Google AI Overviews can reduce clicks to conventional search results by answering a query on the results page, but the evidence does not support one universal traffic-loss percentage for publishers. The observed effect depends on what is measured, which searches are included, and how Google presents the results.
What the available evidence says about clicks
The figures below describe different populations and outcomes. They should not be treated as competing estimates of the same thing—or as a forecast for any one site.
As an Amazon Associate I earn from qualifying purchases.
| Evidence | Reported result | What it measures |
|---|---|---|
| Pew Research Center, published July 22, 2025 | In Pew’s March 2025 data, users clicked a traditional search result on 8% of visits with an AI summary, versus 15% of visits without one. A link in the summary was clicked on 1% of visits to pages with a summary. Pew also reported that 58% of respondents had at least one Google search that produced an AI-generated summary. | Observed search behavior in the study; not an individual website’s expected change in clicks. |
| Google, August 6, 2025 | Google said overall organic click volume from Search to websites was “relatively stable year-over-year” and that its measure showed slightly more “quality clicks.” Google defines those as clicks after which users do not quickly click back. | Google’s own aggregate reporting and quality-click definition, not an independent causal estimate for AI Overviews. |
| Saharsh Agarwal and Ananya Sen, field-experiment paper posted in 2026 | The paper reports that, conditional on an AI Overview appearing in its study setting, outbound organic clicks fell 39.8% and zero-click searches rose 34.5%. | Experimental estimates for the study’s participants and setting; not a universal publisher-wide rate. |
These results can coexist. Pew reports what users did in observed searches that did or did not show a summary. Google describes aggregate Search click volume across its own data and applies a particular quality measure. The experiment estimates outcomes conditional on an AI Overview appearing in its setting. Differences in population, time, query mix, product version, and measurement all matter.
Why AI Overviews can change publisher referrals
Some searches may end on the results page
An overview can provide enough information for a user to stop searching without visiting a source. In that case, the publisher may lose a potential click even if its information helped inform the answer. The observed zero-click increase in the 2026 experiment is evidence for this mechanism in that study, not a prediction for every query or site.
#1 Best Overall
Links inside an overview are not the same as regular result links
Google’s results page can present links in different places and formats. In August 2024, Google said an early experiment showing links to supporting pages directly within AI Overviews was driving higher traffic to publisher sites, but it gave no effect size. That statement concerns an early experiment, so it does not establish how current placements perform across queries or publishers. Google’s announcement
A referral count does not establish visitor value
Clicks, sessions, engagement, conversions, and revenue are separate outcomes. The UK Competition and Markets Authority’s publisher conduct document records publishers’ concern that they cannot accurately assess whether generative-search referrals are more valuable than traditional search referrals. It identifies a measurement concern, not a quantified difference in visitor quality. CMA document
Rank #2
How to assess whether your site is affected
Your own analytics can reveal where Google referrals changed, but a sitewide decline alone cannot prove that AI Overviews caused it. Search rankings, query mix, seasonality, site changes, and changes to the results page can also move traffic.
- Set a comparison window. Compare like-for-like periods and annotate major content, technical, and site changes. Account for seasonal patterns that affect your subject.
- Break the data down. In Search Console, review Google Search impressions and clicks by page and query. In your analytics, review sessions by landing page and referral source. Focus first on pages and queries where an answer-style overview could plausibly satisfy the search.
- Compare impressions with clicks. If impressions are steady or rising while clicks fall, the change may be in click-through behavior or result presentation rather than simply fewer searches or less visibility. This pattern is a clue to investigate, not proof of a cause.
- Check sessions and outcomes separately. Look for corresponding changes in landing-page sessions and the engagement or conversions that matter to your business. Do not assume a change in search clicks equals the same change in visits or revenue.
- Use a comparison group where possible. Compare potentially exposed pages with similar pages that appear less exposed, while accounting for differences in rankings, query intent, and seasonality. A comparison can strengthen a diagnosis, but without a suitable causal design it still cannot isolate AI Overviews with certainty.
- Record what you observe in Search. For important queries, note whether an AI Overview appeared and how links were presented when you checked. Availability and layout can vary by query, country, language, device, and time; a spot check is not a complete exposure history.
Do not infer that an overview caused a drop merely because the timing lines up. The three sources above answer different questions, and none supplies a universally valid attribution method for an individual publisher.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Can a publisher opt out of AI Search features?
In June 2026, Google announced a Search Console toggle giving site owners control over whether a site may appear in and help ground generative AI Search features. Google said sites opting out would not receive traffic or impressions from those features. Google’s announcement
Because this is a changeable product control, check Google’s current Search Console documentation and the consequences shown for your property before making a decision. The announcement establishes the stated trade-off: opting out removes visibility in those features as well as their associated traffic and impressions. It does not establish what effect opting out would have on a particular publisher’s overall Search performance.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Free tools Windows power users keep installed
One-click scans. No signup required.
