and volume over time
Source: https://docs.tryprofound.com/cookbook/citations/citation-share-over-time

Build the daily owned-domain Citation Share line and the daily total citation volume, both from a single API call

The Citation Share tile shows a daily **share** line (your owned-domain
Citation Share over time, with the previous-period overlay). This recipe
builds that line and also derives the daily total citation **volume**,
both from the **same** API call: group by `["date", "root_domain"]`,
request `count` and `citation_share`, then derive each series from the
response.

> [!WARNING]
>
> **Don't average daily rows to derive a period score**: they're computed
> differently. See [Conventions & gotchas](/cookbook/setup/conventions#dont-average-daily-rows-to-get-a-period-score).

## How this example works

1. **Group by `["date", "root_domain"]`.** Each row is one (day, domain)
   pair with its own `count` and `citation_share`.
2. **Derive both series in one pass over the response**:
   * **Daily volume** = sum of `count` across all rows for each date.
   * **Daily owned-domain share** = sum of `citation_share` across rows
     whose `root_domain` is in your owned set, per date. (Per-AI-model
     averaged, same as the headline tile.)
3. **Repeat for the prior window** to get the overlay line.

```python Python theme={null}
import os
from datetime import date, timedelta
from urllib.parse import urlparse
from collections import defaultdict
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
DAYS          = 7
INCLUSIVE_END = date(2026, 5, 11)    # the last day of your current window


def normalize(host_or_url):
    """Strip scheme, www, and trailing slash so domains compare cleanly."""
    p = urlparse(host_or_url if "://" in host_or_url else f"https://{host_or_url}")
    return (p.hostname or "").replace("www.", "").rstrip("/")


def get_owned_domains(category_id):
    """All domains across assets flagged is_owned in the category."""
    out = set()
    for a in client.organizations.categories.assets(category_id):
        if not a.is_owned:
            continue
        if a.website:
            out.add(normalize(a.website))
        for d in a.alternate_domains or []:
            out.add(normalize(d))
    return out


def get_daily_citations(category_id, owned_domains, start, end):
    """One call → both daily volume and daily owned-domain share.

    Returns [(date, volume, owned_share), ...] sorted by date.
    """
    res = client.reports.citations(
        category_id=category_id,
        start_date=start.isoformat(),
        end_date=end.isoformat(),
        metrics=["count", "citation_share"],
        dimensions=["date", "root_domain"],
        date_interval="day",
        filters=[{"field": "prompt_type", "operator": "is", "value": "visibility"}],
        pagination={"limit": 50000, "offset": 0},
    )
    m = res.info.query["metrics"]
    d = res.info.query["dimensions"]
    i_count  = m.index("count")
    i_share  = m.index("citation_share")
    i_date   = d.index("date")
    i_domain = d.index("root_domain")

    volume_by_day      = defaultdict(int)
    owned_share_by_day = defaultdict(float)
    for row in res.data:
        day    = row.dimensions[i_date]
        domain = normalize(row.dimensions[i_domain])
        volume_by_day[day] += int(row.metrics[i_count])
        if domain in owned_domains:
            owned_share_by_day[day] += row.metrics[i_share]

    return sorted(
        (
            (day, volume_by_day[day], owned_share_by_day[day])
            for day in volume_by_day
        ),
        key=lambda p: p[0],
    )


def current_and_prior_windows(inclusive_end, days):
    """Two (start, end_exclusive) pairs of equal length, back-to-back."""
    current = (
        inclusive_end - timedelta(days=days - 1),
        inclusive_end + timedelta(days=1),     # +1 day → exclusive end
    )
    prior = (current[0] - timedelta(days=days), current[0])
    return current, prior


def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


# Resolve name → ID, then run both windows.
category_id    = find_category_id(CATEGORY_NAME)
owned          = get_owned_domains(category_id)
current, prior = current_and_prior_windows(INCLUSIVE_END, DAYS)

current_rows = get_daily_citations(category_id, owned, *current)
prior_rows   = get_daily_citations(category_id, owned, *prior)

print(f"{'Date':<12} {'Volume':>8} {'Owned share':>14}")
print("Current period:")
for day, volume, share in current_rows:
    print(f"  {day:<10} {volume:>8} {share:>13.1%}")

print("\nPrevious period:")
for day, volume, share in prior_rows:
    print(f"  {day:<10} {volume:>8} {share:>13.1%}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

# Current window
curl -X POST "https://api.tryprofound.com/v1/reports/citations" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["count", "citation_share"],
    "dimensions": ["date", "root_domain"],
    "date_interval": "day",
    "filters": [{"field": "prompt_type", "operator": "is", "value": "visibility"}],
    "pagination": {"limit": 50000, "offset": 0}
  }'

# Previous window — same body, prior dates.
curl -X POST "https://api.tryprofound.com/v1/reports/citations" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-04-28",
    "end_date": "2026-05-05",
    "metrics": ["count", "citation_share"],
    "dimensions": ["date", "root_domain"],
    "date_interval": "day",
    "filters": [{"field": "prompt_type", "operator": "is", "value": "visibility"}],
    "pagination": {"limit": 50000, "offset": 0}
  }'
```

## Common follow-ups

> **[Headline Citation Share + delta](/cookbook/citations/citation-share)**
>
> The single-number KPI above the chart, plus its period-over-period delta.

> **[Citation Rank by domain](/cookbook/citations/top-citing-domains)**
>
> Every cited domain ranked by share, with the per-row pp delta column.


# Citation Rank by domain (with deltas)
Source: https://docs.tryprofound.com/cookbook/citations/top-citing-domains

Build the Citation Rank leaderboard: domains sorted by share, with the per-row pp change vs the previous period

The "Citation Rank" panel beside the Citation Share tile in the Profound
app is the citations endpoint grouped by `root_domain`, sorted by
`citation_share` descending, with each row showing the change in share
vs the previous window. Two calls (current + prior), merge by domain,
diff the shares.

## How this example works

1. **`dimensions: ["root_domain"]`** — one row per cited domain in each
   call. Run both the current and prior windows.
2. **Merge by domain.** For each current-window row, look up the
   previous-window share (default 0 if missing) and diff.
3. **Sort by current `citation_share` descending** to match the UI.

> [!NOTE]
>
> **`count` and `citation_share` measure different things.** `count` is the
> raw number of citations to that domain in the window. `citation_share`
> is `count / total_citations` **per AI model, then averaged across
> models**. A domain cited heavily on a low-volume model (e.g. Perplexity)
> can have a higher share than a domain with more total citations spread
> across high-volume models. The two columns won't sort in lockstep.
> Pick the one that matches the question you're answering and sort on it.

```python Python theme={null}
import os
from datetime import date, timedelta
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
DAYS          = 7
INCLUSIVE_END = date(2026, 5, 11)    # the last day of your current window
TOP_N         = 10


def get_domain_shares(category_id, start, end):
    """Return {domain: share} for every cited domain in the window."""
    res = client.reports.citations(
        category_id=category_id,
        start_date=start.isoformat(),
        end_date=end.isoformat(),
        metrics=["count", "citation_share"],
        dimensions=["root_domain"],
        filters=[{"field": "prompt_type", "operator": "is", "value": "visibility"}],
        pagination={"limit": 50000, "offset": 0},
    )
    m = res.info.query["metrics"]
    d = res.info.query["dimensions"]
    i_count  = m.index("count")
    i_share  = m.index("citation_share")
    i_domain = d.index("root_domain")
    return {
        row.dimensions[i_domain]: {
            "count": row.metrics[i_count],
            "share": row.metrics[i_share],
        }
        for row in res.data
    }


def current_and_prior_windows(inclusive_end, days):
    """Two (start, end_exclusive) pairs of equal length, back-to-back."""
    current = (
        inclusive_end - timedelta(days=days - 1),
        inclusive_end + timedelta(days=1),     # +1 day → exclusive end
    )
    prior = (current[0] - timedelta(days=days), current[0])
    return current, prior


def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


# Resolve name → ID, run current + prior, then build the leaderboard with deltas.
category_id    = find_category_id(CATEGORY_NAME)
current, prior = current_and_prior_windows(INCLUSIVE_END, DAYS)

current_domains = get_domain_shares(category_id, *current)
prior_domains   = get_domain_shares(category_id, *prior)

rows = sorted(
    (
        {
            "domain":    name,
            "count":     d["count"],
            "share":     d["share"],
            "delta_pp":  (d["share"] - prior_domains.get(name, {"share": 0})["share"]) * 100,
        }
        for name, d in current_domains.items()
    ),
    key=lambda r: r["share"],
    reverse=True,
)[:TOP_N]

for rank, r in enumerate(rows, start=1):
    sign = "+" if r["delta_pp"] >= 0 else "−"
    delta_str = f"{sign}{abs(r['delta_pp']):.1f} pp"
    print(f"{rank:>2}. {r['domain']:<32} {r['share']:>6.1%}  {delta_str}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

# Current window
curl -X POST "https://api.tryprofound.com/v1/reports/citations" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["count", "citation_share"],
    "dimensions": ["root_domain"],
    "filters": [{"field": "prompt_type", "operator": "is", "value": "visibility"}],
    "pagination": {"limit": 50000, "offset": 0}
  }'

# Previous window — same body, prior dates. Merge by root_domain client-side
# to compute the per-row share delta.
curl -X POST "https://api.tryprofound.com/v1/reports/citations" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-04-28",
    "end_date": "2026-05-05",
    "metrics": ["count", "citation_share"],
    "dimensions": ["root_domain"],
    "filters": [{"field": "prompt_type", "operator": "is", "value": "visibility"}],
    "pagination": {"limit": 50000, "offset": 0}
  }'
```

## Page-level breakdown

Swap `root_domain` for `url` in `dimensions` to get the top citing **pages**
(one row per URL) instead of the top citing domains.


# Introduction
Source: https://docs.tryprofound.com/cookbook/introduction

End-to-end recipes for recreating the Profound dashboards with Profound API

The Cookbook is a collection of **end-to-end recipes** that show how to recreate
each chart, KPI, and table you see in the Profound app using only public API
calls. Every recipe is self-contained: it lists the exact endpoints used,
explains the request shape, and shows runnable Python and `curl` you can copy
into your own code.

