AI Overviews UK Searches: The 2026 Numbers, Sourced

Every UK AI Overviews figure in this post carries a source and a month. Here's what has been measured, what it misses, and how to find your own number.

Take an illustrative case. You're at your desk in Stowmarket with Search Console open. Rankings haven't moved. Impressions are flat. Yet clicks are down by about a third. Then a business partner forwards a screenshot: "AI Overviews now appear on 41% of UK searches." What are you spending to fix it?

That screenshot has no source, no date and no method. Most of the figures going round are the same. FutureProofs is a Webflow and SEO agency, so we gain when you take search seriously, and you should read this with that bias in view, because a post telling you to spend on everything would just be an advert. We'll also say where the numbers justify spending nothing. SEO is already in every plan, so there's no AI add-on to sell.

How many UK searches show an AI Overview?

Around 30%, per Ofcom's Online Nation report from December 2025. It's the only regulator-published UK figure we know of, and it predates Google's latest rollouts.

Here's what has been measured for the UK, next to what each figure leaves out.

Why Ofcom is the number to put in front of a partner

Ofcom is a regulator with nothing to sell on the back of the result, which is more than most vendor blogs can say. The catch is speed. Regulators publish slowly, so the figure describes late 2025, and today's real number is probably higher.

Say so when you quote it. "About 30% in December 2025, per Ofcom" survives a sceptical partner. "30% of searches" doesn't.

The zero-click figure, and what it actually says

The SparkToro and Similarweb study covered six countries. The UK came out highest, with 69.5% of searches ending without a click. The US followed at 68.0%, then France at 65.3%, Canada at 63.8%, Italy at 63.4% and Germany at 62.1%.

The same study counted 232 clicks per 1,000 UK searches reaching a site Google doesn't own. Zero-click isn't new, and AI Overviews don't explain all of it. Nobody ever needed a click to check the weather. Still, 232 in 1,000 shows how thin the slice left for your website has become.

Where does the 41% figure come from?

We couldn't trace it to any named study, sample size or measurement window. It circulates in roundups that cite other roundups. If you find the original, send it over and we'll rewrite this section.

The real measurements disagree with each other, though not because any of them is wrong. Each tracks a different set of keywords.

Four keyword sets, four different things.

A percentage of tracked keywords isn't a percentage of searches. Vendor sets tend to skew towards informational terms, where Overviews fire most often, and none of them is a UK-only count of everything people type. So don't multiply, divide or "correct" them. Use Ofcom for a UK headline and treat the vendor figures as snapshots of what each vendor happens to watch.

How far have click-through rates really fallen?

For the top organic result, by 58% on queries that show an Overview. That's Ahrefs, February 2026, and the decline built up through 2025.

The top result

Ahrefs found that when an Overview appears, the top organic result keeps 42 of every 100 clicks it once got. Second place fell by 50.8% and third by 46.4%. An earlier Ahrefs study of 300,000 keywords, covering March 2024 to March 2025, put the top-result fall at 34.5%. The gap widened within a year.

What a click inside the Overview is worth

Pew Research tracked a US browsing panel in July 2025. About 8% of visits ended in a click on a result when an Overview appeared, against 15% without one. Just 1% clicked a link inside the Overview itself.

Seer Interactive saw something that cuts the other way. Brands cited in an Overview earned roughly 35% more organic clicks and about 91% more paid clicks than uncited ones (September 2025). Being named helps. Being absent from a crowded answer is where the cost lands.

The caveat nobody prints

These are averages across big keyword sets, mostly high-impression queries and often US panels. They aren't your village butcher's numbers. A 58% fall on a national average shows the direction. It won't tell you what happens to a plumber's page for "boiler service Sudbury".

Timing matters too. If your clicks dropped around the May 2026 core update, work through ruling out the core update first before you blame AI.

Which of your searches are exposed, and why do local ones behave differently?

Exposure follows query type, and the national average tells you little. Informational questions take the biggest hit. Buying-intent searches take far less, and a "service plus town" search is usually purchase-ready.

Informational and advice queries

This is where Overviews are densest. BrightEdge's sector data (February 2025 to February 2026) had healthcare at 88%, a sign that advice-heavy topics get answered on the results page itself. If your traffic came from explainer articles, you're in the exposed group.

Transactional and product queries

Google has pulled back here. In the same BrightEdge data, ecommerce and transactional coverage sat at about 4%, down from 29% a year earlier. People about to buy get shopping results and listings rather than a summary paragraph.

Local queries

We haven't found a dated UK measurement of how often Overviews appear on "service plus town" searches. So be wary of any local percentage you see, and note that we aren't offering one. What's safe to say is that these searches tend to be close to a purchase, which is where Google has been showing fewer summaries. Whether you're named when one does appear is the question that matters, and you can only answer it by checking your own searches by hand, as below.

AI Mode is a separate surface with its own behaviour, so keep its numbers out of Overview figures. We cover that split in our post on the May 2026 AI Mode update. Which pages get named in either one, we can't promise.

