The click gap: what AI Overviews cost you

The click gap is the widening distance between how often your page ranks and how often that ranking actually sends you a visitor. You can hold the same position you held two years ago and receive a fraction of the traffic, because an AI summary now answers the query before the user ever reaches your link.

That gap is no longer a forecast. It shows up in behavioral data.

#What the data shows

The clearest look comes from the Pew Research Center, which tracked the browsing activity of 900 U.S. adults across 68,879 unique Google searches in March 2025. On pages where an AI summary appeared, users clicked a traditional search result in 8% of visits, compared with 15% of visits on pages without a summary. The organic click roughly halved.

The link inside the summary fared worse. Users clicked one of the sources cited in the AI answer in just 1% of visits to pages that carried a summary. Being named as a source is not the same as being visited.

Users were also more likely to stop searching entirely. A browsing session ended on 26% of pages with an AI summary versus 16% of pages with only traditional results. The answer was good enough. There was nothing left to look up.

Zoom out to the whole results page and the picture is the same. Analysis of Similarweb clickstream data for the first four months of 2026 found that 68.01% of U.S. Google searches ended without any click, and only about 374 of every 1,000 searches sent a click to the open web. Fewer than a third of searches now reach a site outside Google. AI Overviews, which appear on more than 20% of searches and cut click-through by close to 60% when they show, are a large part of why.

Google disputes the Pew figures and questions the methodology, so treat any single number as directional rather than precise. The direction, across independent panels, is consistent: the ranking still exists, the click does not follow it as reliably as it used to.

#Why ranking and traffic came apart

For most of search's history, SEO rested on a simple chain: rank well, earn the click, get the visit. AI answers break the chain in the middle. The summary extracts the fact, the definition, or the short comparison the user wanted and renders it on the page. The user reads it and leaves satisfied. Your position was never the problem. The interface no longer requires a visit to deliver the value.

This hits informational and definitional queries hardest, the "what is," "how does," and "best way to" questions that used to feed top-of-funnel traffic. Those are exactly the queries an AI summary is built to resolve in place.

#What still earns a click

Not every query collapses. Clicks survive where the answer cannot be fully contained in a paragraph:

• Transactional and product queries. People still click through to buy, sign up, download, or start a trial. The AI can describe a tool. It cannot complete the action.
• Deep or high-stakes research. When the decision carries money or risk, users verify. They open the source to check the reasoning, the methodology, or the fine print.
• Original data and primary sources. When your page is the thing being summarized, the summary becomes an advertisement for it. Users who need the full dataset, the full study, or the exact quote still come.
• Brand and navigational queries. If someone is looking for you by name, the summary does not stand between you and them.

The pattern is that clicks concentrate on pages that offer something a paragraph cannot replace: an action, a full dataset, a tool, a document.

#The response: optimize for the answer, not just the rank

If half your former clicks are being absorbed by the summary, ranking for those queries is no longer the finish line. Two shifts follow.

First, treat citation inside the answer as its own goal. When the AI resolves a query without a click, being named as the source it drew from is the visibility that remains. That comes from the fundamentals we have written about before: an answer-first structure that states the claim in the first sentence, entity clarity so the model knows who you are, and clean, machine-readable structure that lets an engine lift a passage cleanly. You are writing to be quoted, not only to be ranked.

Second, change what you measure. A rank-tracking dashboard that shows position five and calls it a win is now telling you less than half the story, because it cannot see whether the click ever arrived or whether an AI answer intercepted it. Pair rank with two other signals: share of citation, meaning how often you appear as a named source in AI answers for your target queries, and the traffic that actually lands. We covered the reporting shift in from SEO reporting to GEO reporting, and why share of citation is becoming the new share of voice.

In Crawlmind audits, the most common surprise is not a ranking problem. It is a page that holds a strong position and still appears in almost no AI answers for the same query, because it buries its answer three paragraphs down or never states the entity plainly. The rank was fine. The page was not built to be extracted.

#What to do this quarter

You do not need to rebuild your content program. You need to sort it by what the click gap does to each type of page.

1. Segment your pages by intent. Separate the informational pages, whose clicks the summary is most likely to absorb, from the transactional and data pages that still pull visits.
2. Rewrite the informational pages to be quotable. Lead with the answer, define the entity, and structure the page so an engine can lift a clean passage. If the click is going to be intercepted, be the source it cites.
3. Protect and expand the pages that still convert. Product pages, calculators, original research, and comparison content are where surviving clicks concentrate. Give them more attention, not less.
4. Add citation share to your reporting. Track how often you are named in AI answers alongside rank and landed traffic, so a hollow ranking does not read as a success.

The click gap is not a reason to abandon search. It is a reason to stop treating position as a proxy for traffic. Rank where you can, get cited where the click disappears, and put your click-earning pages where the clicks still go.