If you just need the endpoint reference (every parameter, every metric), see
the [REST API](/rest-api/introduction) tab. The Cookbook is the layer above
that: common things people want to **build** with the API.

## Give your AI assistant context

Building with Claude, ChatGPT, Cursor, or another AI coding assistant?
Paste this URL. It's a compact index of every page in these docs that
your assistant can fetch on demand:

```
https://docs.tryprofound.com/llms.txt
```

## What you'll need

**1. An API key**

Generate one in **Settings → API Keys** in the Profound app. See
[Authentication](/rest-api/authentication) if you don't have one yet.

**2. A category ID**

Every report query is scoped to a category. See
[Find your category ID](/cookbook/setup/find-your-category-id) to
look one up by name.

**3. The Python SDK (optional)**

The recipes default to the Python SDK because it's the most common path.
Install with `pip install profound`. Every recipe also has a `curl` tab
if you'd rather hit the REST API directly.

## One-time setup

Every recipe starts by constructing a client like this (you only need it
once per script, even if you're chaining several recipes):

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here
```

## Conventions used in every recipe

Three things to internalize before you start. These come up in every recipe
and trip up most new users:

> **end\_date is exclusive — bump it by one day**
>
> The API parses `end_date` at the start of day, so it **excludes** the date
> you send. To include all of `2026-05-10`, send `end_date="2026-05-11"`.
>
> Display the inclusive value to your users; send `+1 day` to the API.

> **Read array positions from info.query**
>
> Every response includes `info.query.metrics` and `info.query.dimensions`,
> which echo back the exact order the API used when packing each row's
> `metrics` and `dimensions` arrays.
>
> ```python theme={null}
> order = res.info.query["metrics"]   # e.g. ["visibility_score"]
> i = order.index("visibility_score")
> score = res.data[0].metrics[i]
> ```
>
> Don't hardcode positions. Reading them from `info.query` keeps your code
> correct for any combination of metrics you request.

> **The API doesn't compute deltas — you do**
>
> Period-over-period changes (the green/red `+1.0%` you see on every KPI
> tile) are computed client-side. Run the same call twice (once for the
> current window, once for the previous window of equal length) and diff
> the two scores in your code.

## Before you build

> **[Conventions & gotchas](/cookbook/setup/conventions)**
>
> Rate limits, exclusive end-dates, `info.query` ordering, errors,
> pagination. Read this once, paste into your AI assistant.

> **[Data model](/cookbook/setup/data-model)**
>
> How Categories, Topics, Prompts, Tags, Assets, and Personas relate.

> **[Endpoints at a glance](/cookbook/setup/endpoints-at-a-glance)**
>
> Every endpoint, its metrics/dimensions, and what it's for: one
> scannable page.

> **[How metrics are calculated](/cookbook/metrics/how-metrics-are-calculated)**
>
> The formula behind every metric, a worked example, and how to reproduce
> each number from the Answers API.

> **[Why past numbers change](/cookbook/metrics/why-past-numbers-change)**
>
> Topics, tags, and brands are applied from your current configuration, so
> editing your setup changes historical numbers.

## Recipes

### Setup

> **[Find your category ID](/cookbook/setup/find-your-category-id)**
>
> List the categories your key can see and pick one programmatically.

> **[List your owned assets](/cookbook/setup/list-owned-assets)**
>
> Get every asset in a category with its `is_owned` flag and domains.

### Visibility

> **[Visibility Score for one asset](/cookbook/visibility/asset-visibility-score)**
>
> An asset's score for the window plus the change vs the prior window.

> **[Visibility over time](/cookbook/visibility/visibility-over-time)**
>
> Build the daily / weekly / monthly line chart for an asset.

> **[Headline and daily, together](/cookbook/visibility/headline-and-daily)**
>
> Fetch both at once. Understand why they're different calls.

> **[Top-N leaderboard](/cookbook/visibility/leaderboard)**
>
> Rank every asset in a category by any visibility metric.

> **[Compare to competitors](/cookbook/visibility/compare-competitors)**
>
> Multi-line chart of any hand-picked set of assets.

> **[Segment by model / region / persona](/cookbook/visibility/segment-by-model)**
>
> Break a single asset's score down by the AI surface that answered.

### Citations

> **[Citation Share + delta](/cookbook/citations/citation-share)**
>
> Owned-domain share of all citations, with the period-over-period change.

> **[Citation Share and volume over time](/cookbook/citations/citation-share-over-time)**
>
> Daily owned-domain share line **and** daily total citation volume,
> both from one call.

> **[Citation Rank by domain](/cookbook/citations/top-citing-domains)**
>
> Every cited domain ranked by share, with the per-row pp delta column.


# How metrics are calculated
Source: https://docs.tryprofound.com/cookbook/metrics/how-metrics-are-calculated

The formula behind every report metric, a worked example, and how to reproduce each number from the Answers API.

Every report metric is computed for each AI model separately, then averaged
across models with equal weight. A model with many runs counts the same as a
model with few.

$$
\text{score} = \frac{\text{score}_{\text{model 1}} + \text{score}_{\text{model 2}} + \dots + \text{score}_{\text{model } M}}{M}
$$

`M` is the number of models with at least one brand mention in the period. A
model that mentioned other brands but not yours counts in `M` with a score of
0\. Average position is the exception: it is averaged over the models in which
your brand appeared. Counts, such as the number of mentions or citations, are
summed across models rather than averaged.

## Per-model formulas

### Visibility score

$$
\frac{\text{runs mentioning your brand}}{\text{runs mentioning any brand}}
$$

Any brand named in the answer counts, not only your configured competitors,
unless you have excluded it under Configure Competitors in brand settings.
Runs in which no brand was mentioned are excluded from both the numerator and
the denominator. Returned as 0 to 1 by the API and shown as a percentage in
the platform.

### Share of voice

$$
\frac{\text{mentions of your brand}}{\text{mentions of all brands}}
$$

Each brand counts once per run. Returned as 0 to 1 by the API and shown as a
percentage in the platform.

### Average position

$$
\frac{\text{sum of your brand's positions}}{\text{runs mentioning your brand}}
$$

The first brand mentioned in an answer is position 1. Lower is better.

### Citation share

$$
\frac{\text{citations of your domain}}{\text{all citations}}
$$

Returned as 0 to 1 by the API and shown as a percentage in the platform. The
same formula applies per page when grouping by `page`.

### Positive and negative sentiment

$$
100 \times \frac{\text{positive claims about your brand}}{\text{all claims about your brand}}
$$

Negative sentiment is the same with negative claims. Every claim is scored
positive or negative, so the two values sum to 100. Both are returned as 0 to
100\.

## Example: two models

| | Model A | Model B | Result |
| - | - | - | - |
| Runs that mentioned any brand | 10 | 20 | |
| Runs that mentioned your brand | 4 | 12 | |
| **Visibility score** | 4 / 10 = 0.40 | 12 / 20 = 0.60 | (0.40 + 0.60) / 2 = **0.50** |
| Mentions of all brands | 25 | 40 | |
| **Share of voice** | 4 / 25 = 0.16 | 12 / 40 = 0.30 | (0.16 + 0.30) / 2 = **0.23** |
| Sum of your brand's positions | 7 | 27 | |
| **Average position** | 7 / 4 = 1.75 | 27 / 12 = 2.25 | (1.75 + 2.25) / 2 = **2.00** |
| Citations, all domains | 50 | 100 | |
| Citations of your domain | 5 | 25 | |
| **Citation share** | 5 / 50 = 0.10 | 25 / 100 = 0.25 | (0.10 + 0.25) / 2 = **0.175** |
| Claims about your brand | 20 | 10 | |
| Positive claims | 12 | 8 | |
| **Positive sentiment** | 12 / 20 = 60 | 8 / 10 = 80 | (60 + 80) / 2 = **70** |
| Negative claims | 8 | 2 | |
| **Negative sentiment** | 8 / 20 = 40 | 2 / 10 = 20 | (40 + 20) / 2 = **30** |

> [!WARNING]
>
> Pooling runs across models first (16 / 30 = 0.53 for visibility) is wrong
> because each model has equal weight. Averaging daily scores to produce a
> monthly score is also wrong. A monthly score is computed from all runs in the
> month. See [Conventions & gotchas](/cookbook/setup/conventions#dont-average-daily-rows-to-get-a-period-score).

## Reproducing metrics from the Answers API

[Get Answers](/rest-api/reports/query-answers-v2) returns one row per run.
The fields you need:

| Field | Contents |
| - | - |
| `run_id` | Identifier of the run. |
| `model` | The AI model, as an id and a name. |
| `mentions` | Names of the brands in the answer, in order of first appearance, each brand once. Every brand Profound detected is listed, including any you have excluded under Configure Competitors. |
| `citations` | URLs cited in the answer. `citation_details` adds the hostname and title of each and must be requested explicitly in `include`. |
| `sentiment_claims` | Claims extracted from the answer, each with `asset` (the brand it is about), `theme`, `claim`, and `sentiment` (`positive` or `negative`). |
| `date`, `topic`, `tags`, `prompt`, `prompt_id`, `region`, `persona` | Current labels for the run. Use them to select rows. |

Select the rows for the dates, topics, tags, and regions you want. If you have
excluded brands under Configure Competitors, remove them from `mentions` first.
For each metric, compute the fraction below for each model separately, then
average the per-model values. Multiply the sentiment values by 100.

| Metric | Numerator | Denominator |
| - | - | - |
| Visibility score | Distinct `run_id` where your brand is in `mentions` | Distinct `run_id` where `mentions` is not empty |
| Share of voice | Distinct `run_id` where your brand is in `mentions` | Entries in `mentions`, summed over all runs |
| Average position | Sum of your brand's positions in `mentions`, with the first entry as 1 | Distinct `run_id` where your brand is in `mentions` |
| Citation share | Entries in `citations` whose hostname is your domain | Entries in `citations`, summed over all runs |
| Positive sentiment | Entries in `sentiment_claims` where `asset` is your brand and `sentiment` is `positive` | Entries in `sentiment_claims` where `asset` is your brand |
| Negative sentiment | Entries in `sentiment_claims` where `asset` is your brand and `sentiment` is `negative` | Entries in `sentiment_claims` where `asset` is your brand |

Count by `run_id`, join topics and prompts on their ids, match brands by name,
and use Eastern Time dates.

> [!NOTE]
>
> The Answers API returns the topic, tags, and brand list as they are configured
> today, so numbers you rebuild from it change when your configuration changes.
> See [Why past numbers change](/cookbook/metrics/why-past-numbers-change).


# Why past numbers change
Source: https://docs.tryprofound.com/cookbook/metrics/why-past-numbers-change

Answers are stored once. Topics, tags, and brands are applied from your current configuration on every request, so editing your setup changes historical numbers.

* Profound stores each answer once. The response text, citations, and date do not change.
* Topic, tags, and brand attribution are not stored with the answer. They are read from your current configuration each time a report loads or the API is called.
* Editing your configuration therefore changes past numbers. Moving a prompt to a new topic moves all of its past answers. Excluding a brand under Configure Competitors removes its past mentions from every report.
* Reports reflect the current configuration only. An older configuration cannot be viewed.

## How it works

Your configuration defines which prompts run and how their answers are
labeled. Collection runs each prompt and stores the answer. Reports and the
Answers API join stored answers to the current configuration on every request.

| Stage | Contents | Mutability |
| - | - | - |
| Configuration | Category, topics (one per prompt), prompts (permanent id, editable text), tags (many-to-many with prompts), brands and aliases, the Configure Competitors include or exclude list, models, regions, personas | Changes at any time |
| Collection | One run per prompt, model, region, persona, and day. Brand names in the response are extracted once and stored. | Written once, never modified |
| Serving | Reports and the Answers API join stored answers to the configuration as it is now | Evaluated on every request; changes apply to all dates |

Each stored run carries `run_id`, `date`, `prompt_id`, `model`, `region`,
`persona`, `response`, `citations`, `search_queries`, and the brand names
extracted from the response. At request time, `prompt_id` resolves to the
current topic, tags, text, and status, and the extracted brand names resolve
to the current brand list. Deleted prompts are excluded. Paused prompts are
included.

A **run** is one prompt sent to one model from one region on one day. Each
run produces one answer and one row in the Answers API. A **mention** is a
brand named in an answer. A **claim** is a statement about a brand extracted
from an answer, scored positive or negative. See the
[data model](/cookbook/setup/data-model) for how categories, topics, prompts,
tags, and assets relate.

Prompts run once a day per model and region. Days are counted in Eastern Time.

## Effect of configuration changes on past data

| Change | Effect on past data |
| - | - |
| Move a prompt to another topic | All of its past answers move to the new topic. |
| Add or remove a tag | All of the prompt's past answers gain or lose the tag. |
| Rename a topic or tag | The name changes. Numbers do not. |
| Edit a prompt's wording | Numbers do not change. Past answers display the new wording even though they were generated from the old wording. |
| Pause a prompt | The prompt stops running. Its past answers remain in reports. |
| Delete a prompt | The prompt stops running and its past answers are removed from reports and the API. Restoring the prompt restores them. |
| Exclude a brand under Configure Competitors | Its past mentions stop counting on every date. Share of voice changes for every brand. Your visibility score rises if past answers named only the excluded brand and no other, because those answers leave the denominator. Your average position improves in answers where the excluded brand was named before your brand. Switching to an include list does the same for every brand not on it. |
| Add a prompt, topic, model, or region | Data begins on the first day it runs. |

Profound does not re-run or re-analyze old answers. Only the labels applied to
them change. Answers API rows reflect configuration changes immediately. Some
summary reports refresh on a schedule and can lag briefly.

## Store your own configuration snapshots

The Answers API returns the current topic, tags, and brand list, so metrics
you compute from it change whenever your configuration changes. To keep fixed
historical numbers, save the following once a day and join to the saved copies
instead:

| Save | Reason |
| - | - |
| The topic of each `prompt_id` | A moved prompt does not change your past topic totals. |
| The tags of each `prompt_id` | The same, for tags. |
| Each prompt's wording | You know which text generated an old answer. |
| Your Configure Competitors include or exclude list | The Answers API does not apply it, so you need the list as of that date to match the report. |
| The rows you have fetched | Deleted prompts are removed from the API. Saved rows are not affected. |

## Recommendations

* Pause prompts instead of deleting them to keep their history.
* Move prompts between topics and tags deliberately. Each move changes past numbers for both the old and new group.
* Expect share of voice to change across all dates when you exclude a brand or switch to an include list.
* If a number must not change, for example in a quarterly report, export it at the time of reporting. The live report always reflects the current configuration.


# Conventions & gotchas
Source: https://docs.tryprofound.com/cookbook/setup/conventions

Read once, paste into your AI assistant, and skip the most common day-one mistakes

## Authentication

Every request needs your API key in the `X-API-Key` header. The Python
SDK reads it from the `PROFOUND_API_KEY` env var.

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])
categories = client.organizations.categories.list()
```

```bash curl theme={null}
curl https://api.tryprofound.com/v1/org/categories \
  -H "X-API-Key: your_api_key_here"
```

