When an AI Overview appears, people click less: several measurements agree on that, though not on how much. Cited brands do better than uncited ones, but below what's achieved with no AI Overview, and nobody has shown that the citation is the cause. Nobody we've read has measured lost clicks moving from some sites to others, so this guide no longer claims it.
What has been measured
| Source | Who they are | What was measured | Result |
|---|---|---|---|
| Pew, 22 July 2025 | Research centre | 900 US adults, March 2025 | Clicked a result in 8% of visits with a summary and 15% without. Clicked a link in the summary itself: 1% |
| Seer, September 2025 | Agency | 3,119 informational searches from 42 clients, Q3 2025 | Organic CTR: 0.52% uncited, 0.70% cited, 1.45% with no AI Overview. Year-on-year falls: 65.2%, 49.4% and 46.2% |
| Seer, April 2026 | Agency | 53 brands, 5.47 million searches | −61% CTR from Q3 to Q4 2025 for cited brands, mainly because impressions rose, Seer says. In 2026 the forecast decline has reversed. Without an AI Overview, informational CTR rose in 2025 from 2.93% to 3.97% |
| Ahrefs, 4 February 2026 | Tool vendor | 300,000 keywords, position 1, desktop, December 2023 to December 2025 | CTR from 0.073 to 0.016 with an AI Overview. From 0.076 to 0.039 for informational keywords without. Ahrefs headlines −58%: it compares the actual 0.016 with the 0.037 expected if CTR had fallen as it did for informational keywords without an AI Overview. It's an estimate |
| Sistrix, February 2026 | Tool vendor | Germany, from its keyword data | Just over 20% of keywords with an AI Overview in February. On its page updated on 5 November 2026, around 40% of searches. Organic clicks, from the February study: 57% of searches on average, 33% with an AI Overview. Across all organic clicks, −6.6% |
| Reuters Institute with Chartbeat data | Research institute, vendor data | Over 2,500 news sites, November 2024 to November 2025 | Google organic traffic: −33% globally and −38% in the US. The report says it isn't clear how much of this is down to AI Overviews |
| Google, 6 August 2025 | Interested party | Publishes no data | Total organic clicks "relatively stable" year on year, and it says third-party reports use flawed methods |
Why the figures don't match
You'll read −61%, −58% and "stable", and they don't measure the same thing.
- CTR isn't clicks. For the −61% figure, Seer writes that impressions more than doubled while clicks stayed flat. The rate fell because the denominator grew. Always look at both numbers.
- One position or all. Ahrefs and part of Sistrix's data refer to position 1. In Sistrix the fall there is around 60%, while across all organic clicks in Germany it's 6.6%, as an AI Overview doesn't appear on every search.
- Who's in the sample. An agency's clients, a panel of people, a keyword database or news sites.
- Google talks about Search as a whole and doesn't show the data, so it can't be checked.
Searches without an AI Overview complicate both simple stories, because the data doesn't point the same way. In its September 2025 study, Seer measured a 46.2% fall in Q3 against the previous year on informational searches without an AI Overview, and Ahrefs one from 0.076 to 0.039 between December 2023 and December 2025. In its April 2026 study, with a different sample, Seer sees the opposite across 2025: on informational searches without an AI Overview, CTR rose from 2.93% in January to 3.97% in December. Its hypothesis is a selection effect: the AI Overview takes the searches of people who only want an answer, and those left without one draw people who do intend to click. It adds that CTR fell on commercial searches without an AI Overview, and that AI Overviews aren't the only force at work. This data can't separate the cause.
Does being cited help?
Relatively, yes, with an important caveat. At Seer, searches with a citation have a CTR about a third higher than those without (0.70% against 0.52%), but half that of searches with no AI Overview (1.45%). Seer warns that it can't prove the citation causes the difference, as stronger brands may already have had more clicks and also get cited more. This is data from its clients, on informational searches. In April 2026, across 53 brands, Seer gives 120% more clicks per impression with a citation than without, and still 38% fewer than with no AI Overview. Be careful with that second comparison: according to Seer, AI Overviews appear mainly on query types that already got fewer clicks, as happened with featured snippets. Comparing with searches that have no AI Overview also means comparing different searches.
