What Google AI Overviews Are Doing to Your Clicks

When Google places an AI Overview above your result, roughly half your clicks disappear. In the clearest behavioural study to date, Pew Research Center found that users clicked a traditional search result in only 8% of searches that returned an AI summary, versus 15% of searches with no summary. That is about a 47% relative drop in the click rate. The AI Overview answers the question on the results page, so fewer people need to visit your site. This article sets out what the data actually shows, why your Search Console impressions can stay flat while your clicks fall, how to tell whether AI Overviews are behind it, which queries lose the most, and what you can realistically do about it.

What are Google AI Overviews doing to my clicks?

Three independent studies, using three different methods, all point the same way: AI Overviews suppress clicks to the underlying results.

The clearest evidence is behavioural. Pew analysed 68,879 unique Google searches from 900 US adults in March 2025 and watched what real people did. When an AI summary appeared, they clicked a normal search result in just 8% of searches, compared with 15% when no summary was present. Clicks on the links inside the AI summary itself were rarer still, happening in only 1% of visits to a page that carried a summary. So the AI Overview does not redistribute your click to a link within the box. It mostly removes the click altogether.

The second line of evidence is CTR correlation at scale. Ahrefs studied 300,000 keywords in Search Console, split evenly between those with and without AI Overviews. In April 2025 it found that the presence of an AI Overview correlated with a 34.5% lower clickthrough rate for the position-one result, compared with similar informational keywords that had no Overview. For those AI Overview keywords, position-one CTR came in at an actual 0.026 against a forecast of 0.040 without the Overview. By December 2025 Ahrefs had widened that to a 58% lower CTR, with position-one CTR dropping to an actual 0.016 versus a projected 0.037. The effect is not static. It is getting worse as AI Overviews mature and expand.

The third line is zero-click measurement. Similarweb's panel data showed Google searches ending with no click at all rising from 56% to 69% in the year after AI Overviews launched, a 13-point jump. Different tool, different definition, same direction.

The headline numbers differ because the methods differ. Pew is a behavioural panel measuring what real users did. Ahrefs is a keyword-level correlation of Search Console CTR. Similarweb is panel-based zero-click measurement. You should not average these into a single figure, and you should be sceptical of anyone who does. The robust finding is not any one number. It is that all three, built independently, agree that AI Overviews cut clicks.

There is a second, quieter effect worth naming. AI Overviews do not just suppress the first click. In the Pew data, users ended their browsing session entirely after 26% of pages that carried an AI summary, versus 16% of pages with only traditional results. So the summary does not just cost you the visit. It shortens the whole journey, which erodes onward browsing and, for anyone downstream of that search, referral traffic too.

None of this is a niche concern. Per Alphabet's Q2 2025 earnings, AI Overviews reached 2 billion monthly users across 200 countries by July 2025, up from 1.5 billion in May. This is a default search surface now, not an experiment you can wait out.

How big is the click loss, really?

Here is the same evidence laid out side by side, so you can see the direction without conflating the numbers.

Why are my impressions steady but my clicks falling?

This is the pattern most publishers actually notice first, and it is the single clearest fingerprint of AI Overview click loss. In Search Console you see impressions holding steady or even rising on informational queries, while clicks and CTR fall. That divergence, flat or up impressions with down clicks, is the signature.

The mechanism is straightforward. When Google builds an AI Overview and cites your page inside it, your URL still appears on the results surface, so Search Console still records an impression. But the user reads the answer in the Overview and does not click through. Google showed your page. It just did not send the visitor. Impression preserved, click lost.

This is compounded by how Search Console counts AI Overviews. Generative-AI data began appearing in the Performance report from around 16 June 2025, but an AI Overview is treated as a single position, and every source link inside it inherits that same position value. So a citation buried in the Overview can report the same position as a top blue link. That distorts your average-position metric and makes CTR look artificially low for those queries, because Google is blending a low-click Overview citation into the same position bucket as a normal ranking.

There is a further wrinkle. If the same URL appears in both the AI Overview and the blue links on one search, Search Console counts it as a single impression, not two. That deflates apparent impression counts and makes any attempt to reconcile Overview visibility against blue-link visibility unreliable.

