Do Google AI Overviews Reduce Clicks to Your Site?

A client asked me the blunt version of this last month. Her organic traffic was down, her rankings weren't, and she'd read a dozen conflicting headlines. Half said AI Overviews were an extinction event for websites. The other half quoted Google saying everything was fine. She wanted to know which one was true, and what she was supposed to actually do about it.

It's the right question, and the honest answer is that both camps are shading it. The data does show AI Overviews depress clicks, clearly enough that pretending otherwise would be dishonest. But the "websites are dead" version oversells it, and Google's "nothing to see here" version leaves out the part where it hasn't shown its work. So let me walk through what the numbers actually say, where they disagree, what Google's own position is, and the specific moves that hold up once you strip out the panic and the spin.

Do AI Overviews actually reduce clicks?

Yes. On the whole, the strongest independent research says AI Overviews reduce clicks to websites, and the cleanest evidence comes from a source with no product to sell. In July 2025, the Pew Research Center studied the real browsing behavior of 900 U.S. adults across nearly 69,000 Google searches. On result pages that contained an AI summary, people clicked a traditional search-result link 8% of the time. On pages without one, they clicked 15% of the time, close to twice as often. And the AI summary didn't hand that click to itself instead: users clicked a link inside the summary just 1% of the time.

Two things make Pew the anchor here. It measures what people did, not what they said they did, drawn from actual clickstream data rather than a survey. And Pew doesn't sell SEO software, so there's no commercial reason to lean the finding one way or the other. When a neutral, behavioral study finds the click rate roughly halving in the presence of an AI summary, that's about as close to a straight answer as this messy field offers.

What The Data Shows

An AI summary nearly halves the click rate.

How often people clicked a traditional search result, measured across nearly 69,000 real Google searches.

Result page WITHOUT an AI summary 15%

Result page WITH an AI summary 8%

Bars scaled to the 15% baseline · Clicks on a link inside the AI summary itself: just 1%.

A neutral, behavioral study, not a survey and not a vendor with software to sell. When the click rate drops from 15% to 8% just because a summary appeared, the burden of proof shifts to anyone calling AI Overviews harmless.

Carie My Marketing

cariemymarketing.com

Source: Pew Research Center, real browsing behavior of 900 U.S. adults across ~69,000 searches (July 2025).

Pew also found the effect doesn't stop at the click. People ended their browsing session entirely 26% of the time after landing on a page with an AI summary, versus 16% without one. The summary doesn't just redirect the click. Sometimes it ends the search.

How much are clicks really dropping?

Here's where you have to hold several numbers at once, because they measure different things and they don't perfectly agree. The click-through hit is real, but its size depends on what exactly you're counting. This table lays the major 2025 and 2026 studies side by side so you can see the shape of it rather than clinging to one scary figure.

The Ahrefs study of 300,000 keywords is the one people quote for the magnitude. In April 2025 it found the presence of an AI Overview correlated with a 34.5% lower click-through rate for the number-one result. An updated run on December 2025 data put that closer to 58%. Worth saying plainly: these are correlational studies from a company that sells SEO tools, comparing the same keywords across time rather than running a controlled experiment. That doesn't make them wrong. It means you read the direction and the scale as strong signal, not as a lab result.

The Magnitude

The hit to the top result nearly doubled in 2025.

Ahrefs' measured drop in click-through rate for the number-one organic result when an AI Overview is present. Same 300,000 keywords, two points in time.

April 2025 −34.5%

December 2025 −58%

Lower click-through rate for the #1 result vs. the same result with no AI Overview · bars scaled to the 58% figure.

Correlational, from a company that sells SEO tools, comparing the same keywords over time rather than a controlled experiment. Read the direction and the scale as strong signal, not as a lab result.

Carie My Marketing

cariemymarketing.com

Source: Ahrefs, 300,000 keywords (April 2025 study · December 2025 update).

Zoom out from any single result to the whole search page and the trend holds. SparkToro's analysis of Similarweb clickstream data found that in the first four months of 2026, 68.01% of U.S. Google searches ended without a click to any website, up from 60.45% in 2024. By that data, only about 276 of every 1,000 searches now send a click to the open web. One honest caveat: zero-click is driven by many things, including featured snippets, direct answers, and Google's own properties, not AI Overviews alone. So it's the best number for the broad "Google keeps more clicks" story, and the wrong number to pin entirely on AI.

The pain concentrates where you'd expect. Similarweb's publisher research found news-related searches ending without a click rose from 56% in May 2024 to 69% in May 2025, with organic search traffic to news sites down roughly 26% over that year. Informational queries are the most exposed, because a definition or a quick fact can be fully answered in the summary. That's the pattern worth internalizing, and I'll come back to it, because it's the difference between which of your pages are at risk and which aren't.

The honest counter-evidence

I'd be selling you a tidier story than the data supports if I stopped there. Semrush's AI Overviews study, published December 2025, ran a same-keyword before-and-after comparison and found zero-click behavior slightly decreased, from 33.75% to 31.53%, when an AI Overview appeared. That cuts against Pew and Ahrefs. It's one study with its own method, and it doesn't overturn the weight of the evidence, but a clean, AI-only causal number is genuinely contested. Anyone giving you a single confident percentage for "how much AI Overviews cost you" is rounding off that uncertainty. The aggregate trend is clear. The precise attribution is not.