What it means in Europe
Pew measures the US, and the market isn't stated in what we read from Seer and Ahrefs. For Germany there's Sistrix, with the figures above and sector differences from 1% to over 24% lost. We found no comparable figure for Spain. Don't carry US or German data over to your market as if it were your own: check your Search Console.
It isn't all AI Overviews
Searches with no click are older than that. SparkToro reported 49% in 2019, and in June 2026 gave 68.01% in the US for January to April, using another panel (Similarweb), so the series isn't fully comparable. We cover this separately in zero-click searches and the glossary.
Each tool searches in its own way
These figures are about Google. For other tools we only know what each company says, and none of the official pages we read publishes its criteria for choosing sources:
- ChatGPT uses third-party search providers, as OpenAI announced on 31 October 2024, without saying which.
- Perplexity searches the internet and each answer has numbered citations, according to its help centre.
- Claude can search the web and gives direct citations when it uses that information, as Anthropic announced on 20 March 2025.
What you can do
This is a working method and not a proven lever. We know of no study showing these steps win clicks back.
- Look at clicks and impressions separately in Search Console, against the same period a year earlier, for the reason Seer gives.
- Split three groups: searches with your brand, informational and commercial. A fall in informational searches doesn't mean the same as one in commercial ones.
- Compare with and without AI Overviews, if your tool allows it or by hand for your important searches. For generative AI features, Google points to the Search Console report.
- Don't look for a trick. Google says there are no additional requirements or special optimizations to appear in AI Overviews, and no special markup is needed either, as we explain in our Schema.org guide.
- Decide which pages deserve the effort. If a page only answers something that fits in a summary, ask what else it can offer. This is our judgement.
What nobody knows yet
- Whether a citation without a click brings anything. We looked for a measurement showing brand searches rise after an AI citation and didn't find one, only the possibility Seer raises.
- Whether clicks change hands. The measurements above show fewer clicks on these searches. They don't show where they go, and Pew only says more sessions end right there. Google says traffic is shifting to different sites, such as forums and videos, but publishes no data.
- How long it will last. Seer saw a turn against its forecast in 2026, and the figures may change again.
This is what our searches and sources turned up, not a full literature review.
How to find out for your brand
Search Console gives you the clicks. If you also want to know whether you're mentioned, Truffle, our AI visibility tracker, measures with fixed questions how often a brand appears in answers from ChatGPT, Claude, Gemini and Perplexity and in Google's AI Overviews, and whether they link to it. For those models it uses their APIs, not the app people use. It doesn't measure the effect on your traffic.
FAQ
Do AI Overviews take my traffic?
On searches where they appear, measurements show fewer clicks, and Sistrix calculates −6.6% across all organic clicks in Germany. How much it affects you depends on your searches: check Search Console.
Is being cited worth it?
At Seer, cited brands have a better CTR than uncited ones and a worse one than with no AI Overview, and Seer can't prove the citation is the cause.
Why does Google say traffic is stable?
It refers to total clicks across Search, without publishing the data. That may be compatible with falls on certain searches, but what it has published doesn't let anyone check.
Is there data for the UK or Spain?
We found no comparable figure for Spain, and didn't look for the UK in this article. The closest we read is Sistrix's for Germany.
Do I need special optimization to appear?
Google says there are no additional requirements. It's worth reading other promises of a method with their source alongside.
Sources
Retrieved on 6 November 2026.
- Pew Research Center, 22 July 2025
- Seer: September 2025, Seer: April 2026
- Ahrefs, 4 February 2026
- Sistrix, February 2026 (in German)
- Sistrix: AI Overviews, updated 5 November 2026 (in German)
- Reuters Institute: trends and predictions 2026
- Google: queries and clicks, 6 August 2025, AI features, generative AI guide
- SparkToro: 2019, SparkToro: June 2026
- OpenAI, Perplexity, Anthropic
How this article was made
The previous version was headed "AI Overviews aren't killing your traffic: they're redistributing it" and stated that as fact, though no source showed it. We withdrew it and changed the title and address. This text was rewritten in November 2026 with AI assistance, from the pages linked above, and reviewed before publication. We haven't measured the effect on traffic ourselves, and it contains no client data.