The net effect for a publisher is a set of numbers that no longer mean what they used to. A steady impression count can hide a real decline in visibility, and a falling CTR can reflect Google's position-blending rather than anything wrong with your title or snippet. Before you rewrite a title tag in a panic, check whether the query returns an AI Overview. Often the CTR did not fall because your listing got worse. It fell because an AI answer went in above it.

How do I tell if AI Overviews are taking my traffic?

The honest answer is that no single tool shows you this cleanly, and you should distrust anyone who claims otherwise. You have to triangulate.

Start with what GA4 cannot do. AI Overview clicks that do land on your site arrive as ordinary Google organic search referrals. GA4 has no dimension that says this session came from an AI Overview, so you cannot isolate Overview-driven traffic in GA4 at all. GA4 tells you what landed. It cannot tell you what Google showed but did not send.

That missing half is exactly what Search Console holds: impressions and CTR for queries where you appeared but the user did not click. So the method is to combine the two. Read GA4 landings against Search Console impressions and CTR for the same queries and pages. Where Search Console shows impressions steady or up while clicks and CTR fall, and GA4 confirms landings from those queries dropping, you are looking at AI Overview click decoupling.

Do not expect Google's own AI reporting to rescue you here. The dedicated generative-AI reporting was initially impressions-focused. An impression fires when your URL appears inside an AI feature, such as a citation in an AI Overview or a referenced source in AI Mode, but that reporting has not given a separate clicks, CTR, position or query breakdown for AI features on its own. In other words, Google will confirm you appeared inside AI, but it will not tell you how much that appearance cost you in clicks. You still have to infer the loss from the impressions-clicks divergence.

A practical, repeatable check:

1. In Search Console, pull your top informational queries and pages over the last 12 to 16 months.
2. Flag any where impressions are flat or rising but clicks and CTR are clearly falling. That divergence is your candidate list.
3. Manually search a sample of those queries and note which ones now return an AI Overview. If your suspect queries reliably trigger Overviews, you have your answer.
4. Cross-check GA4: are organic landings for those pages falling in line with the CTR drop, even though your ranking position has not?

This is deliberately manual work, and doing it monthly across a large site is tedious. It is also exactly the read that a purpose-built dashboard is meant to automate: showing you the impressions-holding, clicks-falling divergence and the AI-referred slice of traffic in one view rather than three spreadsheets. If you want that read on your own GA4 without adding a tracking script, you can see it in the free Ramprt demo, which reads your existing GA4 read-only and surfaces the AI Overview decoupling directly.

What is the difference between AI Overviews and AI Mode in the data?

They are counted differently, and the distinction matters if you are trying to reconcile reports. An AI Overview occupies a single position, with every link inside it sharing that position. AI Mode does not. Each component of an AI Mode answer, such as a link card, image block or carousel, is positioned under Google's standard element rules, and each follow-up question in an AI Mode conversation is treated as a separate query with its own attribution. So AI Mode fragments into many positioned components and queries, while an AI Overview collapses into one. When you compare AI feature performance, you are not comparing like with like, and Google's tools do not currently let you split them apart cleanly. AI Mode itself is already substantial, with over 100 million monthly users in the US and India by mid-2025.

Which queries lose the most clicks to AI Overviews?

Two things drive where the click loss concentrates: whether the query triggers an Overview at all, and where you rank relative to it.

On triggering, informational queries are the front line. In the Pew study, 18% of all Google searches produced an AI summary, and 88% of those summaries cited three or more sources. The exposure rises sharply with query type: Pew found that 60% of question-format searches and 53% of longer searches of ten words or more produced a summary, against just 8% of one-to-two-word searches. The queries most exposed are the ones an AI answer can satisfy in a sentence or two: definitions, how-to steps, quick facts, comparisons, and the classic what-is and how-do-I questions that make up a large share of many publishers' informational traffic. If your top pages win exactly those queries, they are the ones most at risk.

On ranking position, the intuitive assumption is that only position one suffers. That is wrong. Ahrefs' December 2025 per-position data shows CTR declines across positions two through ten when an AI Overview is present. The drop runs from roughly minus 50.8% at position two down to minus 19.4% at position ten. Ranking below the Overview does not shelter a page. The whole set of blue links sits lower on the screen and gets clicked less, because the answer has already been given above them.

