Are people more likely to stop browsing after seeing an AI Overview?

Do searches end more often when an AI Overview appears?

Yes: in Pew’s panel, browsing ended after 26% of AI Overview pages, against 16% of other pages. An AI Overview is the AI-written summary Google shows at the top of many results pages. Chapekis, Lieb, Shah and Smith (opens in a new tab), researchers at Pew Research Center, studied a month of real browsing from March 1 to 31, 2025.

The 900 US adults came from a panel recruited by address-based sampling, chosen to be as representative of the country as possible. Over the month they made 2,457,176 web page visits. 91,121 of those visits were to Google search pages, covering 68,879 different searches.

For each Google page, the researchers recorded what the person did next.

How did the researchers define stopping?

A session counted as ended when the person left the web browser for five seconds or more. Other next steps were a click on an AI Overview source, a click on a search result, another Google search, or a visit elsewhere.

Clicks on AI Overview sources counted only the first three cited links. In April 2025, three was the most a desktop browser showed without the user asking for more.

Five seconds is a short pause. Some people who “ended” a session may have returned minutes later. The study also cannot see what they did elsewhere, such as in an app.

Is the AI Overview the cause, or the kind of search?

Probably partly the AI Overview itself: an experiment and a traffic study both point that way.

The worry is fair. Pew found that longer searches, questions and searches with both a noun and a verb were more likely to show an AI Overview. Those searches might end sooner anyway.

Pew’s researchers adjusted for those features of each search and for each person’s own habits, and the gaps held. Their data is still observational, so it cannot prove cause.

Stronger evidence comes from a field experiment by Wang and colleagues (opens in a new tab) with 1,100 US Google users in March 2026. People whose AI Overviews were hidden clicked through to outside websites 8.8 percentage points more often. A Google page change meant the tool hid only 51.1% of AI Overviews, and the researchers adjusted their estimate for that.

Those people’s search sessions also became shorter, by 0.59 minutes. A study of Wikipedia traffic (opens in a new tab) adds a real-world estimate. After AI Overviews launched in the US in 2024, English Wikipedia’s search visits fell 5.45% relative to German editions.

How often will your buyers see an AI Overview?

It depends heavily on the search: published rates range from about one search in seven to nearly all of them. The differences come from which searches each study counted.

Questions are the clearest signal. Xu and colleagues found AI Overviews on 64.7% of trending searches phrased as questions, against 9.5% of the rest. In our study, 98.1% of commercial keywords phrased as questions showed one.

Those are the research-style searches where buyers often first meet a brand.

Does stopping mean the searcher got what they needed?

Unknown: the data shows people stopped, not whether they were satisfied. Researchers call a search that ends because the answer was on the page “good abandonment.” Neither study separates that from giving up.

The experiment offers a clue. Hiding AI Overviews produced no detectable change in how much people trusted information on Google or rated their search experience. The authors read this as a sign that any benefits of AI Overviews do not show up in those measures.

The answers people stop at are not always fully backed by their sources. Across 98,020 statements in AI Overviews, Xu and colleagues found 11.0% were not supported by the pages cited. Most were points the source did not mention at all.

Some of that may come from social media pages they could not read; even so, they put the floor at roughly 5.3%. In a small pilot of our own, judged by AI coders rather than people, 12.7% of 71 cited sentences were not supported by the page cited. Someone who stops there leaves with that answer, and your page gets no chance to correct it.

What should you do about it?

Assume fewer searches will reach your site, and work to be the source the AI Overview names. Practical steps:

1. Find your exposed searches. Check which of your important searches show an AI Overview, starting with question-style and longer searches.
2. Aim to be cited, not only ranked. In our citation study, the first organic result was cited in 49.5% of AI Overviews and the ninth in 15.5%. Ranking helps but does not guarantee a mention.
3. State key facts plainly. If the summary is all people read, your prices, features and claims need to be easy to lift accurately.
4. Measure more than clicks. Track how often your brand appears in AI Overviews, alongside branded searches and direct visits.
5. Watch AI Mode. Google’s conversational search cut outside clicks further in the same experiment, as we explain in our article on AI Mode and traffic.

If you want help finding which of your searches end on Google’s page, see our generative engine optimization service.

What does the research not tell us yet?

Only one study has measured session endings directly, so several questions remain open. The main gaps:

• Breadth. The direct evidence is one US panel over one month in early 2025, and AI Overviews have changed since.
• Cause. The Pew findings are associations; the experiment measured clicks and session length, not session endings.
• What happens next. A five-second exit may hide a return visit, a phone call or a switch to another app.
• Business searches. None of the studies reports session endings for B2B or high-value purchase searches.
• Satisfaction. No study has asked people who stopped whether they got what they needed.

Frequently asked questions

Do AI Overviews reduce clicks to websites?

Yes. In Pew’s panel, people clicked a result on 8% of AI Overview pages against 15% of others. In an experiment, hiding AI Overviews raised outside clicks by 8.8 percentage points.

Do people click the links inside AI Overviews?

Rarely. In Pew’s panel of 900 US adults, just 1% of visits to AI Overview pages led to a click on a cited source.

Why do some searches show an AI Overview and others don’t?

Longer searches and questions trigger them most. One 2026 study of trending searches found AI Overviews on 64.7% of questions against 9.5% of other searches.

Is a search that ends on Google bad for my brand?

Not if your brand is named in the answer. In our study, the top organic result was cited in 49.5% of AI Overviews, so ranking well still helps you appear where the search ends.

Sources

• Chapekis, A., Lieb, A., Shah, S. and Smith, A. (2026), Investigating Click Behaviors On Google Search Result Pages That Produce an AI Overview (opens in a new tab), arXiv:2608.04831.
• Wang, S. T., Gleason, J., Bart, Y., Wilson, C. and Metaxa, D. (2026), AI in Search Reduces Publisher Referrals Without Improving User Experience: Experimental Evidence (opens in a new tab), arXiv:2608.18352.
• Khosravi, M. and Yoganarasimhan, H. (2026), Impact of AI Search Summaries on Website Traffic: Evidence from Google AI Overviews and Wikipedia (opens in a new tab), arXiv:2602.18455.
• Xu, H., Iqbal, U. and Montgomery, J. M. (2026), Measuring Google AI Overviews: Activation, Source Quality, Claim Fidelity, and Publisher Impact (opens in a new tab), arXiv:2605.14021.
• Grossman, R., Liu, S., Chen, M. K., Smith, M., Borcea, C. and Chen, Y. (2026), How Generative AI Disrupts Search: An Empirical Study of Google Search, Gemini, and AI Overviews (opens in a new tab), arXiv:2604.27790.
• Underneath (2026), When does Google show an AI Overview? 1,248 US searches
• Underneath (2026), Do AI Overviews cite the pages that rank? 4,051 citations