Find Out How Often ChatGPT Names You (Before Your Buyer Shortlists Someone Else)

Most teams check AI search once, screenshot the answer and call it a measurement. One answer is a roll of the dice. The same list of brands came back less than 1 time in 100 when SparkToro asked the same questions thousands of times.

This is the visibility check from the AI Search 101 session. It is a Claude folder that takes your 3 real buyer questions, the answers you save from 10 fresh ChatGPT chats each, and turns them into a count, three piles and a fix list. Claude counts what ChatGPT said, answer by answer, and never guesses.

Each question takes about 10 minutes of ChatGPT runs, and Claude counts the lot in about 5 minutes. Rerun it next month with the same questions and compare the two tables.

What's in the download

• A CLAUDE.md that holds Claude to one rule, every number comes from a saved answer

• Three prompts that count who got named, sort each question into one of three piles, and write the fix list

• A questions file for your 3 buyer questions, tagged research or buying

• A facts file for your real prices, so the fix list flags anything ChatGPT gets wrong about you

• A pages file for your own URLs, so the fix list never invents a page

• Templates for the results table and the fix list, and one example answer file showing the format

• Build steps that follow the session step for step, and a prerequisites page

Who built it

Will Leatherman runs Catalyst. His team has helped 150 B2B companies build more than $14M in pipeline from content, and Catalyst runs a free AEO audit that checks what ChatGPT, Claude, Gemini and Perplexity say about a company.