AI Overviews and Zero-Click Search: How a UK Business Should Measure When Clicks Fall

If you run a UK business with a website, you have probably noticed the same thing at some point in the last year: impressions steady, clicks down. The explanation arriving in your inbox is nearly always the same. AI Overviews are answering the question, so nobody clicks, and your site is a casualty.

Sometimes that is true. Often it is only part of the story, and occasionally it is the wrong story entirely. I would rather you measured the thing than assumed it, so here is what the published numbers actually say, why they disagree, and what a local business should track instead of clicks.

How often do AI Overviews actually appear?

Nobody agrees, and the disagreement is informative. Three sources I read in October 2026 gave three different answers. LOCALiQ's UK guide cites Search Engine Land data of AI Overviews on around 25% of all searches. A Google-focused news site states that they appear in 43% of searches, up from 15% a year earlier. A vendor guide says roughly a third of US English queries.

The spread is not an error. Each figure measures a different set of queries in a different market with a different method. The vendor guide says as much: exact frequency varies by query, market, device and measurement method. For a UK local business, the only number that matters is how often Overviews appear on the searches you care about, and that you can only find by looking.

What does Google say about clicks from AI features?

Google says clicks from pages with AI Overviews are higher quality, meaning users are more likely to spend more time on the site. It also says AI Overviews and AI Mode appear in the overall Web search traffic in Search Console, and that you should consider conversion indicators such as sales, sign-ups and information lookups about your business, not click counts alone.

That is a Google statement and Google has an interest in the framing, so I treat it as a hypothesis worth testing on your own data rather than a fact. But the direction is sensible: if a system answers the simple question itself, the people who do click may be the ones who want more, which could mean fewer visits and a higher share of serious ones.

Is falling click-through always caused by AI?

No. A drop in click-through rate has several possible causes, and an AI feature is only one of them. A rise in impressions from broad queries lowers CTR with no change in clicks. A ranking slip lowers it. A competitor's snippet beats yours. A SERP feature other than an AI Overview, such as a Business Profile or a knowledge panel, takes the click. A seasonal pattern shifts demand.

A vendor guide makes the same point: a low click-through rate alone does not prove an AI feature caused it. I agree, and I would add that this is why the first step is segmentation. Split branded from non-branded queries, split pages, split device, and compare with the same period a year ago, because London local demand is seasonal and a straight before-and-after comparison misleads.

What is happening to clicks overall?

Some publishers are reporting steep falls, but that evidence is about publishers, not local businesses. A Chartbeat study shared with Axios in March 2026 reportedly found search referral traffic fell 60% for small publishers and 47% for mid-sized publishers, as relayed by a news site. That concerns news and content sites whose product is the information itself.

A plumber in Walthamstow is in a different position. Most of their valuable searches are local, and for those the Map Pack and the Business Profile are the main event. Applying publisher statistics to a local service business is the commonest mistake in this conversation. Your risk is real but it looks different: the click you lose may be replaced by a call you never see in your website analytics.

Why do calls matter more than clicks for a local business?

Because a Business Profile can generate calls and direction requests that never touch your website. The Guard Up Pest Control case study on this site shows it: in five months the profile produced 327 phone calls, up 38% on the same period a year earlier, while sending 351 website clicks. The calls bypassed the site entirely.

If you judged that business by website clicks, you would undercount its results by a wide margin. The same logic applies when an AI answer names you and a customer rings. The click never happens, and the enquiry still does. Measuring only clicks makes success look like decline.

What should I measure instead?

1. Calls, direction requests and bookings from your Google Business Profile performance data, month by month.
2. Enquiries by source, using a tracked phone number or a short 'how did you hear about us' field, so you can see AI, Map Pack and organic separately.
3. Search Console, segmented: branded versus non-branded, by page, with the Generative AI performance report for AI feature visibility.
4. Conversion rate on the clicks you do get, since Google says AI-referred visitors may be more engaged.
5. The same period last year, to strip out seasonality.
6. Revenue or booked jobs, the only measure that cannot be gamed.

How do I set up a measurement baseline this week?

1. Export sixteen months of Search Console data, so you can compare this year with last at the same time of year, by page and by query.
2. Split queries into branded and non-branded, and again into informational ('how much does...') and transactional ('emergency plumber near...').
3. Download twelve months of Business Profile performance: calls, direction requests, website clicks and bookings.
4. Add a tracked phone number or a 'how did you hear about us' field to your main contact routes.
5. Screenshot the results page for your ten key searches monthly, noting whether an AI Overview, a Map Pack or neither appears.

That is an afternoon of work, and it gives you something almost no UK small business has: a baseline. Without one, every later statement about AI taking your traffic is a guess, in either direction.

What would tell me AI really is costing me clicks?

A fall concentrated in informational queries where an Overview appears, with branded and local transactional clicks steady. That pattern points at AI features answering the simple question. A fall across everything points at ranking, seasonality or a site problem instead.

Informational pages are the most exposed, because a summary can satisfy a simple question without a visit. Local, transactional searches, where the buyer wants a provider rather than an explanation, are less exposed, since the listing and the call are the product. Knowing which kind of query you depend on tells you how worried to be.

If your baseline shows exactly that pattern, the response is to strengthen the pages that are being summarised rather than to abandon them. Add the depth a summary cannot give, and make sure your name and contact route sit where an interested reader will see them. If it does not show that pattern, the explanation lies elsewhere and the fix does too.

What should I do about it?

Make the answer to a simple question come from you, and make the next step obvious. If an AI Overview answers 'how much does a boiler service cost', you want it to quote your page, and you want your phone number and booking link where an interested reader will see them. Being the cited source is the new top position.

• Write the direct answer yourself, with a number or range, as covered in the post on writing content AI quotes.
• Own the follow-up. The reader who clicks wants depth, so give them something the summary could not: a worked example, a checklist, a comparison.
• Keep your Business Profile complete, because for local searches it is often the listing that wins the action.
• Make contact frictionless: tap-to-call, short forms, visible hours.
• Track the pattern over months, not single weeks.

The honest summary is that AI features change where the attention goes, they do not abolish the customer. The businesses that handle it well will be the ones measuring enquiries and not just visits. If you want help setting that measurement up, it is the first thing a free audit looks at.