Same study, useful context on how often you're even exposed to this: Semrush found AI Overviews appeared in 6.49% of searches in January 2025, spiked to 24.61% in July, and settled to 15.69% by November. Prevalence swings hard by tracker and by month, so treat any single figure as a snapshot. The point is that AI Overviews show up on a real and growing slice of results, not all of them.

What does Google say about it?

Google disputes the whole framing, and its position deserves a fair hearing even though I don't fully buy it. In an August 2025 post, Liz Reid, the head of Google Search, wrote that "total organic click volume from Google Search to websites has been relatively stable year-over-year," and that "average click quality has increased and we're actually sending slightly more quality clicks to websites than a year ago." Her argument is that AI Overviews get people searching more, and that the clicks that do come through are worth more because the person already got the lay of the land and clicked to go deeper.

That's a coherent story, and parts of it are probably true. The problem is what's missing. The post carries no charts, no percentages, no year-over-year figures to support "stable" or "higher quality." And "quality click" is Google's own term, defined on Google's terms. Meanwhile Pew, Ahrefs, and Similarweb have all put actual numbers on the table pointing the other way. So the honest way to hold Google's position is this: it may be right that surviving clicks are more valuable, but it's asking you to take the reassuring half on faith while independent researchers have shown their work on the worrying half.

So how worried should you actually be?

Less panicked than the doom headlines, more proactive than Google would like. The immediate revenue threat for most businesses isn't "an AI chatbot stole my customer." Direct referral traffic from AI answer engines is still small: Conductor's benchmarks, drawn from billions of sessions, put AI referral traffic at 1.08% of total web visits. The bigger, nearer threat is the one hiding in the zero-click numbers: Google is answering more questions on its own results page and keeping the click.

Which means the real question isn't "how much traffic will I lose," it's "which of my pages are actually exposed." A page that ranks for a quick informational question, the kind a summary can fully absorb, is at genuine risk. A page built around commercial or comparison intent, where the searcher still has to reach a specific business to buy or book, holds up far better. The searcher who wants to hire someone can't complete that inside an AI summary. They still have to land somewhere.

This is also why "are we invisible to AI search" and "are we losing clicks to AI Overviews" are two sides of one coin. If you're not even eligible to appear in the answer, you lose both the citation and the click. Before you worry about the click-loss math, it's worth confirming you clear the basic gate at all, which I walk through step by step in whether your site is invisible to AI search. Absence is a bigger problem than a lower click rate, and it's more common than most businesses realize.

What should you do about AI Overviews?

Here's the part your client, and mine, actually want. The good news is that the most defensible playbook comes straight from Google's own documentation, not from a vendor selling an AI-optimization package. Google's guide to optimizing for generative AI features is blunt that appearing in AI Overviews takes no special markup or files, because the AI features run on core Search ranking and quality systems. It's still SEO. These six moves follow from that, in roughly the order I'd tackle them.

1. Confirm you're eligible at all. Per Google's AI features documentation, a page must be indexed and eligible to show with a snippet to appear in AI Overviews or AI Mode. That's the floor. If key pages aren't indexed or snippet-eligible, you're not in the running for the citation or the click. This is where a surprising number of "AI is killing us" problems actually start.
2. Don't suppress your own snippets. It's tempting to think you can opt out of AI Overviews with nosnippet, data-nosnippet, or max-snippet:0. Google documents that those controls do keep you out of AI Overviews, but they also strip your normal search snippet, which usually lowers your click-through rate in regular results too. For most businesses that trades a small AI concern for a real organic-traffic cost. It's a kill switch, not a dial.
3. Shift weight toward intent a summary can't finish. Audit which of your pages target quick informational answers a summary can fully absorb, and which target commercial, local, or comparison intent that still needs your site. Lean your content investment toward the second kind. You don't abandon informational content, it still builds authority and can earn the citation, but you stop expecting it to carry the click on its own.
4. Earn the citation inside the overview. If Google is going to answer the question on its own page, you want to be the source it names and links. That's the whole discipline of getting cited by AI, which I lay out in the guide to generative engine optimization. Being the cited source keeps your brand in front of the searcher even on a query that no longer sends a reliable click.
5. Keep the citation once you have it. A citation you earn today can quietly disappear next month as the engine reaches for something fresher. That decay is its own problem with its own fixes, and I got into the data on it in why AI citations have a shelf life. Showing up once isn't the same as staying shown.
6. Build to close the clicks that survive. Remember the part Google was right about: the click that makes it through an AI Overview is a more decided visitor. That changes what your landing page has to do. It's no longer there to convince a cold browser, it's there to let a ready buyer act without friction. I wrote about that shift in detail in why AI search sends ready buyers your site may not be built to close. Fewer clicks that convert beats more clicks that bounce.

On measurement, there's finally a real tool, with a real limit. Google announced a Generative AI performance report in Search Console in June 2026 that shows how many times your links appeared in AI Overviews and AI Mode. The catch: it reports impressions only, no clicks and no click-through rate, and it's still in limited rollout, so you may not have it yet. It proves you're showing up. It can't yet tell you what that visibility is worth in traffic. Measuring the real impact of AI Overviews still means combining a few imperfect signals rather than reading one clean number.

If you're staring at a traffic chart trying to work out how much of the dip is AI Overviews, how much is a ranking change, and what to actually do first, that's exactly the kind of untangling an engagement is built for. It's rarely one cause, and the fix is rarely one lever.

The click didn't disappear. It got scarcer and more valuable. Stop mourning the volume and start earning the ones that still convert.