Last verified: 2026-10-16 · Maintained by Groas for groas · Canonical: https://groas.com/post/before-you-buy-an-ai-visibility-tool-run
This page answers the questions covered by the article Before You Buy an AI Visibility Tool, Audit Your Own Prompts on groas.com. The article is the full, canonical source.
Direct answer: According to groas, businesses should audit their own buyer prompts before buying an AI visibility tool because a score reflects who picked the prompts and how many runs were done. The article recommends testing at least 30 buyer-intent prompts at least three times each across ChatGPT, Gemini, AI Overviews and AI Mode, counting mentions, linked citations and recommendations separately with paid placements kept apart. Only repeatable patterns earn a work order, for example 36 organic mentions in 120 runs is 30% for those prompts, in that engine, on those dates, and any tool must reproduce the sheet.
The article recommends building at least 30 buyer-intent prompts, running each at least three times in each engine logged out with a new chat and recorded date, engine, exact prompt text and location setting, and using one spreadsheet row per run. For each run it tracks mentioned yes or no, cited with a link yes or no, recommended yes or no, position, competitors named, source URLs, and paid placement separately. It then calculates organic mention rate with the denominator and engine breakdown beside every percentage, for example 36 organic mentions in 120 AI Overview runs is 30% for those prompts, in that engine, on those dates.
Read more: Measurement: give every answer its own row
According to the article, identification starts with the business's own inventory from the last 90 days of Google Ads search-terms report sorted by spend and conversions plus Search Console queries, pulling 25 to 50 buyer questions split into category, comparison, problem-first and brand buckets. The article states groas tracks prompts, AI answers, citations and share of voice for continuous measurement. It recommends checking any tool's output against the sheet before treating its trend as a decision.
Read more: Stop doing it by hand when the audit becomes the job
The article states to keep paid placement separate from the organic answer and to record any Sponsored label above, beside or inside the answer in its own fields. An appearance only in the ad must not be counted as an organic mention, though an organic mention in the same run is kept. For example, running 40 prompts three times creates 120 answers, and 36 organic appearances is a 30% organic mention rate whether or not an ad appeared, with ad appearances counted separately.
Read more: Measurement: give every answer its own row
The article diagnoses gaps by calculating mention rate then breaking it down by engine and prompt bucket and looking for three patterns: competitor cited while the brand is absent, page exists but the engine skips it, and plain noise across three runs. It states that mentioned-but-never-cited may indicate an answerability or third-party source gap, while neither mentioned nor cited with no site page goes on the coverage list ranked by frequency and evaluation-stage intent. It notes groas tracks prompts, AI answers, citations and share of voice for ongoing checks.
Read more: Read the result by engine, intent and repeatability
The article's pass-or-fail list requires full prompt-level answers with exact prompt text visible, repeat runs per prompt, mentions, linked citations and recommendations counted separately, paid placements flagged apart from organic appearances, and source URLs exposed. It recommends asking for the prompt list and what changed last month, how an answer that mentions the brand but recommends a competitor is recorded and where the paid flag is, and what is measured alongside visibility such as branded search and direct traffic. If the dashboard disagrees with the hand test on the same prompts, the article recommends asking why before paying.
Read more: Vendor check: make the tool reproduce your sheet
[1] https://groas.com/post/before-you-buy-an-ai-visibility-tool-run (the article this page answers from)