Two businesses, opposite problems

A hypothetical. A clinic in Newmarket gets its enquiries from "treatment plus town" searches. A firm in Colchester gets its traffic from a library of explainer articles. Same month, same rollout, opposite problems. The clinic needs to check whether it's named. The firm needs to decide whether those articles ever produced an enquiry. For the first case, SEO for Newmarket businesses covers what we do.

How do you work out your own number in Search Console?

Compare query-level click-through year on year, then cross-reference with the generative AI performance report. Google's Search Console help says what that report does and doesn't include, so read it before you lean on it. We've gone through it in our Search Console AI reports post.

1. Set a year-on-year comparison (10 minutes)

In the Performance report, compare the last 3 months with the same 3 months a year earlier. Select the UK as the country.

2. Work at query level (15 minutes)

Sitewide numbers hide everything. Sort queries by click change, export the top 50 losers, and mark each one as informational or a buying search.

3. Open the generative AI report (10 minutes)

Note which of your pages appear in it. A thin list tells you nothing yet, because you're only building a shortlist.

4. Cross-reference (15 minutes)

A page that shows up in the AI report while its click-through falls is exposed. A page with falling clicks and no sign in the AI report probably has another problem, such as the core update or a ranking drop.

5. Hand-check five money queries (10 minutes)

Pick the five searches that most often lead to enquiries. Run them from a UK phone, logged out, and note whether an Overview appears and who it names.

6. Repeat quarterly

Write the date on the spreadsheet. Undated numbers are how the 41% claim got started.

What do these numbers justify spending on, and where are we the wrong answer?

They justify finding which pages are exposed and fixing those. They don't justify buying a tool on the strength of an unsourced percentage.

Worth doing

Pages that answer what people ring you about, with prices and service areas in plain words. Add structured data so Google knows what each page is, which we cover in why your website needs schema markup. Pages that earn enquiries are specific and local, and our SEO service is built around them.

Not worth doing

Buying "guaranteed" AI placements, because nobody can guarantee one. Rewriting the whole site for AI. Chasing citations on informational queries that never produced an enquiry. Google's own advice is in our write-up of its first AEO guide.

Where we're the wrong answer

If your traffic is mostly local and flat, leave it alone. If you serve one town and already rank, you don't need us. And if your enquiries came from explainer content, no plan will bring that traffic back.

What we charge, plainly

SEO is included in every plan and isn't sold separately. Starter is from £189 a month, Business from £249 a month and Market Leader from £329 a month. It's all set out on our pricing page.

Frequently asked questions

What percentage of UK searches have AI Overviews?

About 30%, per Ofcom's Online Nation report of December 2025. Ofcom is a regulator with nothing to sell, which is why this is the most defensible UK figure. It's dated, though, so today's real number is probably higher. Vendor figures run from 9.46% to about 48% because each tracks a different keyword set. Quote Ofcom, with the month.

Is the 41% AI Overviews figure real?

We can't trace it. There's no named study, sample size or measurement window behind it, and it mostly turns up in roundups citing other roundups. It sits inside the range of vendor figures, so it sounds plausible. Plausible isn't sourced. Use the Ofcom number and date it.

What is the UK zero-click search rate in 2026?

69.5% of UK Google searches ended without a click between January and April 2026, per SparkToro and Similarweb (published 16 June 2026). That's the highest of the six countries studied, with the US next at 68.0%. The same study counted 232 clicks per 1,000 UK searches reaching a site Google doesn't own.

How much traffic have AI Overviews cost small businesses?

It depends on the page. Ahrefs measured a 58% fall in clicks to the top organic result where an Overview showed (February 2026), but that's an average across large keyword sets. We haven't found a UK study of small-business traffic. Your own year-on-year Search Console data, at query level, is the only figure that applies to you.

Do AI Overviews affect local searches in Suffolk?

Possibly, and we can't give you a dated percentage for it. We haven't found a UK measurement for "service plus town" searches, so we won't invent one. The useful question is whether your business is named when an Overview does appear. Check five of your main searches by hand, from a UK phone, logged out, and record what shows.

How do I check AI Overviews in Search Console?

Start in the Performance report and compare click-through year on year at query level. Then open the generative AI performance report and note which of your pages appear in it. A page that appears there while its click-through falls is the one to look at first. Date the findings and repeat every quarter.

What to do on Monday morning

Pull the 20 queries that actually produce enquiries for you. Compare click-through with the same period last year and note which pages are exposed. Then decide what each exposed page was ever for. An explainer that never produced a call isn't worth rescuing.

Write the date on your numbers and recheck in January. It's dull. It's also the only way to end up with a figure you can trust.

Send us your current site and the towns you cover. We'll tell you which pages are exposed, even if you never hire us. That works as well from Bury St Edmunds as from Colchester or anywhere between. Start here.

Written by the FutureProofs team in Suffolk. Updated 10 October 2026.