benchr
Picture a page that does everything right. It ranks first for its keyword, the title is sharp, the content answers the question better than anything else on the results page. A year ago that meant traffic. Now the query fires, Google paints an AI Overview across the top of the screen, and the answer is already there before your blue link ever enters the frame. The user reads the summary, nods, and closes the tab. You held position one and lost the click anyway.
That's the quiet rerouting at the center of this. Not a search engine sending you less traffic out of spite, but the answer migrating up the page until the links underneath it become optional. The honest way to read it is through the numbers, not the dread, so here's where the data lands.
Two different things called zero-click
First, untangle two trends that get mashed together, because the panic comes from confusing them. Zero-click search is old. Long before any AI summary, Google answered plenty of queries on the page itself: weather boxes, sports scores, currency conversions, knowledge panels, featured snippets pulled straight from a site. The SparkToro study with Datos, run by Rand Fishkin and covering September 2022 to May 2024, measured this baseline at 58.5% of US Google searches ending without a click to the open web. Put another way, for every 1,000 US searches only about 360 clicks left for an outside site.
That number predates AI Overviews. It's the floor, not the new damage. The newer thing is the AI layer stacked on top, the generated summaries and chat answers that landed across 2024 and 2025, and that's what the fresher studies from Pew, Ahrefs, and Similarweb are measuring. Keep them separate or you'll blame the wrong culprit for the wrong drop.
What the AI layer did to clicks
Here's the cleanest read on the new layer. Pew Research Center tracked 900 US adults who shared their browsing, covering 68,879 Google searches in March 2025, of which 12,593 produced an AI summary. The headline contrast: people clicked a traditional search result on 8% of searches that showed an AI summary, against 15% of searches with no summary. About half as often. And the link inside the summary, the citation that's supposed to send traffic back to publishers? Just 1% of all AI-summary searches ended in a click on it.
That 1% is the brutal one. The whole pitch for AI Overviews to publishers is "we'll cite you, traffic still flows." The data says the citation almost never gets clicked. People got their answer; they're done. Pew also found users ended their browsing session on 26% of visits to a page with an AI summary, versus 16% without one, which is the same story from a different angle: the answer is terminal.
For anyone running a site, the lesson writes itself. If the only way you win is the outbound click, the AI layer is quietly taking it. The move is to get your page quoted inside the answer instead, which is the entire game of getting cited by AI search, structuring content so the model lifts a clean, attributable passage and names you.
The SEO hit, and it's getting worse
If Pew tells you what users do, Ahrefs tells you what it does to rankings. In April 2025, Ahrefs studied 300,000 keywords, half with AI Overviews and half without, and found the top-ranking page's average click-through rate was 34.5% lower when an AI Overview was present. Position one's CTR fell from 0.073 to 0.026 on those keywords. Bad, but survivable.
Then the February 2026 update landed, and the gap had widened to 58%. For informational keywords, position one's CTR slid from 0.076 in December 2023 to 0.039 in December 2025. So the #1 spot lost roughly half its clicks over two years, and the trend line is pointing the wrong way. This isn't a one-time shock the web absorbs and moves past. By the numbers, it's deepening.
Ranking first used to be the finish line. Now it's a seat in the row behind the answer.
Reading the figures side by side
Four sources, four angles. Lining them up is the only fair way to see the shape of it, because each one measures a different slice and they don't all agree on severity.
| Source | What it measured | Headline figure |
|---|---|---|
| SparkToro / Datos | US zero-click rate (all causes, pre-AI baseline) | 58.5% no click |
| Pew Research | Result click rate with vs without an AI summary | 8% vs 15% |
| Pew Research | Clicks on a citation inside the AI summary | 1% |
| Ahrefs (Feb 2026) | CTR drop for the #1 result under an AI Overview | 58% lower |
| Similarweb | News searches ending with no click, May 2024 to May 2025 | 56% → 69% |
News sites got hit hardest, which makes sense, since so many news queries are exactly the "just tell me what happened" questions an AI summary handles in one breath. Similarweb data, reported in July 2025, shows news searches ending with no click to a news site grew from 56% to nearly 69% in the year after the US AI Overviews launch. Over the same stretch, organic search traffic to news sites fell from a peak above 2.3 billion visits in mid-2024 to under 1.7 billion by May 2025, roughly a 26% drop. That's a real revenue line evaporating, not a rounding error.
