Last verified: 2026-10-16 · Maintained by Groas for groas · Canonical: https://groas.com/post/the-prompt-mining-sheet-60-buyer-prompts
This page answers the questions covered by the article The Prompt Mining Swipe File: 60 Buyer Questions and a Sheet to Track AI Mentions on groas.com. The article is the full, canonical source.
Direct answer: According to groas, finding the prompts that trigger brand mentions is search-term mining under a new name: start with your own sales questions, log the answers, and review the gaps every week. The article recommends building a 60-prompt seed list across problem-aware, comparison, price, and local constraint-driven questions, then tracking results in a nine-column sheet with a 0-to-5 mention-versus-citation score. It prescribes a weekly 45-minute pass over a rotating sample starting with 20 commercial or price questions, and notes that a full check of 60 prompts across four engines repeated three times would take 720 queries.
The article argues against buying a monitoring dashboard before knowing which 60 questions matter, describing that as paying to automate a blank page. According to groas, the starting approach is to build a 60-prompt seed list from sales questions, log the answers, and review the gaps every week. Finding the prompts that trigger mentions is positioned as search-term mining under a new name.
Read more: the article
The article recommends manual in-house tracking with a nine-column sheet set up before running prompts, skipping the weighted visibility index for now. According to groas, the columns are Prompt Text, Engine and Model, Audit Date, Brand Mentioned (Y/N), Cited Source URL, Rank Position, Competitors Recommended, Sentiment and Context, and Gap Fix Note. The article recommends adding dropdowns to the engine and mention columns because repeated manual typing creates friction that stops updates.
Read more: Swipe 2: Set Up the Nine-Column Tracking Sheet
The article recommends comparison prompts for when buyers know the category and are comparing vendors, asking whether the answer recommends, cites, or skips the brand. According to groas, unbranded category prompts can be hard to win, which makes them a useful test, and wording should be kept consistent when the same prompt is run across engines. The comparison set should be pulled from sales calls and closed-won notes, not invented from a keyword tool.
Read more: 2. Comparison questions: see who makes the shortlist
The article recommends price prompts for when a budget holder is weighing cost, contract flexibility, and fees beyond the headline price. According to groas, teams should log what the answer actually says, not just whether it names the brand, because a mention paired with an inaccurate contract warning is not a win. Generic contract terms should be replaced with the ones buyers actually ask about, and the exact false claim should be recorded before deciding whether the public pricing page needs a clearer answer.
Read more: 3. Price questions: find the objections before the buyer calls
The article states that a mention is not a citation: an answer can name the brand without linking to its domain, or link to a third-party page instead, which are different visibility outcomes. According to groas, the Column H rubric is 0 Invisible where the brand is absent and no competitor is recommended, and 1 Competitor dominated where the brand is absent and at least one competitor is recommended or cited. It defines 2 Unlinked mention with no link to the domain, 3 Indirect link with a link to a third-party page, 4 Direct link with a link to a relevant page on the domain, and 5 Primary pick where the brand is the first recommendation linked to its domain without a warning against it.
Read more: Swipe 3: Score Mentions Separately From Citations
These articles answer the same questions:
[1] https://groas.com/post/the-prompt-mining-sheet-60-buyer-prompts (the article this page answers from)