4 November 2026 · 7 min read

Why Websites Get Fewer Google Clicks in 2026, and What Still Works

Randomized studies link AI Overviews and AI Mode to fewer clicks. What they measured, what Google says, and what small sites can do that an AI answer can't.

Websites may be getting fewer Google clicks because more searches are now answered on the results page itself. In experiments posted online in 2026, people shown AI Overviews or put into Google's AI Mode clicked through to outside sites noticeably less often. Google disputes the wider claim of dramatic traffic declines, and your own traffic may have other causes. What still tends to earn a visit is what a summary cannot hand over: working tools, original photos, first-hand data, and pages that are fast to use.

What do the studies actually show?

Randomized experiments are useful here because they compare similar people who were assigned different versions of search, rather than simply comparing searches that happened to show an AI answer with those that did not. The first two studies below are randomized; the third, from Pew, is observational.

Be careful comparing these numbers, because they measure different things in different ways. 'Percentage points' and 'percent' are not interchangeable, and each study has its own definition of a click and its own set of searches.

What are the limits of those numbers?

• The Penn and Northeastern paper forced everyone in one group into AI Mode. Before the experiment, AI Mode made up just 0.6% of its participants' searches, so the authors note that this reflects a full-adoption scenario rather than gradual, opt-in use.
• Google changed its page code during that study, which broke the hiding of AI Overviews part-way through. About half were hidden overall. The 8.8-point figure is an estimate that adjusts for this; the simple comparison by assigned group was 2.7 points and was not statistically significant.
• The same paper's sample skewed younger and more educated than the US population, and the exposure lasted a week. It is also marked as a draft.
• The ISB and Carnegie Mellon paper reportedly found the effect mainly on informational searches, with no statistically clear effect on navigational or transactional ones, although those estimates were noisier. We read it through PPC Land's summary, not the paper.
• Pew's analysis is observational: searches that trigger a summary may differ from those that do not.
• None of this tells you what happened to your site. Each experiment measures clicks to the web in general, not a specific business.

What does Google say about AI Overviews and clicks?

Google has pushed back on the broader story. In August 2025, Liz Reid, its head of Search, wrote that total organic click volume had been 'relatively stable year-over-year', that average click quality had increased, and that third-party reports of dramatic declines in aggregate traffic were inaccurate. That statement predates the 2026 experiments. We did not find a Google response specific to those papers.

These claims can partly coexist. Aggregate clicks across the whole web can look stable while particular kinds of site, such as those that mainly answer simple questions, lose a lot. Google's own guidance says there are no additional requirements or special optimizations for appearing in AI Overviews or AI Mode, and that normal SEO best practices still apply.

How do you check if AI search is hurting your site?

• Open the Performance report in Google Search Console. Google says traffic from AI features is included in the overall figures, under the Web search type. It is not split out there, so you are looking for patterns.
• Compare the same weeks year on year. Seasonal swings can look like a collapse.
• Look for pages where impressions held but clicks fell. That is consistent with people seeing your result, or an answer, without needing to click, though it is not proof.
• Sort the affected pages by type. Definitions and quick facts falling while tools and product pages hold steady tells you something useful.
• Rule out the rest: a site migration, a broken page, a ranking change, or a competitor. Not every drop is AI.

What still works when AI answers the question?

No one can promise a formula. The practical question is simply: what does a person still need to visit your page to get?

• Answer the question early and then go deeper. People who do click have a reason, so give them the detail, the example or the next step.
• Build direct routes: an email list, a useful bookmarkable tool, a social presence. Then a single search change matters less.
• Consider Google's preview controls with care. Its documentation points to nosnippet, data-nosnippet and max-snippet for limiting what can be shown, but these can also limit how your page appears in Search.
• Measure conversions, not just visits. Fewer, better visits may be fine. Google itself argues that click quality has improved, but check that against your own numbers.

Why do fast pages and light images matter more now?

If each visit is harder to win, you do not want to lose it to a slow page. Google's web.dev guidance suggests a Largest Contentful Paint of 2.5 seconds or less, measured at the 75th percentile of page loads. On image-heavy pages, oversized photos are often the first thing to fix: export them at the size they are displayed, then compress them. Speed will not bring back a click that an AI answer has already satisfied, but it does help you keep the people who arrive.

Quick answers

• Should I block Google's AI features? Think hard first. Limiting snippets can reduce what appears, but it may not bring clicks back and could reduce your visibility.
• Is SEO dead? The studies show fewer clicks in some conditions, not zero clicks. Search still sends people to websites.
• Will AI Mode become the default? We do not know, and neither do the studies. Plan for more AI answers without assuming any particular outcome.