AI Overviews are generated per-query from a set of retrieved pages, and the sources chosen frequently sit outside the classic organic top 10. To be pulled in, you need crawl access for Google-Extended, content that answers a specific question in a clean extractable passage, accurate structured data, and corroborating signals off your own site. Ranking well helps but is neither necessary nor sufficient.
An AI Overview is a generated summary placed above the traditional results, with a small set of source links attached. When one appears, it changes user behaviour substantially — the summary answers the question, and the links below it get considerably fewer clicks. Zero-click searches on Google have risen to roughly 69% (Similarweb).
For most businesses this reads as a slow, unexplained decline in organic traffic even while rankings hold steady. The rankings didn't move; the clicks did.
Studies of AI Overview citations consistently find that a large majority of cited URLs are not in the organic top 10 for that query. Google is retrieving and ranking passages for the sub-questions its model generates, not simply summarizing the page-one results.
The practical consequence: a page sitting at position 14 that answers one specific sub-question extremely well can be cited in an Overview while the page ranked #1 is not. If you have been told you must rank first to appear, that is not what the data shows.
Google-Extended is the crawler control that governs whether your content can be used for Gemini and grounding AI experiences. If it's disallowed in robots.txt — sometimes set years ago as a blanket AI opt-out — you have excluded yourself. Check it first; it is the single most common own-goal we find.
Google fans your query out into related questions. Content organised as a set of clearly-headed, self-contained answers gives it many more surfaces to grab than one long essay. In practice:
The Princeton GEO study measured a 41% visibility lift from adding relevant statistics and a 115% lift for lower-ranked pages that cite credible sources. Quantified, attributed claims are the raw material a summarizer prefers, because they are the safest thing to repeat.
Schema doesn't force a citation, but it removes ambiguity about what a page is and who published it. Organization, Person, Service, FAQPage and Breadcrumb markup that accurately reflects visible content is the baseline. Markup that contradicts the page is worse than none.
Google's systems have always weighed what the wider web says about an entity, and the generated layer is no exception. Ahrefs found brand mentions correlate with AI visibility about three times more strongly than backlinks. Consistent business information across directories, reviews and press is doing more work here than any on-page tweak.
Google Search Console does not break out AI Overview appearances separately, so measurement takes a bit of craft:
Our full method is in how to measure AI visibility.
Book a free GEO audit: where you appear across 6+ AI engines today, the prompts competitors own, and a prioritized 90-day plan you keep either way.