Are You Getting Fewer Clicks?

And if you are, would your reports tell you?

The UK has the highest zero-click rate of six countries measured, a controlled experiment now puts a causal figure on what AI Overviews do to clicks, and standard reporting does not separate the effect. If it has not reached your pipeline yet, it might - and the time to test an alternative is while you do not need one.

01

Nearly 7 in 10 British searches end without a click

More searches in Britain end without a click than in any other country measured.

In early 2026, 69.5% of UK Google searches ended without a click. That was the highest of six countries measured, above the US at 68.0% and Germany at 62.1%. So your organic channel sits in the market where a search is least likely to end in a visit. Whether that is costing you yet is a separate question. This briefing helps you answer it. If organic carries a large share of your pipeline, a second route to buyers that does not depend on being found gives you somewhere else to turn.

Further reading

The figure measures the whole market, not any one site. It says nothing yet about your own traffic, which may be holding up. What it does set is the backdrop. More of the answer now arrives on the results page, before the click, here than anywhere else measured.

The evidence. SparkToro's analysis of Similarweb's desktop and mobile panel, January to April 2026, published June 2026. Searches ending without a click: UK 69.5%, US 68.0%, France 65.3%, Canada 63.8%, Italy 63.4%, Germany 62.1%. For every 1,000 UK searches, 232 clicks went on to the open web, a figure that includes paid ads.

The panel does not include Google's own mobile app, and SparkToro expects the true share to be, if anything, higher. This is the first UK figure from this source, so it shows a level, not a trend.

Sources
  • SparkToro (Rand Fishkin), Zero-click searches highest in the UK, lowest in Germany, June 2026, using Similarweb panel data for January-April 2026.
02

Check yours in 60 seconds

Six questions. About a minute.

Six questions on how your pipeline depends on search. The first is the question this briefing asks. The other five gauge your exposure. You get a read you could take into a planning conversation.

1Have your organic search clicks fallen in the past year?
2Does more than 40% of your qualified pipeline start in organic search?
3Do non-branded searches bring you more visits than searches for your name?
4Is your content programme your main source of new prospects?
5Would your current reporting struggle to show how many clicks AI answers are costing you?
6If organic search delivered a third fewer leads next quarter, would you struggle to replace them?

0 of 6 answered

03

When Google answers the question itself, people click about 40% less

The only controlled test published so far found a large fall in clicks.

A pre-registered field experiment hid AI Overviews from a random group of users. Where an AI Overview appeared, organic clicks fell 39.8%. Zero-click searches rose 34.5%. Unlike the industry studies before it, this measures cause, not correlation. That is the case for a second route to buyers - one that does not depend on Google passing on the click.

Further reading

Exposure follows intent. In August 2026, a UK test of 304 B2B queries found AI Overviews on 60% of supplier searches. They appeared on 96% of pricing searches and 88% of comparison searches. But they appeared on only 6% of brand searches and 4% of 'near me' searches.

Search Console data from 18 B2B websites tells a similar story. At the same ranking, click-through on research queries fell 38% in a year. On commercial queries it fell 20%. The final buying search remains largely clear. The research stage, where the shortlist forms, does not.

The evidence. Agarwal (Indian School of Business) and Sen (Carnegie Mellon), The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment, July 2026. A pre-registered randomised trial: 1,065 US participants, desktop, January-February 2026, everyday searches across all topics. Where an AI Overview appeared, outbound organic clicks fell 39.8% (0.62 to 0.37 per search) and zero-click searches rose 34.5%. AI Overviews appeared on about 41% of searches. This is a working paper and not yet peer-reviewed.

The UK Competition and Markets Authority (January 2026) has reviewed a randomised trial by Google showing fewer clicks to websites when an AI Overview appears; the size is redacted.

AEOscore tested 304 UK B2B searches once, on desktop, in August 2026 - a small sample. Grey Matter's figures come from 18 B2B accounts' Search Console data, Q2 2025 against Q2 2026; the country is not stated.

