Skip to main content
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.
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. Putting your name in the question inflates your Visibility Score. Use it only in validation and FactCheck 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 include broad prompts for each business line. Add two to five depth topics 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.

​
Ask AI Marketer

Use AI Marketer (Aim) 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.

​
Add tags to your prompts

Tags can span topics, and one prompt can have several. Use these tag examples and rename them to match how you work.
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, 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 and read what it returns. A discovery prompt should list brands, a comparison should pick one, and a validation should say yes or no. Learn more about different types of prompts in Profound’s AEO guide.
  • Tag every prompt you add. Learn more in Categorize Your Prompts With Tags.

​
Next steps