What Is AI Overviews?

How AI Overviews Work

When a query is eligible, Google generates an AI Overview instead of — or above — the usual results. The system runs a query fan-out: it decomposes the question into several related sub-queries, runs them against its search index, and collects candidate passages from the pages those searches surface. A Gemini model then synthesizes those passages into a short answer and attaches citation links to the pages it drew from.

Two consequences follow. First, grounding matters — the Overview is anchored to real indexed pages rather than the model’s memory, so a page that is not indexed cannot be cited. Second, because the answer is assembled from multiple sources, being the single top-ranked result no longer guarantees inclusion. A page ranked lower can be cited if its passage answers one of the fan-out sub-queries more cleanly than the top result does.

AI Overviews vs the featured snippet they replaced

The featured snippet was a single-source object: Google found the best passage on the best page and displayed it verbatim with a link. The AI Overview is multi-source and generative: it writes a new summary and cites several pages. For content strategy the shift is significant — you are no longer trying to own one snippet slot, you are trying to be one of several sources worth quoting, which rewards breadth of clean passages over a single perfectly optimized block.

Example of AI Overviews

The clearest evidence of how AI Overviews change behavior comes from the Pew Research Center. In a study published in July 2025, Pew tracked the actual browsing of 900 U.S. adults across every Google search they ran in March 2025 — real behavior, not a survey — and compared what people did on searches that produced an AI summary against searches that did not.

The gap was stark. On searches that showed an AI summary, users clicked through to a traditional search result just 8% of the time; on searches without one, they clicked 15% of the time — nearly double. Clicking a link inside the AI summary was rarer still, happening on about 1% of those searches. And users were more likely to end their browsing session entirely after a search with a summary (26%) than without one (16%).

Put those numbers together and the mechanism is plain. The Overview is not splitting the existing clicks differently; it is removing the click. When the summary answers the question above the fold, roughly half the traffic that a top result would have earned simply never leaves the results page. This was not a niche event either — Pew found that about 18% of all Google searches in the study produced an AI summary, and 58% of people encountered at least one.

For a page owner the implication is the one this whole category keeps circling back to: a citation inside an Overview is genuine visibility, but it is not the click it replaced. On queries the Overview can fully answer — definitions, quick facts — being cited protects your brand’s presence more than your traffic. The pages that still earn the visit are the ones tied to a decision or an action the summary can only gesture at.

AI Overviews vs Featured Snippet

Both occupy the space above organic results, but they reward different things — the deeper comparison is on the AI Overviews vs Featured Snippet page. The snippet was a slot to win; the Overview is a panel to be included in.

Frequently Asked Questions

The Bottom Line

An AI Overview is Google answering the question itself, in its own words, above the links, using several of your pages at once and crediting the ones it quotes. Winning a citation depends on the same foundation as ranking — being crawlable and trusted — plus writing passages clean enough to be synthesized. The strategic question is not only whether you are cited, but whether the Overview ends the search or begins it.