Sources
  • Agarwal, S. and Sen, A., The Impact of Google AI Overviews on Publisher Traffic and User Experience: Evidence from a Field Experiment, SSRN 6513059, 8 July 2026 (working paper).
  • Competition and Markets Authority, Publisher conduct requirement, 28 January 2026.
  • AEOscore, AI Overviews and click-through rates in UK B2B search, 17 August 2026.
  • Grey Matter, The Great Declick, July 2026.
04

You can rank first and still lose the click

The research still happens. The website visit does not.

Google gives every link inside an AI Overview the same position. So a cited page can report a top ranking while the answer takes the click. GA4 counts AI Overview and AI Mode clicks as ordinary organic search. Search Console's new AI report shows appearances, not clicks. Rankings can hold while clicks fall, and standard reporting won't tell you why. That is one more reason to have a second route to buyers - one whose results you can measure directly.

Further reading

Fewer than 1 in 10 marketers say they can measure everything that matters in AI search. Not all AI traffic is hidden. ChatGPT now tags the clicks it sends, and many assistants pass a referrer. The gap is the reader who sees the answer and does not click. They come back later by searching your name. In your reports, that looks like brand search, not AI.

The evidence. Google's Search Console help: an AI Overview occupies a single position, and all links in it are assigned that same position. Google's GA4 documentation counts traffic from AI Overviews and AI Mode within Organic Search; its AI assistant channel excludes them. Search Console's generative AI report, available worldwide from 31 August 2026, shows impressions, not clicks.

Semrush surveyed 481 marketers (June 2026): 9% said they could measure all the AI-search metrics that matter. Ahrefs' tests (May 2025) found some AI tools pass no referrer, so the visit is recorded as direct; others do pass one, and ChatGPT adds a tracking tag to the links it sends.

Sources
  • Google Search Console Help, What are impressions, position, and clicks?, consulted October 2026.
  • Google Analytics Help, [GA4] Default channel group, consulted October 2026.
  • Google Search Central, Generative AI performance reports in Search Console, June 2026.
  • Semrush, The operational gap: AI SEO study, 3 June 2026.
  • Ahrefs, Generative engines are breaking web analytics, 26 May 2025.
05

The AI now decides who gets the visit

Buyers still visit supplier websites. Increasingly, they visit the ones an AI named.

AI sits in front of the website, not instead of it. In a survey of 519 US B2B professionals, 71% visited a vendor's site after an AI mentioned it, and 63% searched for it on Google. You cannot control whether you are named. And the space around the answer is already for sale. ChatGPT has carried ads in the UK since August 2026, now bought by bidding per click. Google sells the space around its AI answers through its normal search auction. A second route to buyers, one that does not depend on being named, takes some of that pressure off.

Further reading

Paid search began as free visibility and became an auction. A bigger budget buys more of the space, and Google's automated bidding needs a steady flow of conversions before it works well. The answer engines are on the same path from free to paid. An auction is how an advertising business earns the most from limited space, so that seems the likely end point. If so, visibility in AI answers may go the same way: to whoever can afford to bid most. Not every engine has gone this way. Perplexity tried ads and dropped them. But the largest have started.

Even being named is a weak substitute for the click. On the controlled evidence, citations inside AI answers send very little traffic.

The evidence. TrustRadius: 21% of software buyers used generative AI in their buying process in its 2024 study; 63% had used AI at some point in its 2026 study of 1,862 buyers - AI tools were the only research source consulted more often. Semrush (July 2026): 519 US B2B professionals who use AI for work; after an AI mentioned a vendor, 71% visited its website and 63% searched for it on Google. Most buyers check what AI tells them: 94% of those using it said they fact-check it (TrustRadius).

UK Department for Science, Innovation and Technology, UK Business Data Survey 2026 (4,450 businesses): the most common use of AI was 'to research information (28%, for example, in place of a traditional search engine such as Google)'. Ofcom (April 2026): 54% of UK adults use AI tools.

OpenAI began testing ads in the US in February 2026, launched them in the UK in August 2026 on its free and low-cost plans, and introduced cost-per-click bidding in May 2026. Google's help pages state that ads above and below AI Overviews 'serve following the existing auction ranking system'; ads inside AI Overviews run in the US and other English-speaking markets, not yet the UK. Google's automated Target ROAS bidding asks for at least 15 conversions in 30 days. Perplexity began sponsored questions in 2024 and later dropped them.