Generate a key in the app under **Settings → API Keys**. Treat it like a
password: it has full access to your org's data, including write endpoints
such as creating prompts and running agents.

## Rate limit

**600 requests per hour, per key.** Anything above returns `429 Too Many
Requests`. Cache responses where you can and batch period-over-period or
multi-asset queries instead of fanning out.

## `end_date` is exclusive — add one day

`end_date` is parsed at the **start** of that day in **Eastern Time**,
so it's excluded from the response. To include all of `May 10`, send
`end_date="2026-05-11"`.

| Window you want to display | What to send |
| - | - |
| `May 4 → May 10` (7 days, inclusive) | `start_date="2026-05-04", end_date="2026-05-11"` |
| `April 1 → April 30` (full month) | `start_date="2026-04-01", end_date="2026-05-01"` |

The `date_interval` buckets (`"hour"` / `"day"` / `"week"` / `"month"` /
`"quarter"` / `"year"` / `"relative_week"`) are also computed in ET. Weeks
start on Sunday.

## Read column positions from `info.query`, not your request

Each row in the response packs its `metrics` and `dimensions` as arrays.
The order of values in those arrays is the order listed in
`info.query.metrics` and `info.query.dimensions`, which echo your request.
Look positions up there instead of hardcoding them:

```python theme={null}
order   = res.info.query["metrics"]
i_score = order.index("visibility_score")
score   = res.data[0].metrics[i_score]
```

A response always looks like this:

```json theme={null}
{
  "info": {
    "total_rows": 12345,
    "query": {
      "metrics": ["visibility_score", "share_of_voice"],
      "dimensions": ["asset_name"]
    }
  },
  "data": [
    { "metrics": [0.42, 0.17], "dimensions": ["<your-asset>"] }
  ]
}
```

## Period-over-period deltas are client-side

The API doesn't return change vs the previous period. Run the same call
twice (current window and a prior window of equal length) and subtract.

## Don't average daily rows to get a period score

A call with `dimensions=["date"]` returns one row per day. A call without
`date` returns one row for the whole window. These are **different
numbers**: the period score is traffic-weighted, an average of daily
rows is not. Use the no-`date` call for headlines; use the with-`date`
call for charts. Never derive one from the other.

## Pagination

Default `pagination.limit` is `10,000`. Max is `50,000`. Use
`info.total_rows` (returned on every response) to decide whether to
paginate. Almost all queries fit in a single 50k page; only heavy
`dimensions=["url", ...]` citation queries usually need a second page.

```python theme={null}
pagination={"limit": 50000, "offset": 0}
```

If you do need more, increment `offset` by `limit` until you've covered
`total_rows`.

## Filters

Every report endpoint accepts a `filters` array of `{field, operator,
value}` objects:

```json theme={null}
{ "field": "asset_name", "operator": "is", "value": "<your-asset-name>" }
```

| Operator | What it does |
| - | - |
| `is` | Exact match (scalar value) |
| `not_is` | Not equal |
| `in` | Match any value in an array |
| `not_in` | Match none of the values in an array |
| `contains` | Substring match (case-sensitive) |
| `not_contains` | Negated substring match (case-sensitive) |
| `contains_case_insensitive` | Substring match (case-insensitive) |
| `not_contains_case_insensitive` | Negated substring match (case-insensitive) |
| `matches` | Regex match |

UUID filters (`model_id`, `region_id`, `topic_id`, `tag_id`, `persona_id`)
accept only `is`, `not_is`, `in`, and `not_in`. The substring and regex
operators apply to name-based fields such as `asset_name`, `root_domain`,
and `url`.

`prompt_type` (with values like `"visibility"`) maps to the app's
**Prompt Type** filter. Send `prompt_type=visibility` on Citations queries
to mirror the default UI scope. `prompt_type` is deprecated in favor of
`analysis_type` (values `visibility`, `sentiment`, `sentiment_v2`,
`accuracy`); both are accepted today.

## Error responses

| Status | Meaning | What to check |
| - | - | - |
| `422` | Validation error | The response body's `detail` field. |
| `401` | API key missing or invalid | The `X-API-Key` header; whether the key has been revoked. |
| `403` | Key valid but no access | The `category_id` may belong to a different org. |
| `404` | Wrong path | Typo or wrong API version. |
| `429` | Rate limited | Back off; throttle to ≤600/hr. |
| `5xx` | Server error | Retry with exponential backoff. |

## Timezones

All bucketing happens in **Eastern Time**. A "last 7 days" range anchored
to your local clock can land on a different ET day than you expect.
Anchor scheduled jobs to ET:

```python theme={null}
from datetime import datetime
from zoneinfo import ZoneInfo

today_et = datetime.now(ZoneInfo("America/New_York")).date()
```


# Data model
Source: https://docs.tryprofound.com/cookbook/setup/data-model

How everything in Profound nests together, from your Organization down to the slice dimensions on every report

Your **Organization** holds many **Categories**. Each Category contains
its **Assets** (the brands you're tracking), grouped into **Topics**, made
up of **Prompts**. Each Prompt then fans out across four dimensions
(**Platform**, **Region**, **Persona**, and **Tag**), which is how reports
slice the data.

