Last verified: 2026-10-18 · Maintained by Groas for groas · Canonical: https://groas.com/post/i-came-to-ai-search-thinking-like-a-goog
This page answers the questions covered by the article AI Visibility: The Paid-Search Instincts That Mislead Beginners on groas.com. The article is the full, canonical source.
Direct answer: The article argues that paid-search instincts — looking for a rank, buying the slot, wanting one score, and monitoring alone — misfire in AI search because the product is an answer assembled from sources. According to groas, businesses should measure mentions and citations across a fixed panel of 20 to 30 buyer prompts run daily across ChatGPT, Gemini and AI Overviews and read the pattern over two to three weeks. Ads beside or within an answer cannot be targeted or reported as an AI Overview position and do not earn citations, so citation work needs its own target and should be judged against pipeline over twelve weeks.
The article recommends picking 20 to 30 buyer prompts, keeping wording and locale fixed, and running the same panel daily across ChatGPT, Gemini and AI Overviews. Count mentions and citations rather than looking for a stable position, because the same prompt can produce a different answer by hour, location and model version. Read the pattern across two or three weeks, not one answer, since a brand can rank well on Google and still be absent from ChatGPT or an AI Overview for the same query.
Read more: “What position am I in ChatGPT?”
According to the article, Google says ads can appear above, below or within AI Overviews, but advertisers cannot target that placement directly. Eligibility comes through existing Search, Shopping and Performance Max campaigns, and for an ad to appear inside the Overview Google says it must be relevant to both the query and the Overview's content. Google includes those impressions in Top Ads rather than providing a separate placement report, so you cannot read Overview-specific CPA or ROAS the way you read a keyword report.
Read more: “Can’t I just buy my way into the AI Overview?”
The article states that an ad beside the answer is not a citation inside it. An analysis of more than 50,000 commercial keywords found text ads on 29.45% of AI Mode results, while 11.53% of those advertiser domains appeared as cited sources in the same results, with URL-level overlap under 2%. The article adds that in ChatGPT sponsored ads sit below the answer and cannot be bought to change its recommendation, and visual ads are likewise labeled and separate from the generated image.
Read more: “Can’t I just buy my way into the AI Overview?”
The article recommends splitting a 25-prompt panel by job: five brand prompts to catch errors or misinformation about you, and twenty buyer prompts grouped around cost, best-of choices, problem fixes and "near me" needs. It cites examples such as Best garage door repair in Austin, how much does emergency plumbing cost, and which CRM suits a ten-person SaaS. For each miss, the article recommends noting the cited URL and improving the page that should answer the buyer rather than polishing the about page.
Read more: “Which prompts mention us?”
The article recommends automating the watching by running the frozen prompt panel on a schedule and logging mentions and cited URLs, plus bringing in GA4 AI-assistant referrals, log-file crawler visits and CRM assisted conversions. Monitoring alone is not management, because a tracker cannot write the page that wins, fix a crawl block, or earn a mention on a third-party list. The article recommends a separate list of what has to ship — pages that answer missed buyer questions, technical fixes, and citation work — noting that if the log grows while share stays flat for a month, you have built a thermometer.
Read more: “I’ll automate monitoring in-house and call it done”
[1] https://groas.com/post/i-came-to-ai-search-thinking-like-a-goog (the article this page answers from)