AI Search Statistics UK 2026

Statistics 21–30 — AI Overviews, Click-Through, Zero-Click Search & Website Traffic

The most significant commercial change created by AI-powered search may not be whether people continue using Google.

It may be what happens after the search.

Generated answers can satisfy part of the information need directly inside the results page, changing whether users click a traditional result, click a cited source, perform another search or leave the search journey entirely.

21. 18% of Google searches in Pew’s behavioural study produced an AI summary

Around one in five searches within the measured dataset therefore produced an AI-generated answer alongside conventional search results.

AI Overview prevalence varies by query type, geography and measurement date, so this figure should not be treated as a universal Google rate.

AI search implication: Generated answers already occur frequently enough to affect conventional SEO traffic across substantial query sets.

Source: Pew Research Center, July 2025.

22. 58% of people in the Pew study encountered at least one Google AI summary during the month

Even when AI Overviews appear on a minority of total queries, they can still reach a majority of active search users during normal monthly search activity.

AI search implication: Exposure to generative search can become mainstream before AI-generated summaries appear on every type of query.

Source: Pew Research Center, July 2025.

23. Users clicked a traditional search result on only 8% of visits when an AI summary appeared

For searches without an AI summary, users clicked a conventional result on 15% of visits.

The observed click rate was therefore almost twice as high when no generated summary appeared.

AI search implication: A page can retain the same organic position while receiving fewer clicks because the search interface above it has changed.

Source: Pew Research Center, July 2025.

24. Only 1% of visits to pages with an AI summary produced a click on a source cited inside the summary

This is one of the most important distinctions in AI search measurement.

Being cited can create visibility and authority without generating an equivalent volume of direct referral traffic.

AI search implication: Citation visibility and traffic should be reported separately. A source can influence the answer without receiving the click.

Source: Pew Research Center, July 2025.

25. 26% of visits ended the browsing session after an AI summary appeared

Users encountering an AI-generated summary were more likely to stop browsing immediately after the search results page.

That pattern is consistent with some information needs being resolved directly within Google.

AI search implication: Search success is becoming less synonymous with website traffic because the search engine can increasingly complete part of the information journey itself.

Source: Pew Research Center, July 2025.

26. 88% of Google’s measured AI summaries cited at least three sources

Only 1% of the measured summaries cited a single source.

This demonstrates that generated search answers usually synthesise information from several sources rather than simply replacing the organic result with one publisher.

AI search implication: AI Overviews create new source competition: businesses and publishers are competing for inclusion inside a multi-source answer as well as for conventional rankings.

Source: Pew Research Center, July 2025.

27. 53% of searches containing ten words or more generated an AI summary

The probability of encountering an AI-generated answer increased sharply as search queries became longer.

Long queries often contain more context, constraints and explanatory intent — the type of information need that generative systems are designed to synthesise.

AI search implication: Long-tail search visibility increasingly requires consideration of generated answers, not only conventional blue-link rankings.

Source: Pew Research Center, July 2025.

28. 60% of searches beginning with question words generated an AI summary

Full-sentence searches were also more likely to trigger AI summaries, with 36% of queries containing both a noun and a verb producing one.

AI search implication: Informational content built to answer direct questions faces particularly high exposure to generative search interfaces.

Source: Pew Research Center, July 2025.

29. AI Overviews were associated with a 58% lower click-through rate for the number-one organic result

Ahrefs re-ran its AI Overview click study using December 2025 data across 300,000 keywords.

For keywords that triggered AI Overviews, the observed average position-one desktop CTR was approximately 1.6%, compared with an estimated 3.7% had the AI Overview not been present.

Ahrefs therefore estimated a 58% reduction attributable to the presence of the AI Overview after adjusting for the broader decline in informational-search CTR.

AI search implication: Number-one ranking can remain valuable while producing materially less traffic when a generated answer occupies the search interface above it.

Source: Ahrefs, Update: AI Overviews Reduce Clicks by 58%, February 2026.