## Entity reference

| Entity | What it is |
| - | - |
| **Organization** | Your Profound account. Has many Categories. |
| **Category** | A workspace, typically one brand and its competitive set. Every report query is scoped to one Category. |
| **Asset** | A brand, company, or product tracked in the Category. Has `is_owned`, `website`, and `alternate_domains`. |
| **Topic** | A grouping of related prompts (e.g. "Pricing", "Integrations"). |
| **Prompt** | A question collected against AI models (e.g. "Best CRM for startups?"). |
| **Platform** | The AI surface that answered (ChatGPT, Claude, Perplexity, …). Called `model` as a dimension and `model_id` as a filter field in API params. |
| **Region** | The country the answer is scoped to (US, UK, Germany, …). |
| **Persona** | An audience profile (e.g. "small business owner"). |
| **Tag** | A free-form label applied to a Prompt. |

## What this means for queries

1. **Every report call is scoped to one `category_id`.** No cross-category
   queries.
2. **`filters` narrow which prompts contribute** — by `asset_name`,
   `topic_id`, `tag_id`, and the like.
3. **`dimensions` break results out by slice** — `model`, `region`,
   `persona`, or `tag`.

## Endpoints by entity

| Entity | List endpoint |
| - | - |
| Categories | [`GET /v1/org/categories`](/api-reference/organization/get-categories) |
| Topics | [`GET /v1/org/categories/{id}/topics`](/api-reference/organization/get-category-topics) |
| Prompts | [`GET /v1/org/categories/{id}/prompts`](/api-reference/organization/list-prompts) |
| Tags | [`GET /v1/org/categories/{id}/tags`](/api-reference/organization/get-category-tags) |
| Assets | [`GET /v1/org/categories/{id}/assets`](/api-reference/organization/get-category-assets) |
| Personas | [`GET /v1/org/categories/{id}/personas`](/api-reference/organization/get-category-personas) |
| Platforms (org-level) | [`GET /v1/org/models`](/api-reference/organization/get-models) |
| Regions (org-level) | [`GET /v1/org/regions`](/api-reference/organization/get-regions) |

Click any endpoint above to jump to its full parameter reference.


# Endpoints at a glance
Source: https://docs.tryprofound.com/cookbook/setup/endpoints-at-a-glance

The Profound API endpoints the Cookbook recipes use, plus the other main endpoint groups, with key metrics/dimensions/filters: one scannable page

Orient yourself in 30 seconds. Each row is one endpoint; click through
for the full parameter reference. This page covers the endpoints the
recipes use plus the main endpoint groups; the complete list (including
Documents, Knowledge bases, Projects, Integrations, and OpenAI Ads) is in
the [REST API](/rest-api/introduction) tab. For deeper context on conventions
(dates, ordering, pagination), see [Conventions & gotchas](/cookbook/setup/conventions).

## Organization & setup

| Method | Path | Use it for |
| - | - | - |
| `GET` | `/v1/org/categories` | List every category your key can see. |
| `GET` | `/v1/org/categories/{id}/assets` | Assets in a category, with `is_owned` + domains. |
| `GET` | `/v1/org/categories/{id}/topics` | Configured topics for the category. |
| `GET` | `/v1/org/categories/{id}/tags` | Configured tags. |
| `GET` | `/v1/org/categories/{id}/personas` | Configured personas. |
| `GET` | `/v1/org/categories/{id}/regions` | Regions enabled on the category. |
| `GET` | `/v1/org/categories/{id}/citation-categories` | Citation-category filter options for the category. |
| `GET` | `/v1/org/categories/{id}/citation-tags` | Custom citation tags defined for the category. |
| `GET` | `/v1/org/assets` | Assets across your org. |
| `GET` | `/v1/org/regions` | Regions enabled on your org. |
| `GET` | `/v1/org/models` | AI models (ChatGPT, Claude, etc.) enabled on your org. |
| `GET` | `/v1/org/domains` | Domains your org owns. |
| `GET` | `/v1/org/personas` | Personas enabled on your org. |

## Reports

Answer Engine Insights report endpoints are `POST` requests scoped by
`category_id` and return the `{info, data}` shape. Visibility, citations,
sentiment, and query-fanouts reports take `category_id`, `start_date`,
`end_date`, and `metrics`, plus report-specific optional fields.

Traffic reports use a different shape: bots and referrals are scoped by
`domain` rather than `category_id`. V2 FactCheck reports and V2 prompt
answers are category-and-date scoped, but do not take a `metrics` field.

### `/v1/reports/visibility`

How often your asset appears in AI answers.

| Metrics | Dimensions | Common filters |
| - | - | - |
| `visibility_score`, `share_of_voice`, `average_position`, `mentions_count`, `executions` | `date`, `asset_name`, `model`, `region`, `persona`, `topic`, `tag` | `asset_name`, `model_id`, `region_id`, `persona_id`, `topic_id`, `tag_id` |

Common recipes: [Visibility Score (with delta)](/cookbook/visibility/asset-visibility-score),
[over time](/cookbook/visibility/visibility-over-time),
[leaderboard](/cookbook/visibility/leaderboard),
[compare competitors](/cookbook/visibility/compare-competitors),
[segment by model/region/persona](/cookbook/visibility/segment-by-model).

### `/v1/reports/citations`

Which URLs and domains AI answers cite.

| Metrics | Dimensions | Common filters |
| - | - | - |
| `count`, `citation_share` | `date`, `root_domain`, `url`, `citation_category`, `model`, `region`, `persona` | `prompt_type` (e.g. `"visibility"`), `model_id`, `region_id`, `persona_id` |

Common recipes: [Citation Share + delta](/cookbook/citations/citation-share),
[share + volume over time](/cookbook/citations/citation-share-over-time),
[rank by domain](/cookbook/citations/top-citing-domains).

> [!NOTE]
>
> `citation_share` is per-AI-model averaged. `count` is the raw number.
> Sorting by one doesn't track the other. Pick the question you're
> answering and sort on the matching metric.

### `/v1/reports/sentiment`

How positively/negatively AI answers reference your asset.

| Metrics | Dimensions | Common filters |
| - | - | - |
| `positive`, `negative`, `occurrences` | `date`, `asset_name`, `theme`, `sentiment_type`, `topic`, `model`, `region`, `persona` | `asset_name`, `theme`, `topic_id`, `model_id`, `region_id`, `persona_id` |

> [!WARNING]
>
> `/v1/reports/sentiment` is the legacy sentiment processor and its numbers
> will not match the Profound app's headline value. The app reads from
> Sentiment v2: use `POST /v2/reports/sentiment` (see
> [Filtering & concepts](/rest-api/reports/reports-v2-overview)). Its
> `positive_sentiment` / `negative_sentiment` metrics are percentages that
> sum to 100 per row, so they are not comparable with v1's counts.

### `/v1/reports/query-fanouts`

The query-fanout view (how prompts cascade across AI surfaces). Same
`{info, data}` shape.

### `/v1/reports/referrals` (and `/v2/reports/referrals`)

Human-referral traffic to your domains attributed back to AI answers.

### `/v1/reports/bots` (and `/v2/reports/bots`)

AI crawler / bot traffic to your domains.

### `/v2/reports/factcheck` and `/v2/reports/factcheck/claims`

FactCheck scores and inaccurate claims for a category. These endpoints take
`category_id`, `start_date`, and `end_date`, but not `metrics`.

### V2 report endpoints

`POST /v2/reports/visibility`, `/v2/reports/citations`,
`/v2/reports/sentiment`, and `/v2/reports/query-fanouts` are the V2
equivalents of the V1 reports above, with an inclusive `end_date`, a
`filter` tree, named metric fields on each row, and cursor pagination.
`/v2/reports/shopping/*` and `/v2/reports/social/youtube/*` are V2-only.
Each has a `/stream` SSE variant. See
[Filtering & concepts](/rest-api/reports/reports-v2-overview).

## Answers (per-prompt data)

| Method | Path | Use it for |
| - | - | - |
| `POST` | `/v1/prompts/answers` | Fetch raw AI answers for prompts in a category. |
| `POST` | `/v2/prompts/answers` | Fetch raw per-execution answer rows for a category. |

## Agents

| Method | Path | Use it for |
| - | - | - |
| `GET` | `/v1/agents` | List configured agents. |
| `GET` | `/v1/agents/{id}` | Get one agent's config. |
| `POST` | `/v1/agents/{id}/runs` | Start a new agent run. |
| `GET` | `/v1/agents/{id}/runs/{run_id}` | Poll a run's status. |

## Prompts

| Method | Path | Use it for |
| - | - | - |
| `GET` | `/v1/org/categories/{id}/prompts` | List prompts in a category. |
| `POST` | `/v1/org/categories/{id}/prompts` | Add a prompt. |
| `PATCH` | `/v1/org/categories/{id}/prompts` | Edit prompts. |
| `PATCH` | `/v1/org/categories/{id}/prompts/status` | Enable/disable prompts. |

## Content Optimization (beta)

| Method | Path | Use it for |
| - | - | - |
| `GET` | `/v1/content/{asset_id}/optimization` | Optimization opportunities for an asset. |
| `GET` | `/v1/content/{asset_id}/optimization/{content_id}` | One optimization in detail. |

***

For the canonical parameter reference (every operator, every field,
required vs optional), see the [REST API](/rest-api/introduction) tab.


# Find your category ID
Source: https://docs.tryprofound.com/cookbook/setup/find-your-category-id

Every report call requires a category_id. Here's how to look one up programmatically

Every report endpoint (`/v1/reports/visibility`, `/v1/reports/citations`,
`/v1/reports/sentiment`) requires a `category_id`. List the categories
your key can see and pick the right one.

## How this example works

1. **One call lists everything.** No filters, no pagination.
2. **Match by name** if you know what you're looking for, or print them all
   if you don't.

