Google AI Overviews now sit above the results for a large share of searches, and AI Mode extends the same idea into a full conversational surface. Both answer the question directly and cite a handful of sources. Being one of those sources is a different objective from ranking, though on Google specifically the two are more closely related than they are anywhere else.
Organic ranking still matters here, unlike elsewhere
This is the most useful thing to understand about AI Overviews, and it separates Google from other AI surfaces.
Analyses of AI Overview citations have consistently found heavy overlap with conventional organic results: roughly half of cited links come from the top 20 organic results, and the overwhelming majority of Overviews cite at least one page from the top 20. Google AI Mode shows a similar pattern.
ChatGPT behaves quite differently, drawing a much smaller share of its citations from Google's top results and applying its own selection layer on top. So the practical implication is specific: conventional SEO remains a meaningful input for Google's AI surfaces, and a much weaker one for ChatGPT. A single "AI visibility" tactic applied identically across platforms misunderstands how differently they source.
What appears to influence selection
Beyond ranking, the pages that get pulled into an Overview tend to share characteristics. They answer a specific question directly rather than burying the answer in preamble. They are structured so a passage can be lifted without losing its meaning. They contain concrete, checkable specifics — numbers, named entities, dates — rather than general assertion.
Content that requires the whole page as context to make sense is harder to cite than content where a single paragraph stands on its own.
Question-shaped pages do better
Overviews are triggered by questions, and pages organised around real questions map onto that more naturally. This is why the FAQ format keeps being recommended — not because the markup is magic, but because the format forces the writing into question-and-answer pairs that are straightforward to extract.
The questions have to be ones people actually ask, phrased the way they actually phrase them. Invented questions written to hold keywords do not match real queries.
The traffic reality, stated plainly
It would be dishonest to present AI Overviews purely as an opportunity. Publishers have reported substantial declines in click-through from results carrying an Overview — reporting around antitrust filings has cited figures in the region of a 58% drop — and Google has faced sustained criticism over the effect on the sites its answers are built from. Google has since shipped changes intended to send more traffic back, including additional exploration links and clearer source context.
For most brands the conclusion is not that Overviews should be resisted, but that the objective has shifted. Being named in the answer carries value even when the click does not follow, and measuring only sessions will understate what is happening.
What to do
- Keep conventional SEO in good order — on Google's AI surfaces it still feeds selection.
- Answer real questions directly and early on the page, in extractable passages.
- Include specifics that can be verified: figures, named entities, dates.
- Keep facts about your company consistent everywhere they appear.
- Track brand mentions in answers, not only clicks, or you will misread the result.
What nobody can sell you
There is no submission process for AI Overviews, no placement to buy, and no agency that controls whether Google cites a given page. Anyone offering guaranteed inclusion is describing something that does not exist. What can be influenced is whether the material Google draws on is clear, verifiable and corroborated.
See our AI visibility service, or how ranking on Google differs from being cited by ChatGPT.