> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryprofound.com/llms.txt
> Use this file to discover all available pages before exploring further.

# 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.