A worked example makes the compounding clear. Suppose a page ranks second for a strong informational query with 100,000 monthly impressions. Before AI Overviews, at an illustrative 10% CTR, that is 10,000 clicks a month. If that query now returns an AI Overview and position-two CTR falls by roughly half in line with the Ahrefs figure, the same page earns around 5,000 clicks from the same impressions. The impression count in Search Console has not moved. Half the traffic has gone. Multiply that across every informational page on a large site and you have the quiet, steady erosion that so many publishers are now reporting, with no ranking drop to explain it.

Can I recover clicks lost to AI Overviews?

Be honest with yourself about the ceiling first. When the answer is fully contained in the Overview and the user's need is a single fact, that click is not coming back through any on-page tactic. The research is unambiguous that clicks on links inside the summary happen in only 1% of visits, so being cited in the Overview is not a click-recovery strategy in itself. The realistic goal is to protect and grow the traffic that AI answers cannot easily absorb, and to measure the loss precisely enough to make good decisions.

Practical priorities, in order:

1. Measure net clicks, not rankings or impressions. The whole point of the impressions-flat, clicks-down pattern is that your old success metrics now lie to you. Track clicks and CTR by query and page, and watch specifically for the divergence. This is the first thing to get right, because every other decision depends on reading the loss correctly.
2. Identify your Overview-exposed pages. Build the candidate list from the Search Console divergence, confirm which queries trigger Overviews, and prioritise by how much traffic is at risk. You cannot defend what you have not measured.
3. Prioritise intent an AI answer cannot satisfy. Queries that need depth, judgement, first-hand experience, original data, tools, calculators, community, or up-to-the-minute information are less easily answered in two sentences by an Overview. Shifting weight toward that kind of content is the most durable response to a summarising search engine.
4. Strengthen the value of the clicks you still get. If AI Overviews are compressing your top-of-funnel informational traffic, the visitors who do arrive matter more. For an ad-funded publisher, that means understanding which surviving traffic actually earns, and where AI-referred visitors sit against your site average, so you can invest in the segments that hold up.
5. Do not chase the summary. Rewriting a title or reworking a snippet to win an Overview citation does not recover the click, because in-summary clicks are vanishingly rare. Effort is better spent on content and formats that keep a human reason to visit.

The uncomfortable truth is that this is a structural shift in how search distributes attention, not a bug you can patch. AI Overviews are a 2-billion-user default, the click suppression is measurable and worsening, and the reporting tools were not built to show it to you cleanly. The publishers who cope best will be the ones who see the decoupling early, in their own data, and reallocate accordingly. For a fuller picture of how AI is reshaping publisher traffic beyond Google, see our guides on what AI is doing to your search traffic and how to read your GA4 for AI referrals.

Frequently asked questions

Do AI Overviews reduce clicks by 34.5% or 58%?

Both figures are Ahrefs' own, at different points in time. In April 2025 it measured a 34.5% lower position-one CTR when an AI Overview was present. By December 2025 it had widened to 58%. The two numbers together tell the real story: the effect is worsening, not a fixed penalty.

Can I see AI Overview traffic in GA4?

No. AI Overview clicks that reach your site arrive as ordinary Google organic search referrals, and GA4 has no dimension identifying them as coming from an Overview. To see the effect you have to combine GA4 landings with Search Console impressions and CTR, which together reveal what Google showed but did not send.

Why are my Search Console impressions flat while clicks fall?

Because Google is still showing your page inside the AI Overview, so it still counts an impression, but the user reads the answer in the Overview and does not click through. That flat-or-rising impressions with falling clicks and CTR is the clearest fingerprint of AI Overview click decoupling.

Does ranking below the AI Overview protect my page?

No. Ahrefs' December 2025 data shows CTR declines across positions two through ten when an Overview is present, from about minus 50.8% at position two to minus 19.4% at position ten. Every blue link sits lower on the screen and gets clicked less once an AI answer appears above it.

Can I filter AI Overviews out of Search Console?

No. An AI Overview is treated as a single position and its link clicks and impressions are blended into your normal Performance totals, so they cannot be filtered out separately. If a URL appears in both the Overview and the blue links on one search, it is counted as a single impression, which further muddies the picture.

Are AI Overviews common enough to worry about?

Yes. Pew found 18% of all searches in its study returned an AI summary, and Alphabet reported AI Overviews reached 2 billion monthly users across 200 countries by July 2025. This is a default search surface, so most publishers with informational content are already exposed.