30. AI Overview click-through losses were measurable across every position in Google’s top ten

Ahrefs reported the following estimated effects:

• Position 1 — −58.0%
• Position 2 — −50.8%
• Position 3 — −46.4%
• Position 4 — −38.8%
• Position 5 — −32.6%
• Position 6 — −30.5%
• Position 7 — −29.7%
• Position 8 — −28.8%
• Position 9 — −29.7%
• Position 10 — −19.4%

The estimated effect therefore extended beyond the first result.

AI search implication: AI Overviews can change the economic value of the entire first page of organic results, not merely position one.

Source: Ahrefs, February 2026.

Statistics 31–40 — AI Citations, Google Ranking Overlap, Brand Mentions & Source Selection

AI search visibility is not the same as traditional search visibility.

A webpage can rank strongly without being cited by an AI system, while another page can receive an AI citation despite not ranking for the user’s original query.

There is also an important difference between being cited as a source and having the organisation’s brand explicitly named in the generated answer.

31. 61.7% of measured AI citations did not produce an explicit brand mention

Semrush and Kevin Indig describe this as a ghost citation: the AI system uses and links to the source, but the brand itself is not named within the generated answer.

The website may therefore contribute information without receiving equivalent brand recognition.

AI search implication: Citation visibility and brand visibility should be measured separately.

Source: Semrush / Kevin Indig / Growth Memo, Why 62% of AI Citations Don’t Lead to Brand Mentions, June 2026.

32. 74.9% of measured appearances included a citation, but only 38.3% included a brand mention

Within the same dataset:

• 61.7% were cited without a brand mention.
• 13.2% were both cited and mentioned.
• 25.1% were mentioned without a citation.

This demonstrates that generative systems can use an organisation’s information without making the organisation prominent to the user.

AI search implication: Reporting one combined “AI visibility” metric can conceal major differences between attribution and brand recognition.

Source: Semrush, June 2026.

33. ChatGPT cited brands in 87% of measured appearances but mentioned them in only 20.7%

ChatGPT showed one of the clearest examples of the difference between being used as a source and being named in the answer.

A publisher can therefore achieve strong ChatGPT citation visibility while still receiving relatively limited explicit brand exposure.

AI search implication: Businesses should not interpret a high ChatGPT citation count as evidence of equivalent brand recommendation visibility.

Source: Semrush, June 2026.

34. Gemini mentioned brands in 83.7% of appearances but cited them in only 21.4%

Gemini produced almost the reverse pattern of ChatGPT within the study.

This is strong evidence that AI platforms should not be treated as one homogeneous search channel.

AI search implication: An organisation may perform strongly for brand mentions on one AI platform and strongly for citations on another.

Source: Semrush, June 2026.

35. Comparative queries produced 2.4 times more brand mentions than informational queries

Informational prompts produced a high citation rate but relatively few explicit brand mentions.

Comparison queries were more likely to force the system to name organisations, products or alternatives directly.

AI search implication: Commercial comparison visibility and informational citation visibility are different strategic objectives and should be tracked separately.

Source: Semrush, June 2026.

36. Only around 12% of AI-cited URLs ranked in Google’s top ten for the original prompt

Ahrefs compared visible citations from ChatGPT, Gemini, Copilot and Perplexity with Google results for the same 15,000 long-tail prompts.

The low overlap shows that AI systems do not simply copy Google’s first page when constructing generated answers.

AI search implication: Conventional rankings remain useful for discoverability, but AI source selection applies an additional retrieval and filtering layer.

Source: Ahrefs, Only 12% of AI-Cited URLs Rank in Google’s Top 10 for the Original Prompt, August 2025.

37. Around 80% of measured AI citations did not rank anywhere in Google for the original prompt

This does not necessarily mean those pages were invisible to Google generally.

An AI system can generate related fan-out queries and discover a source through another search path.

AI search implication: The question a page ranks for may differ from the subquestion through which an AI system retrieves and cites it.

