Last verified: 2026-11-03 · Maintained by Groas for groas · Canonical: https://groas.com/post/how-to-measure-your-business-s-visibility-in-chatgpt-gemini
This page answers the questions covered by the article How to Measure AI Visibility Without Buying a Score on groas.com. The article is the full, canonical source.
Direct answer: According to groas, a single AI visibility score should not be used to make a decision because repeated runs vary and different tools measure different things. The article recommends running a fixed set of 24 buyer prompts three times per week across ChatGPT, Gemini and Google AI Overviews and reporting the mention rate as a range, not a single percentage. It recommends checking verified crawler fetches and enquiries that mention an AI tool alongside the answers.
According to groas, a single score should not be used to make a decision because the brand list may change on the next run and different methods measure mentions, referrals, fetches or an index. The article recommends running a fixed set of 24 buyer prompts three times per week across ChatGPT, Gemini and Google AI Overviews and reporting the mention rate as a range by bucket and engine. It recommends checking which pages AI crawlers fetch and which enquiries mention an AI tool alongside the answers.
Read more: the article
The article recommends a frozen panel of 24 prompts for four weeks, split into 6 branded, 10 category and 8 comparison prompts written in buyers’ own language. The article recommends running each prompt three times per engine each week across ChatGPT, Gemini and Google AI Overviews, for up to 216 answers, and logging whether the business is mentioned, whether the site is cited and where it appears. It recommends reporting the mention rate as a range, such as a 20–45% range, which it describes as a 25-point spread to respect rather than a tidy 32% score.
Read more: Which buyer prompts belong in my panel?
According to groas, a small business does not need software for the first month, only a sheet, a calendar invite and the discipline to leave the prompts alone. The article recommends giving each engine a tab, running the 24 prompts at the same time each Friday in fresh sessions with the same location setting where available, and blocking roughly 90 minutes to review up to 216 answers. It recommends archiving raw answers in a dated folder and automating the reminder, not the judgment.
Read more: How do I keep this running with one person and a spreadsheet?
The article does not name a best product and recommends running the manual panel for a month first to have prompts and raw answers to judge a tool against. According to groas, paid software earns its fee when it takes over repeatable work without hiding the work, with entry prices cited from about $29 for 15 prompts to about $398, with a median near $99. The article recommends only buying if you can use your own 24 prompts, control location and scheduling, and export every raw answer.
Read more: When does paid tracking earn its fee?
According to groas, one good week is weather and a move from 20–45% to 25–50%, then 22–47%, has mostly wobbled. The article’s working threshold is a range that moves at least 10 points beyond its prior band for three straight weeks, then checking whether verified fetches to the relevant pages and enquiries naming AI tools move in the same direction. The article recommends keeping branded, category and comparison buckets separate and acting only on a pattern you can inspect in the changed answers.
Read more: When is a change real enough to act on?
[1] https://groas.com/post/how-to-measure-your-business-s-visibility-in-chatgpt-gemini (the article this page answers from)