How much does an AI summary increase the no-click rate?

In a Pew Research Center study of 68,879 Google searches during March 2025, 26% of pages with an AI summary ended the browsing session with no click at all, compared with 16% of pages showing only traditional results. That is a ten point gap on one panel in one month, not a universal constant.

Is zero-click new, or did AI cause it?

It predates AI summaries by years. Weather answers, currency conversions, sports scores, definitions and knowledge panels have resolved queries on the results page for a long time. What AI summaries changed is the range of questions that can be resolved that way, extending it from short factual lookups into explanatory and comparative questions.

If nobody clicks, is being cited worth anything?

It depends what the page was for. A page whose value is a completed transaction still needs the click. A page whose value is being the reference a buyer remembers can pay off through the mention alone, which is why appearance in answers is worth tracking as its own measure rather than as a proxy for sessions.

How do I measure zero-click exposure?

Not from your analytics, which by definition cannot see a session that never reached you. It has to be measured on the other side: sample a fixed set of queries and prompts on a schedule and record whether you appear and whether a link to you is offered.

Should I make my pages harder to summarise?

Withholding the answer usually costs the citation without recovering the click, because the summary is assembled from whatever sources will supply it. The supported way to be excluded is the snippet controls, which remove you from ordinary snippets at the same time.

Zero-click search - AI SEO & GEO Glossary

A zero-click search is one where the reader gets what they came for on the results page and never visits a website. The term is older than AI search — weather, conversions, opening hours and dictionary definitions have been answered in place for years — but the arrival of generated summaries widened the set of questions that can end this way, from short factual lookups to explanations and comparisons.

The measured size of the effect

The clearest public measurement is the Pew Research Center analysis of 68,879 Google searches made by 900 US adults who shared browsing data during March 2025. Pew reports that a browsing session ended entirely on 26% of pages carrying an AI summary, against 16% of pages showing only traditional results. On the click side, users who saw a summary clicked a traditional result in 8% of visits versus 15% without one, and clicked a link inside the summary in 1% of visits.

Two cautions belong with those numbers. They describe US desktop-and-mobile browsing on Google in a single month, and they measure a panel rather than the whole web. They are a good anchor and a poor constant, which is the honest way to use any single study.

How it differs from a low click-through rate

These are often conflated and they call for opposite responses.

A low click-through rate means you appeared and were not chosen. The listing lost to something else on the same page, and the fix is usually the listing: the title, the description, the match to intent.

A zero-click outcome means nobody chose anything. The question was answered before choosing arose. Rewriting your title does not recover it, because the reader never reached the moment of choice.

Mistaking the second for the first produces months of work on snippets that were never the constraint. The diagnostic is simple: if impressions hold while clicks fall across a whole class of informational queries, you are looking at zero-click, not at a weak listing.

What it means for how you value a page

The useful move is to stop treating a session as the only unit of value, and to separate pages by what they were actually for.

Transactional pages — pricing, signup, checkout, contact — still need the click, and are largely unaffected because their queries are not the ones summaries resolve.
Reference pages — definitions, comparisons, how-to material — are exactly what gets summarised. Their return now arrives partly as being the source an answer draws on, which is a different accounting line from sessions.

Neither of those is an argument for publishing less reference material. The summary has to come from somewhere, and being one of the sources it comes from is the only position available. Source citation is the measure that survives when the click does not.

How to act on it

Measure from the other side. Your analytics cannot see a search that never reached you. Sample a fixed set of queries and prompts on a schedule and record whether you appear at all. Generative engine optimization is the discipline built around that sampling.
Split reporting by page purpose before drawing conclusions. A blended sessions chart hides the effect entirely, because transactional and reference pages move in opposite directions.
Decide the exclusion question explicitly. The supported controls are nosnippet, data-nosnippet, max-snippet and noindex, and the first three also govern ordinary snippets. noindex vs nosnippet sets out the trade.
Keep the answer on the page. Withholding it does not usually recover a click; it forfeits the citation to whoever did answer.

Frequently asked questions

How much does an AI summary raise the no-click rate?

Pew measured 26% of pages with an AI summary ending the session with no click, against 16% without one.

Did AI cause zero-click search?

No. It extended it from short factual lookups to explanatory and comparative questions.

Is a citation worth anything without a click?

For reference pages, often yes. For transactional pages, the click is still the point.

Can I measure it in my own analytics?

No. A session that never reached you cannot appear in your own logs.

Zero-click search