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

categories = client.organizations.categories.list()

# Print every category your key can see.
for c in categories:
    print(f"{c.id}  {c.name}")

# Or: find one by name.
TARGET = "<your-category-name>"
match = next((c for c in categories if c.name == TARGET), None)
if match:
    print(f"\nFound Category '{TARGET}': {match.id}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

curl "https://api.tryprofound.com/v1/org/categories" \
  -H "X-API-Key: $PROFOUND_API_KEY"
```


# List your owned assets and domains
Source: https://docs.tryprofound.com/cookbook/setup/list-owned-assets

Get every asset in a category, with the is_owned flag and any associated domains

Most "us vs. them" recipes need to know which assets in a category belong
to you. This call returns every asset with its `is_owned` flag plus its
website and alternate domains: everything you need for filtering and
domain matching.

## How this example works

1. **One call per category.** Cache the result if you're going to make many
   report calls (assets change infrequently).
2. **Filter by `is_owned`** to get just your assets.
3. **Combine `website` + `alternate_domains`** to build the full set of
   domains you own (used by the [Citation Share
   recipe](/cookbook/citations/citation-share)).

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"


def list_owned(category_id):
    """Return (owned_asset_names, owned_domains) for the category."""
    names, domains = [], []
    for a in client.organizations.categories.assets(category_id):
        if not a.is_owned:
            continue
        names.append(a.name)
        if a.website:
            domains.append(a.website)
        domains.extend(a.alternate_domains or [])
    return names, domains


# Helper — translate the category name to the UUID the report API expects.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


# Resolve name → ID, then list owned assets and domains.
category_id = find_category_id(CATEGORY_NAME)
owned_names, owned_domains = list_owned(category_id)

print("Owned assets:")
for n in owned_names:
    print(f"  - {n}")

print("\nOwned domains:")
for d in owned_domains:
    print(f"  - {d}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here
export CATEGORY_ID=your-category-uuid

curl "https://api.tryprofound.com/v1/org/categories/$CATEGORY_ID/assets" \
  -H "X-API-Key: $PROFOUND_API_KEY"
```

## Companion lookups

The same `/v1/org/categories/{id}/...` pattern works for other category resources:

| Path | Returns |
| - | - |
| `/v1/org/categories/{id}/assets` | Assets in the category (used above) |
| `/v1/org/categories/{id}/topics` | Topics covered by the category |
| `/v1/org/categories/{id}/tags` | Tags configured for the category |
| `/v1/org/categories/{id}/personas` | Personas configured for the category |
| `/v1/org/categories/{id}/prompts` | Prompts the category runs against |


# Visibility Score for one asset
Source: https://docs.tryprofound.com/cookbook/visibility/asset-visibility-score

Get an asset's traffic-weighted Visibility Score for a window and the change vs the prior window

The headline number on every Visibility tile in the Profound app, plus
the `+1.0%` change badge beside it. Same call twice (current window and
prior window of equal length) and diff the two scores.

## How this example works

1. **Aggregate, not time-series.** Omit `date` from `dimensions`; filter
   on `asset_name`. Response has one row with a single traffic-weighted
   score.
2. **Run the same call twice** (once for the current window, once for
   the prior window of equal length), then subtract.
3. **Read positions from `info.query.metrics`.** Don't hardcode array
   indices.

```python Python theme={null}
import os
from datetime import date, timedelta
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
ASSET_NAME    = "<your-asset-name>"
METRIC        = "visibility_score"   # or share_of_voice, average_position, ...
DAYS          = 7
INCLUSIVE_END = date(2026, 5, 11)    # the last day of your current window


def get_visibility_score(category_id, asset_name, metric, start, end):
    """Aggregate score for one asset in one window."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start.isoformat(),
        end_date=end.isoformat(),
        metrics=[metric],
        filters=[{"field": "asset_name", "operator": "is", "value": asset_name}],
    )
    order = res.info.query["metrics"]
    return res.data[0].metrics[order.index(metric)]


def current_and_prior_windows(inclusive_end, days):
    """Two (start, end_exclusive) pairs of equal length, back-to-back."""
    current = (
        inclusive_end - timedelta(days=days - 1),
        inclusive_end + timedelta(days=1),     # +1 day → exclusive end
    )
    prior = (current[0] - timedelta(days=days), current[0])
    return current, prior


# Helpers — translate human-readable names into the IDs the report API needs.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


def find_asset_name(category_id, name):
    """Return the canonical asset name (case-insensitive) inside the category."""
    for a in client.organizations.categories.assets(category_id):
        if a.name.lower() == name.lower():
            return a.name
    raise ValueError(f"No asset named {name!r} in this category")


# Resolve names → IDs, then run both windows.
category_id    = find_category_id(CATEGORY_NAME)
asset_name     = find_asset_name(category_id, ASSET_NAME)
current, prior = current_and_prior_windows(INCLUSIVE_END, DAYS)

current_score = get_visibility_score(category_id, asset_name, METRIC, *current)
prior_score   = get_visibility_score(category_id, asset_name, METRIC, *prior)
delta_pp      = (current_score - prior_score) * 100

print(f"{asset_name} {METRIC}: {current_score:.1%}  ({delta_pp:+.1f} pp vs prev period)")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

# Current window
curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "filters": [{"field": "asset_name", "operator": "is", "value": "<your-asset-name>"}]
  }'

# Previous window — same body, prior dates. Diff the two scores client-side.
curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-04-28",
    "end_date": "2026-05-05",
    "metrics": ["visibility_score"],
    "filters": [{"field": "asset_name", "operator": "is", "value": "<your-asset-name>"}]
  }'
```

## Available metrics

Swap `visibility_score` for any of these to get a different KPI tile:

| Metric | What it is | Format | Direction |
| - | - | - | - |
| `visibility_score` | Fraction of answers the asset appeared in | Percent | Higher is better |
| `share_of_voice` | Fraction of all mentions the asset accounts for | Percent | Higher is better |
| `average_position` | Average rank when the asset is mentioned | Rank | **Lower** is better |
| `mentions_count` | Raw count of mentions | Number | Higher is better |
| `executions` | Number of prompts run | Number | Higher is better |

For `average_position`, flip the delta's color logic: a negative `delta`
is an improvement.

## Common follow-ups

> **[Plot the same metric over time](/cookbook/visibility/visibility-over-time)**
>
> Add `date` to dimensions to build a daily line chart.

> **[Headline + daily series, together](/cookbook/visibility/headline-and-daily)**
>
> Why the headline call and the chart call return different numbers.


# Compare your asset to competitors over time
Source: https://docs.tryprofound.com/cookbook/visibility/compare-competitors

Build a multi-line chart of Visibility Score for any list of assets

The multi-line "your asset vs competitors" chart in the Profound app is one
call. Add `asset_name` to dimensions to get per-asset rows, and use an `in`
filter to limit to the assets you care about.

> [!WARNING]
>
> **Don't average daily rows to derive a period score**: they're computed
> differently. See [Conventions & gotchas](/cookbook/setup/conventions#dont-average-daily-rows-to-get-a-period-score).

## How this example works

1. **`dimensions: ["date", "asset_name"]`** — one row per (date, asset).
2. **`filters` with `in` operator** — limit to a hand-picked set of assets.
   Without the filter you'd get every asset in the category.
3. **Pivot the response into one series per asset** before plotting.

```python Python theme={null}
import os
from collections import defaultdict
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
ASSETS        = ["<your-asset>", "<competitor-1>", "<competitor-2>", "<competitor-3>"]
START_DATE    = "2026-05-05"
END_DATE      = "2026-05-12"   # exclusive — returns data through 2026-05-11


def get_competitor_series(category_id, asset_names, start_date, end_date, interval="day"):
    """Daily Visibility Score for each asset in `asset_names`. Returns {asset: [(date, score), ...]}."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start_date,
        end_date=end_date,
        metrics=["visibility_score"],
        dimensions=["date", "asset_name"],
        date_interval=interval,
        filters=[{"field": "asset_name", "operator": "in", "value": asset_names}],
    )
    m_order = res.info.query["metrics"]
    d_order = res.info.query["dimensions"]
    i_score = m_order.index("visibility_score")
    i_date  = d_order.index("date")
    i_asset = d_order.index("asset_name")

    by_asset = defaultdict(list)
    for row in res.data:
        by_asset[row.dimensions[i_asset]].append(
            (row.dimensions[i_date], row.metrics[i_score])
        )
    for asset in by_asset:
        by_asset[asset].sort(key=lambda x: x[0])
    return by_asset


# Helpers — translate human-readable names into the IDs the report API needs.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


def find_asset_names(category_id, names):
    """Return the canonical names for each input name (case-insensitive)."""
    by_lower = {
        a.name.lower(): a.name
        for a in client.organizations.categories.assets(category_id)
    }
    out = []
    for n in names:
        if n.lower() not in by_lower:
            raise ValueError(f"No asset named {n!r} in this category")
        out.append(by_lower[n.lower()])
    return out


# Resolve names → IDs, then run.
category_id   = find_category_id(CATEGORY_NAME)
canonical     = find_asset_names(category_id, ASSETS)
series_by_asset = get_competitor_series(category_id, canonical, START_DATE, END_DATE)

# Print a labeled table: one row per asset, one column per day.
print(f"Daily Visibility Score · {START_DATE} → {END_DATE} (exclusive)\n")
dates = [d for d, _ in next(iter(series_by_asset.values()))]
print(f"{'Asset':<14} " + "  ".join(f"{d:>10}" for d in dates))
for asset, points in series_by_asset.items():
    cells = "  ".join(f"{s:>10.1%}" for _, s in points)
    print(f"{asset:<14} {cells}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "dimensions": ["date", "asset_name"],
    "date_interval": "day",
    "filters": [
      {"field": "asset_name", "operator": "in", "value": ["<your-asset>", "<competitor-1>", "<competitor-2>", "<competitor-3>"]}
    ]
  }'
```

## Picking which competitors to plot

To avoid hardcoding asset names, fetch the [top-N
leaderboard](/cookbook/visibility/leaderboard) first, take the top 5 by
score, and feed those names into the `in` filter on this call.


# Headline score and the daily series, together
Source: https://docs.tryprofound.com/cookbook/visibility/headline-and-daily

Fetch both the period score and the daily line for one asset, and understand why they're different calls

A real dashboard tile shows two numbers: the **headline period score** (one
big number) and the **daily series** (the line). These come from **two
separate calls** to the same endpoint. This recipe runs both side-by-side
and explains how each is extracted and why you can't derive one from the
other.

