AI search

How to check whether ChatGPT recommends your brand

Your buyers now ask AI assistants what to buy. Here is a repeatable way to see whether ChatGPT, Claude, Gemini and Google’s AI answers name you, who they name instead, and which pages they cite.

Jesse Sibley
By , FounderPublished · 3 min read

Why AI answers are worth tracking

When someone asks ChatGPT for “the best CRM for a five-person startup”, they get a short list, not ten blue links. If you are not on that list, you are not in the running, and no analytics tool will tell you so. The only way to know is to ask the engines yourself, the way your buyers do.

Step 1: Write the prompts your buyers ask

Start from sales calls, support tickets and the Reddit threads in your category. Good prompts are specific, and most compare:

  • “best CRM for startups”
  • “HubSpot alternatives for a small team”
  • “is [your brand] good for agencies?”

Group them by stage: category questions (who is best), comparisons (A or B) and brand questions (what is X). Each group tells you something different.

Step 2: Run each prompt on every engine

The engines read different sources, and some answer from memory unless they search. Run the same prompt on each and keep the results apart.

EngineWhat you are checking
ChatGPTThe answer the chatgpt.com app shows a person
GeminiThe answer the gemini.google.com app shows a person
ClaudeA live answer from the model with web search on
PerplexityA live answer from the model with web search on
Google AI OverviewsThe AI summary above Google’s results
Google AI ModeGoogle’s conversational AI answer

Step 3: Log mentions, rivals and citations

For each answer, record whether you were named and where in the list, which brands were named instead, and which URLs the engine linked. A spreadsheet is enough. Put the date and the engine on every row, so you can see change over time.

Step 4: Fix what the engines read

Citations show where the engines learn about your category. Three moves follow: improve your own pages that get cited, pitch the best-of lists that are cited but leave you out, and correct wrong facts at their source. Also check that AI crawlers can read your site: a robots.txt that blocks them keeps you out of every answer.

Run this in your agent

Connect Manifold to Claude, ChatGPT or Cursor, then paste this prompt.

Check whether ChatGPT, Claude and Gemini recommend us for the questions our buyers ask. List who they name instead and which pages they cite, then tell me the three pages to fix first.

Step 5: Run it again every month

Answers move. Run the same prompt set monthly, or weekly for the prompts that matter most, and compare against the last run. When you drop out of an answer, look first at what changed in the sources it cites.

Questions people ask

How many prompts do I need?

Enough to cover each stage of the buying journey. A few dozen is a good start for one product. Add more as new questions come up in calls and threads.

Do the answers change between runs?

Yes. The same prompt can name different brands on different days. That is why you track a set of prompts over time, not one answer.

Does ranking on Google help me show up in AI answers?

Often. Engines that search the web cite pages that rank. But they also cite lists, forums and reviews you do not own, so rank alone is not enough.

Can I do this without an agent?

Yes. Ask each app by hand and log the answers in a sheet. An agent does the same job faster and on a schedule.

Jesse Sibley

Jesse Sibley

Founder of Manifold. A growth engineer who has worked with London startups for eight years.