(GEO·AI Search)

AI Overview

A top-of-page summary that changes how search traffic flows

AI Overview is a Google feature that uses generative AI to write an answer to a search query and shows it at the top of the search results page (SERP) . Google Search Central’s documentation describes it as a summary that “helps people get the gist of a complex topic or question more quickly,” and says it shows links to the web pages used as sources along with the answer. Google (official)

This top placement changes the flow of clicks. When Ahrefs analyzed 300,000 keywords in 2025, the average click-through rate of the #1 page was 34.5% lower for keywords that showed an AI Overview than for those that didn’t. Ahrefs (field data) When Pew Research Center analyzed the actual browsing of 900 US adults in 2025, users clicked a regular result link on 8% of visits to searches showing an AI summary, about half the rate without one (15%), and clicked a source link inside the summary on just 1% of visits. Pew survey

But Google’s official explanation of the click impact points in a different direction from these measurements. Side by side, the two positions look like this.

Google’s official positionThere are no extra optimization requirements to appear in AI Overviews, and links included in the summary get more clicks than if the same page had appeared as a traditional web listing (supporting figures not disclosed). Google (official)
Expert field dataWhen an AI Overview appears, organic clicks drop sharply. #1 page CTR −34.5% (Ahrefs, 300,000 keywords); regular link clicks 15%→8% (Pew, 900 adults). The appearance itself reduces clicks. Ahrefs (field data)

The difference comes down to perspective. Google talks about the quality of the links that remain inside the Overview, while independent studies look at the volume of clicks lost across the whole search page. Google hasn’t published supporting figures, and field studies have repeatedly confirmed the drop with large samples. In practice, assume by default that “the appearance itself can reduce clicks,” and it’s safer to judge citation visibility as brand reach rather than clicks.

In the end, AI Overview is at the center of the zero-click search debate, and being cited as a source inside the summary has itself become a new visibility goal. The GEO study, in which researchers at Princeton and IIT Delhi tested generative engines with 10,000 queries, reported that the “Cite Sources” technique raised the visibility of the site ranked #5 on the SERP by 115.1% while lowering the #1 site’s by 30.3% on average, so citations level the field in favor of lower-ranked sites. GEO paper How to earn these citations is covered separately in the GEO entry.

How it’s generated – a customized Gemini and query fan-out

In May 2024, Google announced on its official blog that AI Overviews run on “a new Gemini model customized for Google Search.” Google (official) Google Search Central’s documentation describes the generation process: it “may use a ‘query fan-out’ technique – issuing multiple related searches across subtopics and data sources.” Google (official) Putting the two documents together, the process works like this.

  1. It splits the user’s question into several sub-queries and searches each one (query fan-out).
  2. While generating the answer, it identifies the web pages that support it.
  3. It shows the generated summary with source links to those pages.

The Google Search Central blog (May 2025) states that “Google’s AI features use our core ranking systems.” Google (official) In other words, AI Overviews also run on the existing search pipeline of crawling → indexing → ranking, and pages that aren’t indexed can’t become citation candidates.

The correlation between citations and rankings is changing. When Ahrefs analyzed 1.9 million AI Overview citations in 2025, 76.1% of cited pages were in the top 10 for that keyword, and the median rank of the first citation was #2. Ahrefs (field data) But in a follow-up analysis expanded to 863,000 SERPs, this share fell to 38%. Ahrefs (field data) Ahrefs interprets this as documents that appear for fan-out sub-queries being cited more often than the top results for the original query. Ranking high is still an advantage, but #1 doesn’t guarantee a citation, and a lower rank doesn’t rule one out.

Getting cited – checkpoints based on Google’s official guidance

Google Search Central’s documentation says “there are no additional requirements to appear in AI Overviews,” and the Search Central blog stresses that success in AI experiences “starts with the same fundamentals that help you rank in traditional search results.” Google (official) Turning the same blog’s advice into a production checklist gives the following.

  • Unique, valuable content for people – Google advises creating “helpful, reliable, people-first content.” EEAT This is in line with the document’s quality signals.
  • A good page experience – the Search Central blog explicitly names this as something that also works in AI experiences.
  • Crawling and indexing accessibility – check whether Googlebot is blocked and check index status. Pages that aren’t indexed can’t be cited.
  • Review your snippet preview controls – Search Central’s documentation says you can control what’s shown in AI features with nosnippet / data-nosnippet / max-snippet / noindex.
  • Match structured data to visible content – the Search Central blog recommends that structured data match the content actually shown on the page.

At the markup level, review your snippet control settings. Google’s robots meta tag documentation states that the nosnippet rule “applies to all forms of search results (at Google: web search, Google Images, Discover, AI Overviews, AI Mode) and will also prevent the content from being used as a direct input for AI Overviews and AI Mode.” Google (official) If you limit max-snippet too much, you also reduce the chance of being excerpted.

<!-- Remove the excerpt length limit - shown at the length Google judges effective -->
<meta name="robots" content="max-snippet:-1, max-image-preview:large">

<!-- Exclude only a specific area from snippets and AI citations -->
<div data-nosnippet>Information that changes often, like prices, and shouldn't be summarized</div>

Common mistakes

  • Trying to block AI Overviews by blocking Google-Extended – Google’s crawler documentation states that Google-Extended is for controlling model training and grounding for Gemini Apps and Vertex AI, and “does not impact a site’s inclusion in Google Search nor is it used as a ranking signal.” Google (official) Appearing in AI Overviews follows regular Googlebot crawling and your snippet control settings.
  • Applying nosnippet everywhere to avoid being excerpted – according to Google’s docs, nosnippet applies to all forms of search results, so not only AI Overviews but your regular search snippets disappear too.
  • Focusing only on AI while skipping basic SEO – AI features use the core ranking systems, so pages that aren’t crawled and indexed can’t be cited in the first place.
  • The illusion that “#1 guarantees a citation” – in Ahrefs’ follow-up analysis, the share of citations going to top-10 pages was 38%. Ahrefs (field data) Ranking and citation are only correlated, and visibility for fan-out sub-queries is becoming more and more important.

Further reading

References

Jul.2026By Studio JT

More from GEO·AI Search

Like this term?
Give it a like!

(35)