> [!WARNING]
>
> **Don't average daily rows to derive a period score**: they're computed
> differently. See [Conventions & gotchas](/cookbook/setup/conventions#dont-average-daily-rows-to-get-a-period-score).

## How this example works

1. **Headline call** has no `date` dimension. The API returns a single
   traffic-weighted score for the whole window.
2. **Series call** adds `date` to dimensions and sets `date_interval`. The
   API returns one row per day, each containing that day's unweighted rate.
3. **Don't average the daily values to compute the headline.** Days with
   more prompts count more in the headline; the daily mean treats every day
   equally. The two are computed differently on purpose.

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
ASSET_NAME    = "<your-asset-name>"
START_DATE    = "2026-05-05"
END_DATE      = "2026-05-12"   # exclusive — returns data through 2026-05-11


def get_headline_score(category_id, asset_name, start_date, end_date):
    """Traffic-weighted period score (no date dim)."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start_date,
        end_date=end_date,
        metrics=["visibility_score"],
        filters=[{"field": "asset_name", "operator": "is", "value": asset_name}],
    )
    order = res.info.query["metrics"]
    return res.data[0].metrics[order.index("visibility_score")]


def get_daily_series(category_id, asset_name, start_date, end_date):
    """One (date, score) point per day — for the line chart."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start_date,
        end_date=end_date,
        metrics=["visibility_score"],
        dimensions=["date"],
        date_interval="day",
        filters=[{"field": "asset_name", "operator": "is", "value": asset_name}],
    )
    m_order = res.info.query["metrics"]
    d_order = res.info.query["dimensions"]
    i_score = m_order.index("visibility_score")
    i_date  = d_order.index("date")
    points = [
        (row.dimensions[i_date], row.metrics[i_score])
        for row in res.data
    ]
    return sorted(points, key=lambda p: p[0])


# Helpers — translate human-readable names into the IDs the report API needs.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


def find_asset_name(category_id, name):
    """Return the canonical asset name (case-insensitive) inside the category."""
    for a in client.organizations.categories.assets(category_id):
        if a.name.lower() == name.lower():
            return a.name
    raise ValueError(f"No asset named {name!r} in this category")


# Resolve names → IDs, then run both calls.
category_id = find_category_id(CATEGORY_NAME)
asset_name  = find_asset_name(category_id, ASSET_NAME)

headline_score = get_headline_score(category_id, asset_name, START_DATE, END_DATE)
daily          = get_daily_series(category_id, asset_name, START_DATE, END_DATE)

print(f"Headline: {headline_score:.1%}\n")
print("Daily:")
for date, score in daily:
    print(f"  {date}  {score:.1%}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

# 1) Headline (no date dimension)
curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "filters": [{"field": "asset_name", "operator": "is", "value": "<your-asset-name>"}]
  }'

# 2) Daily series (add date dimension + date_interval)
curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "dimensions": ["date"],
    "date_interval": "day",
    "filters": [{"field": "asset_name", "operator": "is", "value": "<your-asset-name>"}]
  }'
```

## How the two responses differ

| | Headline call | Series call |
| - | - | - |
| `dimensions` | `[]` | `["date"]` |
| `date_interval` | not sent | `"day"` (or week/month) |
| Rows returned | 1 | N (one per bucket) |
| Score per row | Traffic-weighted across the window | Unweighted within that day |
| Used for | The big KPI number | The line on the chart |

## Why you can't average the daily values

The headline weights days by their prompt volume: a day with 10,000
prompts moves the headline more than a day with 100. Averaging daily rates
treats those two days as equal. For most assets the gap is 1-3 percentage
points; for assets with bursty traffic it can be more. Always use the
headline call for the headline.


# Top-N leaderboard
Source: https://docs.tryprofound.com/cookbook/visibility/leaderboard

Rank every asset in a category by any visibility metric

The leaderboard on the Visibility tile is the same aggregate call as the
headline, without the asset filter. The API returns one row per asset, and
you sort client-side.

## How this example works

1. **Add `asset_name` to dimensions, drop the filter.** Now every asset gets
   its own row.
2. **Sort client-side.** By default the API sorts rows by the first metric
   descending (you can change this with `order_by`); re-sorting in your
   code keeps the direction explicit. Use `reverse=True` for "higher is better" metrics
   like `visibility_score`, and `reverse=False` for `average_position`
   (lower rank = better).
3. **`pagination.limit`.** The default is 10,000. Set it higher (max 50,000)
   if your category has many assets.

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
START_DATE    = "2026-05-05"
END_DATE      = "2026-05-12"   # exclusive — returns data through 2026-05-11
METRIC        = "visibility_score"   # or share_of_voice, average_position, ...
TOP_N         = 10


def get_leaderboard(category_id, metric, start_date, end_date):
    """All assets in the category ranked by the chosen metric."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start_date,
        end_date=end_date,
        metrics=[metric],
        dimensions=["asset_name"],
        pagination={"limit": 50000, "offset": 0},
    )
    m_order = res.info.query["metrics"]
    d_order = res.info.query["dimensions"]
    i_metric = m_order.index(metric)
    i_asset  = d_order.index("asset_name")

    # Lower-is-better for average_position; higher-is-better for everything else.
    reverse = metric != "average_position"
    return sorted(
        [(row.dimensions[i_asset], row.metrics[i_metric]) for row in res.data],
        key=lambda x: x[1],
        reverse=reverse,
    )


# Helper — translate the category name to the UUID the report API expects.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


# Resolve name → ID, then run.
category_id = find_category_id(CATEGORY_NAME)
leaderboard = get_leaderboard(category_id, METRIC, START_DATE, END_DATE)

for rank, (name, value) in enumerate(leaderboard[:TOP_N], start=1):
    print(f"{rank:>2}. {name:<24} {value:.1%}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "dimensions": ["asset_name"],
    "pagination": {"limit": 50000, "offset": 0}
  }'
```

## Highlighting your owned assets

To mark your own assets in the leaderboard (the "Owned" tag in the Profound
UI), join the response against the
[List your owned assets](/cookbook/setup/list-owned-assets) recipe: it
returns the set of asset names flagged `is_owned` in this category.


# Segment Visibility by model, region, or persona
Source: https://docs.tryprofound.com/cookbook/visibility/segment-by-model

Break a single asset's Visibility Score down by which AI model, region, or persona generated the answer

The Profound app's **Platforms** tab ("Visibility Score by Platform") shows
the same metric split by which AI surface served the answer. Same pattern works for region and
persona. Add the segmentation field to `dimensions` (or use it as a
`filter` to slice to one value).

## How this example works

1. **Add the segmentation field to `dimensions`** along with `asset_name`.
   You get one row per (segment, asset).
2. **Filter to your asset** so the response is one row per segment.
3. **Sort by score** to rank the segments.

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
ASSET_NAME    = "<your-asset-name>"
START_DATE    = "2026-05-05"
END_DATE      = "2026-05-12"
SEGMENT       = "model"   # or "region", "persona"


def get_visibility_by_segment(category_id, asset_name, segment, start_date, end_date):
    """Visibility Score for one asset, broken down by `segment` (model/region/persona)."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start_date,
        end_date=end_date,
        metrics=["visibility_score"],
        dimensions=[segment],
        filters=[{"field": "asset_name", "operator": "is", "value": asset_name}],
    )
    m_order = res.info.query["metrics"]
    d_order = res.info.query["dimensions"]
    i_score   = m_order.index("visibility_score")
    i_segment = d_order.index(segment)
    return sorted(
        [(row.dimensions[i_segment], row.metrics[i_score]) for row in res.data],
        key=lambda x: x[1],
        reverse=True,
    )


# Helpers — translate human-readable names into the IDs the report API needs.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


def find_asset_name(category_id, name):
    """Return the canonical asset name (case-insensitive) inside the category."""
    for a in client.organizations.categories.assets(category_id):
        if a.name.lower() == name.lower():
            return a.name
    raise ValueError(f"No asset named {name!r} in this category")


# Resolve names → IDs, then run.
category_id = find_category_id(CATEGORY_NAME)
asset_name  = find_asset_name(category_id, ASSET_NAME)

rows = get_visibility_by_segment(category_id, asset_name, SEGMENT, START_DATE, END_DATE)
print(f"{asset_name} Visibility Score by {SEGMENT}:")
for segment, score in rows:
    print(f"  {segment:<24} {score:.1%}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "dimensions": ["model"],
    "filters": [{"field": "asset_name", "operator": "is", "value": "<your-asset-name>"}]
  }'
```

## Slicing to a single segment instead

If you only want the score for one model (say, "where do I stand on
ChatGPT specifically?"), drop `model` from `dimensions` and filter on
`model_id` instead. The value is the model's UUID from
`GET /v1/org/models`:

```python theme={null}
filters=[
    {"field": "asset_name", "operator": "is", "value": "<your-asset-name>"},
    {"field": "model_id",   "operator": "is", "value": "<model-uuid>"},
]
```

Now the response is one row: same shape as the [headline
recipe](/cookbook/visibility/asset-visibility-score), scoped to ChatGPT.


# Visibility over time
Source: https://docs.tryprofound.com/cookbook/visibility/visibility-over-time

Build the daily, weekly, or monthly Visibility line chart for one asset

The line chart on every Visibility tile in the Profound app is one call with
the `date` dimension added. Use `date_interval` to switch between day, week,
and month buckets.

> [!WARNING]
>
> **Don't average daily rows to derive a period score**: they're computed
> differently. See [Conventions & gotchas](/cookbook/setup/conventions#dont-average-daily-rows-to-get-a-period-score).

## How this example works

1. **Add `date` to dimensions and set `date_interval`** (`"day"` / `"week"`
   / `"month"`). One row per bucket. Different bucket sizes give different
   numbers. See [Headline + daily](/cookbook/visibility/headline-and-daily).
2. **Filter by asset** so each bucket is one row.
3. **Read positions from `info.query`** instead of hardcoding them.