Sources
  • TrustRadius, The 2026 B2B Buying Disconnect, July 2026; and TrustRadius, generative AI in B2B buying, July 2024.
  • Semrush, How AI shapes B2B buying, 8 July 2026.
  • Department for Science, Innovation and Technology, UK Business Data Survey 2026, 18 June 2026.
  • Ofcom, Adults' Media Use and Attitudes 2026, April 2026.
  • OpenAI, Testing ads in ChatGPT, 9 February 2026, and later updates; Expanding how businesses can buy ChatGPT ads, May 2026.
  • Google Ads Help, About ads and AI Overviews and About Target ROAS bidding, consulted October 2026.
06

Most of next year's customers aren't searching yet

Search and AI answers only reach people who are already looking. Most are not.

Demand capture reaches only the in-market minority. Professor John Dawes estimates around 5% of business buyers in a typical quarter. He offers it as a rule of thumb, not a measurement. In 6sense's 2025 study, buyers had four of the five suppliers they considered on their list from day one. Testing direct access while search still performs gives you a baseline before you might need one.

Further reading

6sense has its own reading of its data. It suggests the visits AI is absorbing are 'more plausibly' from people not yet buying. Those are tomorrow's buyers, not today's. If that is right, the effect lands first on pipeline you have not yet seen. Direct access reaches the same people before they search, by name, on a timetable you set.

The evidence. Professor John Dawes (Ehrenberg-Bass Institute), for the LinkedIn B2B Institute, 2021: 'only 20% of business buyers are "in the market" over the course of an entire year; something like 5% in a quarter'. Dawes describes the 95% figure as 'a heuristic', not a precise rule.

6sense, 2025 Buyer Experience Report (nearly 4,000 buyers, a fifth of them in the UK and Ireland): buyers place about four of the five vendors they evaluate on their shortlist from day one. 6sense's interpretation that traffic lost to AI tools is more plausibly future buyers is its own reading, not a measurement.

Sources
  • Dawes, J., Advertising effectiveness and the 95-5 rule: most B2B buyers are not in the market right now, The B2B Institute, 2021.
  • 6sense, 2025 Buyer Experience Report, November 2025.
07

A route to new customers that nobody else controls

Approaching decision-makers directly, by name. No algorithm or auction decides who sees it.

Direct access means approaching decision-makers directly, by name. No algorithm or auction decides who sees the message. Start from the outcome you are accountable for, and tell us what it is. We will help you choose the data most likely to deliver it. Provenance is the one part of a data purchase you cannot check by sampling. The documentation supplied with every licence is what makes the purchase defensible.

Further reading

The useful conversation starts with the result you need, not with a list. We work back from it to the people who need to hear from you, and the data most likely to reach them. If a target looks hard to reach, we will say so, and suggest what would work better.

The evidence. The Information Commissioner's Office guidance on business-to-business marketing: 'The PECR rule on direct marketing by electronic mail does not apply to corporate subscribers.' You must not disguise your identity, must give a valid address to opt out, and should comply with an opt-out request. Sole traders and some partnerships are treated like individuals and need consent or the 'soft opt-in'. For calls, numbers must be screened against the Telephone Preference Service and Corporate TPS. 'If you are intending to send marketing by post then PECR does not apply.' Named business contacts are still personal data under UK GDPR, and the right to object is absolute. The ICO notes the guidance is under review following the Data (Use and Access) Act 2025, which lists direct marketing as an example of a legitimate interest.

Sources
  • Information Commissioner's Office, Business-to-business marketing, consulted October 2026.
  • Information Commissioner's Office, The Data (Use and Access) Act 2025: what does it mean for organisations?, 2025.

Start with the outcome.

Tell us what the campaign has to deliver. We will come back with the decision-makers and data most likely to deliver it - and say plainly if the target looks hard to reach.

Tell us what you need
  • Over 1.3 millionNamed decision-makers, reachable directly.
  • Over 340,000The largest UK trading companies they sit inside.
  • Over 3.6 millionFurther UK companies also contactable.
  • 1992Researching UK business data ever since, continually verified.
  • 94 daysAverage data age.
  • 2-for-1Goneaway Guarantee.

Related briefings

This argument builds on work published earlier in the 2026 series. None of it is required reading, and each stands on its own.