What happens to clicks when Google shows an AI Overview
When a Google AI Overview appears, Pew Research Center found that US users in its study clicked ordinary result links less often, almost never clicked a source inside the summary, and were more likely to end the session on that results page. Those figures are observational, from one US sample, in one month. They are not a forecast of your traffic, and they are not a UK study. Google has publicly disagreed with some third-party claims that AI features caused dramatic declines in aggregate clicks. Being named or cited in the overview still matters, because that is the shortlist the user sees first. Nobody can switch that on. AI results vary.
Why AI search visibility matters for local businesses uses these figures as part of a wider point. Below is what Pew measured, what it did not, what Google says, and what a local business can still do.
What did Pew actually measure?
Pew published Google users are less likely to click on links when an AI summary appears on 22 July 2025, by Athena Chapekis and Anna Lieb. The analysis used browsing data from 900 US adults in KnowledgePanel Digital, who had agreed to install an app that tracked their activity.
The browsing covered 1 to 31 March 2025. For every google.com search in that set, Pew used a third-party scraping service to re-run the same query and collect the first page of results, including any AI summary and up to three source URLs from that summary. Those result snapshots were collected from 7 to 17 April 2025. Pew notes that an AI summary for a given search may change over time, and that the analysis reflects how results appeared when they were collected, not necessarily how they looked on the day of the original search.
The dataset contained 68,879 unique Google searches. 12,593 of those searches produced an AI summary. Because of technical limits, the analysis includes only Google searches, not other search engines.
To judge the next action, Pew used the timestamp and URL of the next visit, and grouped behaviour as: clicking a link in the summary or in the traditional results; continuing to search Google; leaving Google for a different site that was not one of those links; or ending the browsing session, which Pew defined as exiting the web browser for five seconds or more. If a link appeared in both the summary and the traditional results, the click was counted towards the summary.
What were the headline figures?
These are the figures Pew reported, in its own terms:
- 58% of the 900 adults had at least one Google search in March 2025 that produced an AI-generated summary.
- About 18% of all the Google searches in the study produced an AI summary. Pew also describes this as around one in five.
- Users clicked a traditional search result link in 8% of visits when a summary appeared, and in 15% of visits when it did not.
- They clicked a source inside the summary in 1% of visits to pages that had one.
- They ended the browsing session on 26% of pages with a summary, compared with 16% of pages without one.
- Regardless of whether a summary appeared, the largest share of searches in the study ended with the user either browsing elsewhere on Google or leaving without clicking a result link. Pew says around two-thirds of all searches resulted in one of those actions.
Longer and question-shaped searches were more likely to produce a summary:
- 8% of one- or two-word searches.
- 53% of searches with 10 words or more.
- 60% of queries that began with question words such as "who", "what", "when" or "why".
- 36% of searches that included both a noun and a verb.
Pew also looked at what got cited. Wikipedia, YouTube and Reddit were the most frequently cited sources in both AI summaries and standard results, together accounting for 15% of sources listed in the summaries examined, and 17% of sources in standard results. Government websites were more common in summaries than in standard results: 6% of sources linked in AI overviews were .gov sites, compared with 2% in standard results. News websites were 5% in both. Around 88% of the summaries cited three or more sources, and only 1% cited a single source. The median summary was 67 words long, with a range from 7 to 369 words.
None of those figures is a ranking factor, a UK measurement, or a prediction for a particular local business. They describe what this US panel did, and what those re-run result pages contained, in the spring of 2025.
What does Google say about traffic?
Google has disputed some third-party reports that AI features caused dramatic declines in aggregate clicks to websites. In a 6 August 2025 post, Liz Reid, VP and Head of Google Search, wrote that total organic click volume from Google Search to websites had been relatively stable year over year, and that average click quality had increased, with slightly more "quality clicks" than a year earlier. Google defined quality clicks as those where users do not quickly click back.
Reid wrote that this contrasted with third-party reports that, in Google's view, inaccurately suggested dramatic declines in aggregate traffic, often based on flawed methodologies, isolated examples, or traffic changes that occurred before AI features rolled out in Search. The post also said people are searching more, asking longer and more complex questions, and seeing more links on the page when an AI Overview appears. It accepted that traffic is shifting between sites, so some sites see less and others see more.
Google did not publish the underlying click counts in that post. Treat it as Google's public position, not as an independent replication of Pew.
Pew observed fewer clicks on traditional result links, and rarer clicks on summary sources, in the sessions that had a summary. Google says overall organic click volume has been relatively stable, that click quality has risen slightly, and that some aggregate-decline claims are wrong. Neither source tells you what will happen to your own referrals next month.
What can a local business still do?
You cannot control whether Google shows an overview for a query, and you cannot switch a mention on. You can still make the public facts accurate, and you can measure Google's own AI features rather than guessing from a headline percentage.
For the work itself, Google's guide to generative AI features, last updated 10 July 2026, says this is still SEO: useful, non-commodity content, crawlable pages, and ordinary local details. A complete Google Business Profile can help services be visible in AI responses as well as in ordinary Search. Google says you can ignore llms.txt, special AI markup, "chunking", writing only for machines, inauthentic mentions, and special schema invented for AI Overviews. Do not build a thin page for every wording of a query.
The practical writing method is in how to write a service page that AI tools can quote. The broader local steps are in Google AI Overviews for local businesses.
For measurement, use Search Console's Generative AI performance report. As of 31 August 2026, Google says those insights have been rolled out to all websites worldwide. The report covers impressions in AI Overviews and AI Mode on Google Search. An impression is counted when links to your site were shown to a user in one of those features. The report does not measure ChatGPT, Claude, Gemini or Perplexity. Google says you may not see the report until the site has had enough impressions, and that data from Search Labs experiments is excluded.
If you want a picture of who is named for your customer questions in ChatGPT or Claude, that is a different test, with a dated count, not a click-through rate. The rule is in how The Named List counts a name.
Frequently asked questions
Does this mean my website traffic will fall by the same percentages?
No. Pew's figures are observational US data from one month of browsing, not a forecast for your site and not a UK sample. Google has also disputed some third-party claims of dramatic aggregate declines.
Does the Search Console report measure ChatGPT?
No. It measures generative AI features on Google Search, currently AI Overviews and AI Mode. It does not measure ChatGPT, Claude, Gemini or Perplexity.
If people rarely click the sources, why does being in the overview still matter?
Because the businesses or pages named in the answer are the shortlist the user sees first. They may phone a name, or they may not click anything. Being accurate in the sources an overview might read is still the work you can do.
Is this the same as the why-it-matters post?
That post uses these figures as part of a wider argument. This one covers Pew's method, the extra detail, Google's response, and how to measure Google's own AI features.
Want someone to run the checks for you?
A Who-Gets-Named Check asks the customer questions for your town and shows who is named and which pages are cited. If you would rather hire someone, how to choose an AI visibility service lists the questions to ask, including of me. AI results vary.