```python Python theme={null}
import os
from profound import Profound

client = Profound(api_key=os.environ["PROFOUND_API_KEY"])

# What to fetch — replace with your own values.
CATEGORY_NAME = "<your-category-name>"
ASSET_NAME    = "<your-asset-name>"
START_DATE    = "2026-05-05"
END_DATE      = "2026-05-12"   # exclusive — returns data through 2026-05-11
DATE_INTERVAL = "day"          # or "week" / "month"


def get_visibility_over_time(category_id, asset_name, start_date, end_date, interval="day"):
    """Daily/weekly/monthly Visibility Score for one asset, sorted by date."""
    res = client.reports.visibility(
        category_id=category_id,
        start_date=start_date,
        end_date=end_date,
        metrics=["visibility_score"],
        dimensions=["date"],
        date_interval=interval,
        filters=[{"field": "asset_name", "operator": "is", "value": asset_name}],
    )
    m_order = res.info.query["metrics"]
    d_order = res.info.query["dimensions"]
    i_score = m_order.index("visibility_score")
    i_date  = d_order.index("date")

    points = [
        (row.dimensions[i_date], row.metrics[i_score])
        for row in res.data
    ]
    points.sort(key=lambda p: p[0])
    return points   # → [("2026-05-05", 0.78), ("2026-05-06", 0.81), ...]


# Helpers — translate human-readable names into the IDs the report API needs.

def find_category_id(name):
    """Return the UUID of the category whose name matches (case-insensitive)."""
    for c in client.organizations.categories.list():
        if c.name.lower() == name.lower():
            return c.id
    raise ValueError(f"No category named {name!r}")


def find_asset_name(category_id, name):
    """Return the canonical asset name (case-insensitive) inside the category."""
    for a in client.organizations.categories.assets(category_id):
        if a.name.lower() == name.lower():
            return a.name
    raise ValueError(f"No asset named {name!r} in this category")


# Resolve names → IDs, then run.
category_id = find_category_id(CATEGORY_NAME)
asset_name  = find_asset_name(category_id, ASSET_NAME)

points = get_visibility_over_time(category_id, asset_name, START_DATE, END_DATE, DATE_INTERVAL)
for date, score in points:
    print(f"{date}  {score:.1%}")
```

