AI Overviews optimization: what Google says vs the data

Between July 2025 and April 2026, five organizations published data on what Google’s AI Overviews do to clicks. One found that adding an AI Overview did not raise zero-click rates at all. Another found a 58% lower click-through rate for the top result. Google says total organic clicks have been relatively stable. They are not all wrong; they measured different things.

This post sets out what Google says about AI Overviews optimization, compares the five studies side by side with their methods, and ends with the steps the evidence supports. Every figure links to its source.

What does Google say it takes to rank in AI Overviews?

Google says there is no separate optimization for AI Overviews. A page must be indexed and eligible to show in Search with a snippet, and the site must be included in Search generative AI features in Search Console, which is the default setting. AI Overviews are “rooted in our core Search ranking and quality systems”, so normal SEO decides which pages are linked (Google Search Central).

The same guide is explicit about what you do not need:

• No new machine-readable files, AI text files or Markdown versions of pages.
• No special schema.org markup for generative AI search.
• No breaking content into small chunks; there is no ideal page length.
• No rewriting content specifically for AI, and no chasing long-tail keyword variations.
• No pursuit of inauthentic mentions across the web.

Google also describes how answers are assembled. AI Overviews and AI Mode may use a “query fan-out” technique: the model generates concurrent, related queries to gather more results. Creating separate pages for every variation in order to manipulate those answers, Google warns, violates its scaled content abuse policy.

What do the click studies show?

Source	Published	Data and method	Headline finding
Ahrefs	Feb 2026	300,000 keywords; aggregated Search Console desktop CTR, December 2023 vs December 2025; expected CTR estimated from informational keywords without AI Overviews	Position 1 CTR about 58% lower than forecast when an AI Overview appears (0.016 actual vs about 0.037 expected)
Seer Interactive	Apr 2026	53 client brands, 5.47 million queries, 2.43 billion organic impressions, January 2025 to February 2026	Organic CTR on queries with an AI Overview fell from 3.19% to 1.31% over 2025, then recovered to 2.36% by February 2026; queries without one rose from 2.75% to 3.82%
Semrush	Dec 2025	10 million+ keywords for prevalence; 200,000+ keywords with Datos clickstream data; same keywords before and after an AI Overview appeared (May vs October 2025)	Zero-click rate for those keywords fell from 33.75% to 31.53%; AI Overviews settled at around 16% of queries by November 2025
Pew Research Center	Jul 2025	Browsing data from 900 US adults in March 2025; 68,879 Google searches	People clicked a traditional result on 8% of visits with an AI summary vs 15% without; 1% clicked a link inside the summary
Google	Aug 2025	Internal data; method not disclosed	Total organic click volume “relatively stable” year over year, with “slightly more quality clicks”

Research shows a consistent direction across the independent studies: a search that shows an AI Overview tends to produce fewer clicks on organic results than one that does not. The size of the effect is where they part ways.

Why do the studies disagree?

They answer different questions, with different baselines:

1. Different metrics. Ahrefs measures CTR at each ranking position. Seer measures CTR across its clients’ accounts. Semrush measures the share of searches that end without any click. Pew measures what individual people do on a results page. Google reports total click volume. Our view: total clicks can hold steady while clicks per search fall, if the number of searches grows, so Google’s figure and the others are not necessarily in conflict.
2. Different counterfactuals. Ahrefs compares AI Overview keywords with informational keywords that never had one. Semrush compares the same keywords before and after. Seer compares queries with and without an AI Overview inside the same accounts. Each baseline has its own blind spots, and the Ahrefs authors note theirs relies on two years of change in other SERP features.
3. Different query mixes. Semrush found AI Overviews shifting away from purely informational queries: informational intent fell from 91.3% to 57.1% of triggering keywords during 2025. Studies weighted towards informational queries will see larger effects.
4. Different time windows. Seer’s data shows a trough in late 2025 and a partial rebound in early 2026. A study that stops in December sees a different picture from one that runs into February.
5. Undisclosed methods. Google has not published how it measured “relatively stable” or what counts as a quality click beyond users not quickly clicking back.

For planning, our view is to assume that informational queries with AI Overviews send fewer clicks per impression than they did before 2024, and to treat any single percentage, including the ones above, as approximate.