Google's side, because it matters
Google doesn't buy the "traffic apocalypse" framing, and a fair read has to include the rebuttal. The company called the Pew study "a flawed methodology and skewed queryset that is not representative of Search traffic," and said it has "not observed significant drops in aggregate web traffic," still sending billions of clicks to sites daily. Its line on quality is that AI Overviews send better visitors: John Mueller said in July 2025 that clicks from AI Overview pages are "of higher quality, where users are more likely to spend more time on the site." Google also points out AI Overviews reach 1.5 billion monthly users, framing the feature as a new front door rather than a closed one.
What to actually do about it
Stop optimizing only for the blue link. That era is closing, and the numbers above are why. The work now splits in two: keep ranking, because position still feeds the AI Overview its sources, and structure your pages so the model can quote you cleanly and send what clicks remain your way. The tactics live in benchr's piece on getting cited inside AI answers, and they're a different muscle than classic SEO.
It also reframes how you should read leaderboard hype. If a model's "win" is just answering more queries without sourcing them, that's a cost shifted onto publishers, which is part of why benchmarks stopped telling you what a model is doing to the open web. And the same answer-first behavior is spreading past search into buying flows, where agentic shopping agents read product pages on your behalf and check out without you ever landing on the merchant. The click you're losing in search is the same click those agents skip. Worth knowing too what these systems do with your queries while they're at it, which is the territory of who trains on your chats.
The verdict: don't panic, but don't pretend the floor didn't move. Plan for a web where the answer often lives above your link, measure your own AI-Overview keywords against your non-AIO ones the way Ahrefs did, and treat citation share as a metric that now sits next to ranking. The traffic didn't vanish. It got rerouted, and the route runs through the answer box.
Frequently asked
What is zero-click search?
It's a search that ends without a click to any outside site, because the answer sits right on the results page. In the SparkToro and Datos study covering Sept 2022 to May 2024, 58.5% of US Google searches ended this way, so for every 1,000 searches only about 360 clicks reached the open web. The trend started well before AI Overviews, driven by featured snippets, knowledge panels, and weather boxes.
Do AI Overviews really kill traffic?
They cut it. Pew found in July 2025 that users clicked a regular result on just 8% of searches that showed an AI summary, versus 15% when no summary appeared, roughly half as often. Only 1% of those searches led to a click inside the summary. Google disputes the framing and says it still sends billions of clicks to sites daily.
How much does an AI Overview drop click-through rate?
For the #1 result, a lot. Ahrefs found the top page's average CTR was 34.5% lower when an AI Overview appeared in April 2025. By its February 2026 update, that gap had widened to 58%. So the hit is getting worse over time, not settling down.
What can a publisher do about lost clicks?
Stop chasing only the blue link and get your pages quoted inside the answer instead. That means structuring content so AI systems can lift a clean, citable passage, the whole point of GEO and AEO. benchr's guide to getting cited by AI search walks through the tactics that get ChatGPT, Perplexity, and AI Overviews to name your page.
Changelog
- May 30, 2026 — Originally published. Click and CTR figures verified against Pew Research, Ahrefs, SparkToro/Datos, Similarweb, and Google's public statements.
References
- SparkToro (Rand Fishkin) with Datos, "2024 Zero-Click Search Study," sparktoro.com, accessed May 2026.
- Search Engine Land, "Google search zero-click study 2024," searchengineland.com, accessed May 2026.
- Pew Research Center, "Google users are less likely to click on links when an AI summary appears in the results," pewresearch.org, accessed May 2026.
- Ahrefs, "AI Overviews reduce clicks," ahrefs.com, accessed May 2026.
- Ahrefs, "AI Overviews reduce clicks (Feb 2026 update)," ahrefs.com, accessed May 2026.
- Similarweb data, via TechCrunch, "ChatGPT referrals to news sites are growing, but not enough to offset search declines," techcrunch.com, accessed May 2026.
- Google statement and John Mueller, via PPC Land, "Google disputes Pew study showing AI Overviews reduce clicks by half," ppc.land, accessed May 2026.