```bash curl theme={null}
export PROFOUND_API_KEY=your_api_key_here

curl -X POST "https://api.tryprofound.com/v1/reports/visibility" \
  -H "X-API-Key: $PROFOUND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "category_id": "your-category-uuid",
    "start_date": "2026-05-05",
    "end_date": "2026-05-12",
    "metrics": ["visibility_score"],
    "dimensions": ["date"],
    "date_interval": "day",
    "filters": [
      {"field": "asset_name", "operator": "is", "value": "<your-asset-name>"}
    ]
  }'
```

## Switching the bucket

Swap `date_interval` to roll the data up to coarser buckets:

| `date_interval` | What you get |
| - | - |
| `"day"` | One point per calendar day (Eastern Time) |
| `"week"` | One point per week, weeks starting on Sunday |
| `"month"` | One point per calendar month |

The API also accepts `"hour"`, `"quarter"`, `"year"`, and `"relative_week"`
(weeks counted from your `start_date`). The number of returned rows scales
with the bucket size for the same window.


# Get an agent
Source: https://docs.tryprofound.com/api-reference/agents/get-an-agent

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents/{agent_id}
Retrieve an agent and its schema details.

Agents can have both a live published version and a draft version with newer
unpublished changes. Use the `version` parameter to choose which state to return.



# Get an agent run
Source: https://docs.tryprofound.com/api-reference/agents/get-an-agent-run

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents/{agent_id}/runs/{run_id}
Retrieve the current status and result details for an agent run.



# List agents
Source: https://docs.tryprofound.com/api-reference/agents/list-agents

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents
List agents available to your organization.

Agent status reflects whether an agent has ever been published. `published`
agents have a live published version. `draft` agents have not been
published yet.



# Run an agent
Source: https://docs.tryprofound.com/api-reference/agents/run-an-agent

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/agents/{agent_id}/runs
Start a new run for an agent.

Runs always execute the agent's live published version, so the agent must be
published first with `POST /v1/agents/{agent_id}/publish`. Unpublished drafts
cannot be run.



# Optimization Analysis
Source: https://docs.tryprofound.com/api-reference/content/optimization-analysis

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/content/{asset_id}/optimization/{content_id}



# Optimization List
Source: https://docs.tryprofound.com/api-reference/content/optimization-list

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/content/{asset_id}/optimization



# Add Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/add-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/documents
Add a document to a knowledge base using JSON text or multipart file upload.



# Add Folder
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/add-folder

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/folders
Create an empty folder at the requested knowledge base path.



# Delete Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/delete-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed delete /v1/knowledge-bases/{knowledge_base_id}/documents
Delete an existing document from a knowledge base.



# Delete Folder
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/delete-folder

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed delete /v1/knowledge-bases/{knowledge_base_id}/folders
Delete a folder. With recursive=false, non-empty folders return 409 and no contents are deleted.



# List Knowledge Bases
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/list-knowledge-bases

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/knowledge-bases
List knowledge bases accessible to the API key.



# Search Knowledge Base
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/search-knowledge-base

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/search
Search a knowledge base and return matching snippets or pages.



# Update Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/update-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed put /v1/knowledge-bases/{knowledge_base_id}/documents
Overwrite a knowledge base document using JSON text or multipart file upload.



# Get Account Insights
Source: https://docs.tryprofound.com/api-reference/openai-ads/get-account-insights

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/ads/openai-ads/ad-account/insights
Get ad account insights for the organization's OpenAI Ads partner brand.

`aggregation_level=campaign` returns one row per campaign (with `campaign_id`
/ `campaign_name` and all metrics), so every campaign's insights come back in
a single call; `time_granularity=daily` gives per-day rows (e.g. daily spend).



# Create prompts
Source: https://docs.tryprofound.com/api-reference/organization/create-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/org/categories/{category_id}/prompts
Create one or more prompts in a category. Topics and tags are auto-created if referenced by name and not yet existing. Use dry_run to preview without persisting.



# Get Assets
Source: https://docs.tryprofound.com/api-reference/organization/get-assets

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/assets
Get the organization assets, one row per (asset, organization) pair.

An asset's category can belong to multiple organizations; one asset row is
emitted per owning org so no association is silently dropped.



# Get Categories
Source: https://docs.tryprofound.com/api-reference/organization/get-categories

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories
Get the organization categories, one row per (category, organization) pair.



# Get Category Assets
Source: https://docs.tryprofound.com/api-reference/organization/get-category-assets

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/assets



# Get Category Citation Categories
Source: https://docs.tryprofound.com/api-reference/organization/get-category-citation-categories

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/citation-categories
Get the citation categories for a category: the built-in buckets plus any custom categories.



# Get Category Citation Tags
Source: https://docs.tryprofound.com/api-reference/organization/get-category-citation-tags

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/citation-tags
Get the custom citation tags defined for a category.



# Get Category Personas
Source: https://docs.tryprofound.com/api-reference/organization/get-category-personas

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/personas



# Get Category Regions
Source: https://docs.tryprofound.com/api-reference/organization/get-category-regions

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/regions
Get the regions for a specific category.



# Get Category Tags
Source: https://docs.tryprofound.com/api-reference/organization/get-category-tags

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/tags
Get the tags for a specific category.



# Get Category Topics
Source: https://docs.tryprofound.com/api-reference/organization/get-category-topics

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/topics
Get the topics for a specific category.



# Get Domains
Source: https://docs.tryprofound.com/api-reference/organization/get-domains

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/domains
Get the organization domains.



# Get Models
Source: https://docs.tryprofound.com/api-reference/organization/get-models

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/models
Get the organization models.



# Get Personas
Source: https://docs.tryprofound.com/api-reference/organization/get-personas

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/personas
Get the organization personas, one row per (persona, organization) pair.

Same (item, org) fan-out as ``get_assets``: a persona's category can be
owned by multiple orgs, and each owning org gets its own row so no
association is silently dropped.



# Get Regions
Source: https://docs.tryprofound.com/api-reference/organization/get-regions

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/regions
Get the organization regions.



# List prompts
Source: https://docs.tryprofound.com/api-reference/organization/list-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/prompts
Retrieve prompts in a category with optional filtering by type, topic, tag, region, platform, or persona. Supports cursor-based pagination.



# Update prompt status
Source: https://docs.tryprofound.com/api-reference/organization/update-prompt-status

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed patch /v1/org/categories/{category_id}/prompts/status
Bulk-update the status of one or more prompts. Prompts already in the target status are skipped. Use dry_run to preview without persisting.

Status options:
- 'active': Prompts will run daily.
- 'disabled': Prompts will not run moving forward, but historical data is preserved.
- 'deleted': Prompts are deleted along with historical data



# Update prompts
Source: https://docs.tryprofound.com/api-reference/organization/update-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed patch /v1/org/categories/{category_id}/prompts
Update one or more existing prompts. Only provided fields are changed. Dimension fields (regions, platforms, personas, tags) replace the full set when provided. Use dry_run to preview without persisting.



# Get Answers
Source: https://docs.tryprofound.com/api-reference/prompts/get-answers

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/prompts/answers



# Get Bots Report V1
Source: https://docs.tryprofound.com/api-reference/reports/get-bots-report-v1

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/bots
Get bot traffic report from the daily aggregated materialized view.

This endpoint queries pre-aggregated daily bot data, making it efficient
for large date ranges and high-traffic sites.

Metrics:
- count: unique bot visits
- citations: unique citation events
- indexing: unique indexing events
- training: unique training events
- last_visit: most recent visit timestamp



# Get Bots Report V2
Source: https://docs.tryprofound.com/api-reference/reports/get-bots-report-v2

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v2/reports/bots
Get bot traffic report from the hourly aggregated materialized view (UTC-based).

Supports date_interval="hour", calendar intervals through "year", "quarter", and "relative_week".
When `view_id` is provided, the query is scoped to that domain segment's hosts and paths.

Metrics:
- count: unique bot visits
- citations: unique citation events (ai_assistant bot type)
- indexing: unique indexing events (index bot type)
- training: unique training events (ai_training bot type)
- last_visit: most recent visit timestamp

Dimensions:
- date, path, bot_name, bot_provider, bot_type



# Get Referrals Report V1
Source: https://docs.tryprofound.com/api-reference/reports/get-referrals-report-v1

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/referrals
Get referral traffic report from the daily aggregated materialized view.

This endpoint queries pre-aggregated daily referral data, making it efficient
for large date ranges and high-traffic sites.



# Get Referrals Report V2
Source: https://docs.tryprofound.com/api-reference/reports/get-referrals-report-v2

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v2/reports/referrals
Get referral traffic report from the hourly aggregated materialized view (UTC-based).

Supports date_interval="hour", calendar intervals through "year", "quarter", and "relative_week".
When `view_id` is provided, the query is scoped to that domain segment's hosts and paths.



# Query Citations
Source: https://docs.tryprofound.com/api-reference/reports/query-citations

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/citations
Get citations for a given category.

The ``mentioned`` filter supports ``is true`` and ``is false``. It uses the
latest page analysis available at or before ``end_date``; pages without an
analysis by then are excluded from both values. ``citation_share`` keeps all
otherwise eligible citations in its denominator when this filter is used.



# Query Fanouts
Source: https://docs.tryprofound.com/api-reference/reports/query-fanouts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/query-fanouts



# Query Sentiment
Source: https://docs.tryprofound.com/api-reference/reports/query-sentiment

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/sentiment
Get citations for a given category.



# Query Visibility
Source: https://docs.tryprofound.com/api-reference/reports/query-visibility

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/visibility
Query visibility report.



# Get an agent
Source: https://docs.tryprofound.com/api-reference/agents/get-an-agent

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents/{agent_id}
Retrieve an agent and its schema details.

Agents can have both a live published version and a draft version with newer
unpublished changes. Use the `version` parameter to choose which state to return.



# Get an agent run
Source: https://docs.tryprofound.com/api-reference/agents/get-an-agent-run

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents/{agent_id}/runs/{run_id}
Retrieve the current status and result details for an agent run.



# List agents
Source: https://docs.tryprofound.com/api-reference/agents/list-agents

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/agents
List agents available to your organization.

Agent status reflects whether an agent has ever been published. `published`
agents have a live published version. `draft` agents have not been
published yet.



# Run an agent
Source: https://docs.tryprofound.com/api-reference/agents/run-an-agent

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/agents/{agent_id}/runs
Start a new run for an agent.

Runs always execute the agent's live published version, so the agent must be
published first with `POST /v1/agents/{agent_id}/publish`. Unpublished drafts
cannot be run.



# Optimization Analysis
Source: https://docs.tryprofound.com/api-reference/content/optimization-analysis

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/content/{asset_id}/optimization/{content_id}



# Optimization List
Source: https://docs.tryprofound.com/api-reference/content/optimization-list

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/content/{asset_id}/optimization



# Add Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/add-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/documents
Add a document to a knowledge base using JSON text or multipart file upload.



# Add Folder
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/add-folder

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/folders
Create an empty folder at the requested knowledge base path.



# Delete Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/delete-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed delete /v1/knowledge-bases/{knowledge_base_id}/documents
Delete an existing document from a knowledge base.



# Delete Folder
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/delete-folder

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed delete /v1/knowledge-bases/{knowledge_base_id}/folders
Delete a folder. With recursive=false, non-empty folders return 409 and no contents are deleted.



# List Knowledge Bases
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/list-knowledge-bases

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/knowledge-bases
List knowledge bases accessible to the API key.



# Search Knowledge Base
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/search-knowledge-base

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/knowledge-bases/{knowledge_base_id}/search
Search a knowledge base and return matching snippets or pages.



# Update Document
Source: https://docs.tryprofound.com/api-reference/knowledge-bases/update-document

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed put /v1/knowledge-bases/{knowledge_base_id}/documents
Overwrite a knowledge base document using JSON text or multipart file upload.



# Get Account Insights
Source: https://docs.tryprofound.com/api-reference/openai-ads/get-account-insights

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/ads/openai-ads/ad-account/insights
Get ad account insights for the organization's OpenAI Ads partner brand.

`aggregation_level=campaign` returns one row per campaign (with `campaign_id`
/ `campaign_name` and all metrics), so every campaign's insights come back in
a single call; `time_granularity=daily` gives per-day rows (e.g. daily spend).



# Create prompts
Source: https://docs.tryprofound.com/api-reference/organization/create-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/org/categories/{category_id}/prompts
Create one or more prompts in a category. Topics and tags are auto-created if referenced by name and not yet existing. Use dry_run to preview without persisting.



# Get Assets
Source: https://docs.tryprofound.com/api-reference/organization/get-assets

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/assets
Get the organization assets, one row per (asset, organization) pair.

An asset's category can belong to multiple organizations; one asset row is
emitted per owning org so no association is silently dropped.



# Get Categories
Source: https://docs.tryprofound.com/api-reference/organization/get-categories

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories
Get the organization categories, one row per (category, organization) pair.



# Get Category Assets
Source: https://docs.tryprofound.com/api-reference/organization/get-category-assets

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/assets



# Get Category Citation Categories
Source: https://docs.tryprofound.com/api-reference/organization/get-category-citation-categories

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/citation-categories
Get the citation categories for a category: the built-in buckets plus any custom categories.



# Get Category Citation Tags
Source: https://docs.tryprofound.com/api-reference/organization/get-category-citation-tags

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/citation-tags
Get the custom citation tags defined for a category.



# Get Category Personas
Source: https://docs.tryprofound.com/api-reference/organization/get-category-personas

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/personas



# Get Category Regions
Source: https://docs.tryprofound.com/api-reference/organization/get-category-regions

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/regions
Get the regions for a specific category.



# Get Category Tags
Source: https://docs.tryprofound.com/api-reference/organization/get-category-tags

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/tags
Get the tags for a specific category.



# Get Category Topics
Source: https://docs.tryprofound.com/api-reference/organization/get-category-topics

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/topics
Get the topics for a specific category.



# Get Domains
Source: https://docs.tryprofound.com/api-reference/organization/get-domains

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/domains
Get the organization domains.



# Get Models
Source: https://docs.tryprofound.com/api-reference/organization/get-models

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/models
Get the organization models.



# Get Personas
Source: https://docs.tryprofound.com/api-reference/organization/get-personas

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/personas
Get the organization personas, one row per (persona, organization) pair.

Same (item, org) fan-out as ``get_assets``: a persona's category can be
owned by multiple orgs, and each owning org gets its own row so no
association is silently dropped.



# Get Regions
Source: https://docs.tryprofound.com/api-reference/organization/get-regions

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/regions
Get the organization regions.



# List prompts
Source: https://docs.tryprofound.com/api-reference/organization/list-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed get /v1/org/categories/{category_id}/prompts
Retrieve prompts in a category with optional filtering by type, topic, tag, region, platform, or persona. Supports cursor-based pagination.



# Update prompt status
Source: https://docs.tryprofound.com/api-reference/organization/update-prompt-status

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed patch /v1/org/categories/{category_id}/prompts/status
Bulk-update the status of one or more prompts. Prompts already in the target status are skipped. Use dry_run to preview without persisting.

Status options:
- 'active': Prompts will run daily.
- 'disabled': Prompts will not run moving forward, but historical data is preserved.
- 'deleted': Prompts are deleted along with historical data



# Update prompts
Source: https://docs.tryprofound.com/api-reference/organization/update-prompts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed patch /v1/org/categories/{category_id}/prompts
Update one or more existing prompts. Only provided fields are changed. Dimension fields (regions, platforms, personas, tags) replace the full set when provided. Use dry_run to preview without persisting.



# Get Answers
Source: https://docs.tryprofound.com/api-reference/prompts/get-answers

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/prompts/answers



# Get Bots Report V1
Source: https://docs.tryprofound.com/api-reference/reports/get-bots-report-v1

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/bots
Get bot traffic report from the daily aggregated materialized view.

This endpoint queries pre-aggregated daily bot data, making it efficient
for large date ranges and high-traffic sites.

Metrics:
- count: unique bot visits
- citations: unique citation events
- indexing: unique indexing events
- training: unique training events
- last_visit: most recent visit timestamp



# Get Bots Report V2
Source: https://docs.tryprofound.com/api-reference/reports/get-bots-report-v2

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v2/reports/bots
Get bot traffic report from the hourly aggregated materialized view (UTC-based).

Supports date_interval="hour", calendar intervals through "year", "quarter", and "relative_week".
When `view_id` is provided, the query is scoped to that domain segment's hosts and paths.

Metrics:
- count: unique bot visits
- citations: unique citation events (ai_assistant bot type)
- indexing: unique indexing events (index bot type)
- training: unique training events (ai_training bot type)
- last_visit: most recent visit timestamp

Dimensions:
- date, path, bot_name, bot_provider, bot_type



# Get Referrals Report V1
Source: https://docs.tryprofound.com/api-reference/reports/get-referrals-report-v1

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/referrals
Get referral traffic report from the daily aggregated materialized view.

This endpoint queries pre-aggregated daily referral data, making it efficient
for large date ranges and high-traffic sites.



# Get Referrals Report V2
Source: https://docs.tryprofound.com/api-reference/reports/get-referrals-report-v2

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v2/reports/referrals
Get referral traffic report from the hourly aggregated materialized view (UTC-based).

Supports date_interval="hour", calendar intervals through "year", "quarter", and "relative_week".
When `view_id` is provided, the query is scoped to that domain segment's hosts and paths.



# Query Citations
Source: https://docs.tryprofound.com/api-reference/reports/query-citations

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/citations
Get citations for a given category.

The ``mentioned`` filter supports ``is true`` and ``is false``. It uses the
latest page analysis available at or before ``end_date``; pages without an
analysis by then are excluded from both values. ``citation_share`` keeps all
otherwise eligible citations in its denominator when this filter is used.



# Query Fanouts
Source: https://docs.tryprofound.com/api-reference/reports/query-fanouts

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/query-fanouts



# Query Sentiment
Source: https://docs.tryprofound.com/api-reference/reports/query-sentiment

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/sentiment
Get citations for a given category.



# Query Visibility
Source: https://docs.tryprofound.com/api-reference/reports/query-visibility

https://registry.scalar.com/@profound/apis/external-api/latest/openapi.json?variant=processed post /v1/reports/visibility
Query visibility report.