Zero-Click Britain: Reading the 2026 Search Console Signals Behind AI Overviews

AI Overviews have changed the shape of UK search results, but the aggregate Search Console chart is an unreliable witness. Clicks can fall while impressions rise, and a sitewide trend can conceal a sharp split between informational pages that lose visibility and commercial pages that still earn attention. The useful question in 2026 is not whether AI Overviews ate clicks, but which query cohorts, page types and devices now produce a different click yield — and how that changes measurement and prioritisation.

The aggregate curve is a composite, not a diagnosis

Search Console remains the most complete first-party view of Google performance, but it is not a zero-click attribution tool. It does not give a clean, query-level flag for every AI Overview or other SERP feature. A fall in CTR can therefore mean several things: an AI Overview answered the query, a featured snippet absorbed the click, the page slipped below a new module, or the query mix shifted towards research-led phrases. Treat the sitewide line as a composite. Decompose it before acting.

Segment by intent before drawing conclusions

Start with the Search Console API or bulk export, filtered to the United Kingdom and grouped by query, page, device and date. Build intent buckets with a documented keyword and regex list: definitional, comparison, commercial investigation, transactional, local commercial, brand and support. Keep the rules in version control so that cohort definitions do not drift between reports. Then map pages to types: editorial, category, product or service, location, support and tool. Device should be split at least into mobile and desktop; tablet can be grouped if volume is thin.

Click yield is the first metric: clicks divided by impressions, expressed per 1,000 impressions. Compare each cohort with its own pre-AI baseline rather than with a sitewide average. A drop from 24 to 18 clicks per 1,000 impressions on comparison pages matters more than a similar drop on definition pages if the former carries commercial value. Where volume allows, model expected clicks using position, device and intent-specific CTR benchmarks drawn from the baseline period. The variance between expected and actual clicks is the signal to investigate.

A measurement framework that survives zero-click

Use three layers. The first is visibility: impressions and click yield by cohort. The second is value: sessions, conversions and assisted conversions from GA4 or CRM, segmented by landing-page type and query bucket where possible. The third is SERP presence: manual or third-party sampling of AI Overview citations for a fixed set of high-value queries. Do not rely on a single dashboard. Tie the layers together with a monthly review.

Set a baseline window that predates the current SERP pattern, then compare like-for-like periods. Where possible, create a control cohort of queries or locations that have not been exposed to the same feature mix. A simple difference-in-differences view is often enough: compare the change in click yield for exposed and control cohorts. The aim is not academic precision; it is to avoid attributing every decline to AI Overviews when seasonality, tracking changes or cannibalisation may be responsible.

Where clicks are usually lost

Teardowns typically show three patterns. Definitional and how-to queries lose click yield first, because a concise answer is sufficient. Comparison queries lose clicks at the top of the funnel but retain value when the page offers original testing, pricing context or expert judgement that the overview cannot replicate. Local commercial queries can be more resilient, but they are vulnerable when the map pack and AI Overview appear together and the page lacks proof of proximity, availability or service scope.

Page type matters as much as intent. A category page with thin copy may lose both clicks and conversions, while a well-structured service page with clear pricing, coverage and evidence can hold its own. Mobile behaviour amplifies the pattern: shorter sessions and app switching make the first screen decisive.

Priority fixes for high-value traffic

1. Protect commercial comparison and service pages first. Add original data, tested claims, named expert commentary and visible update dates. Make the page useful after the overview has been read, not merely a restatement of it.
2. Consolidate cannibalised pages. If three articles answer the same definitional query, merge them into one authoritative resource and redirect the rest. Fewer, stronger URLs are easier to measure and easier to cite.
3. Design for citation and for click. Use clear headings, tables, sourced statistics and concise definitions. AI systems reward extractable structure; users reward a reason to continue reading.
4. Strengthen commercial proof. For product and service pages, add availability, delivery or coverage details, pricing ranges in GBP, guarantees and location-specific evidence. These are the elements that turn a zero-click SERP into a branded visit later.
5. Measure assisted value. A research query may not convert in the same session. Use GA4 path analysis and CRM source data to see whether zero-click exposure still contributes to branded search, direct visits and downstream conversions.
6. Audit local pages. Unique address, phone, opening hours and service-area copy should match Google Business Profile and internal records. Inconsistent local data undermines both trust and measurement.

UK-specific checks

Filter to the United Kingdom and review regional modifiers separately. Queries containing near me, city names and postcode districts behave differently from national head terms. Seasonal demand, bank holidays and regulatory language can distort month-on-month comparisons. Use UK spelling and terminology in query lists, and check that currency, delivery and service-area statements are consistent across pages.

Conclusion

AI Overviews have not removed the need for Search Console analysis; they have raised the cost of lazy analysis. The sites that protect high-value traffic in 2026 will be those that segment by intent, page type and device, set a baseline, and measure value beyond the click. The aggregate curve can start the conversation, but it should never end it.