Does being cited in an AI Overview help?

It helps, but it does not restore the old click-through rate. In Seer’s 2026 data on informational queries, brands cited in the AI Overview averaged 2.07% organic CTR, against 0.94% for brands not cited and 3.35% where no AI Overview appeared. On paid results the gap was 15.74% versus 11.19%. The authors state that they cannot claim causation: stronger brands are more likely to be cited and also more likely to be clicked.

Who gets cited correlates with off-site reputation. In an Ahrefs study of 75,000 brands, branded web mentions had a 0.664 correlation with appearing in AI Overviews, far stronger than backlinks at 0.218. Research shows correlation only.

Which optimization steps does the evidence support?

1. Confirm eligibility. The page is indexed, returns a 200 status, and has no nosnippet or max-snippet:0 directive. Google says its nosnippet and related preview controls also limit how content appears in AI features. Leave the Search generative AI setting in Search Console on unless you mean to opt out. A free SEO audit flags snippet blocks and indexing problems in one pass.
2. Cover the sub-questions people ask. Because answers are assembled from fan-out queries, a page that answers the main question and its obvious follow-ups has more chances to be retrieved. Do it within strong pages rather than by spinning up a page per variation. The query fan-out explainer shows how to map those sub-questions.
3. Write what others have not. Google says “Don’t just recycle what others on the internet have already said”. Original data, first-hand testing and specific expertise give an AI Overview a reason to link you rather than a competitor.
4. Earn real mentions. The correlation between brand mentions and AI Overview visibility supports PR, partnerships and research that others cite. Google warns that manufactured mentions are not worth pursuing.
5. Skip the AI-only extras. In a matched test, Ahrefs found that adding schema did not boost citations; for AI Overviews the measured change was slightly negative and could not be attributed to schema. Google says llms.txt and chunking are unnecessary. Spend the time on steps 2 to 4.

How do you measure your AI Overview exposure?

• Search Console’s generative AI reports. They show impressions in AI Overviews, AI Mode and Discover’s generative features by page, country, device and date. They show no queries and, at the time of writing, no clicks. Google says they reached all websites on August 31, 2026.
• Clicks alongside impressions. Compare those pages’ clicks in the standard Performance report over the same period to see whether AI exposure is turning into visits.
• A tracked query set. Record weekly whether an AI Overview appears for your 30 most important queries and whether you are linked in it. Prompt-tracking tools can automate this; the AI visibility checker covers the technical side, from crawler access to snippet eligibility.
• Trends, not snapshots. As the Seer data shows, CTR on these queries moved in both directions within a few months.

What to do next

Treat AI Overviews as part of Search, because Google does. Fix eligibility problems first, since they are binary and quick. Then put your effort into the pages that answer your most valuable informational queries: cover their follow-up questions, add something original, and earn mentions from sites your audience trusts. Measure with Search Console’s generative AI reports and a fixed query set, and judge progress over quarters, not weeks. For how this fits with the rest of your search strategy, read GEO vs SEO, and see zero-click search for the broader trend these studies measure.

Frequently asked questions

How do I rank in Google AI Overviews?

Google says there are no special requirements beyond normal Search eligibility: the page must be indexed, eligible to show with a snippet, and your site must be included in Search generative AI features in Search Console, which is the default. Beyond that, AI Overviews draw on Google's core ranking systems, so relevance and quality decide which pages are linked.

Do AI Overviews reduce clicks to websites?

Most independent studies find fewer clicks per search when an AI Overview appears, but the size of the drop varies widely with method, from no increase in zero-click rates (Semrush) to a 58% lower top-position click-through rate (Ahrefs). Google says total organic click volume has been relatively stable. Treat any single figure as approximate.

Can I opt out of AI Overviews without leaving Google Search?

Yes. Search Console has a Search generative AI control that excludes your site's content from AI Overviews, AI Mode and generative features in Discover without affecting classic results. Snippet controls such as nosnippet also limit AI features but remove classic snippets too. Google-Extended does not affect AI Overviews.

Where can I see AI Overview data for my site?

Search Console's Search generative AI performance reports show impressions in AI Overviews, AI Mode and Discover, broken down by page, country, device and date. Google says the reports reached all websites on August 31, 2026. They do not include query-level data.