Source: Ahrefs, August 2025.

38. Perplexity had 28.6% citation overlap with Google’s top ten, while other measured AI assistants were around 8%

Perplexity was substantially more aligned with conventional Google rankings than ChatGPT, Gemini or Copilot.

This reinforces the broader finding that different AI products use different source-selection processes.

AI search implication: SEO performance may translate more directly into citation visibility on some AI platforms than on others.

Source: Ahrefs, August 2025.

39. Only 37.1% of Google AI Overview cited URLs ranked in the conventional organic top ten in the 2026 update

Ahrefs analysed 863,000 keyword SERPs and approximately four million AI Overview URLs in its March 2026 update.

The cited URLs were distributed as follows:

• 37.1% ranked in Google’s organic top ten.
• 26.2% ranked between positions 11 and 100.
• 36.7% did not rank in the organic top 100 for the same query.

This was materially lower than Ahrefs’ earlier 2025 study, reinforcing how quickly AI source selection can change as Google updates its generative search systems.

AI search implication: Even within Google’s own AI search environment, a conventional top-ten ranking no longer explains the majority of final citation selection.

Source: Ahrefs, Update: 38% of AI Overview Citations Pull From the Top 10, March 2026.

40. ChatGPT matched Google ranking domains 31.8% of the time but exact ranking URLs only 10%

Ahrefs’ separate analysis of 3,311 short-tail terms found that ChatGPT was around three times more likely to agree with Google about the domain than about the precise ranking page.

This means the same trusted website can appear in both systems while each system selects a different page from that domain.

AI search implication: Topic clusters matter because several pages from an authoritative domain can become citation candidates for different fan-out queries and information needs.

Source: Ahrefs, ChatGPT May Scrape Google, but the Results Don’t Match, September 2025.

Statistics 41–50 — Reasoning Depth, Query Fan-Out, Freshness, Referral Traffic & the Emerging AI Search Economy

The final ten statistics show why AI search cannot be understood through prompt tracking alone.

How deeply a model reasons can change which sources it finds, how many searches it performs and which domains appear in the final answer.

At the same time, AI platforms are beginning to generate measurable referral traffic while Google’s own AI-powered search products have reached audiences measured in the billions.

41. ChatGPT’s citation rate increased from 50% to 68% when higher reasoning was used

Semrush tested the same 100 prompts in two reasoning configurations.

When deeper reasoning was enabled, ChatGPT relied more heavily on external research and citations.

AI search implication: The probability of being cited can change according to how difficult the user’s question is and how much research the model performs before answering.

Source: Semrush / Kevin Indig, Only 25% of Cited Sources Overlap Between ChatGPT’s Different Reasoning Modes, June 2026.

42. High-reasoning ChatGPT responses cited 4.5 sources on average versus 2.6 with minimal reasoning

Deeper reasoning creates more opportunities for external sources to enter the answer.

But it also creates more competition because the model considers a wider evidence set.

AI search implication: Complex research queries can create more citation opportunities than simple prompts, while increasing the number of competing sources.

Source: Semrush, June 2026.

43. High-reasoning ChatGPT generated 4.6 times more internal fan-out queries

Across the full study:

• Minimal reasoning ran 245 web searches.
• High reasoning ran 1,130 web searches.

At the comparison stage of the buyer journey, high reasoning generated an average of 24 subqueries per prompt, versus approximately 5.5 in minimal reasoning.

AI search implication: Organisations increasingly compete across a web of secondary questions generated by the AI system, not just the wording of the original prompt.

Source: Semrush, June 2026.

44. Only 25.6% of cited domains overlapped between minimal- and high-reasoning ChatGPT responses

The same prompts therefore produced materially different source sets depending on how ChatGPT approached the task.

AI search implication: One prompt test cannot represent stable AI visibility. Citation monitoring needs repeated testing and should recognise model configuration and reasoning depth.

Source: Semrush, June 2026.

45. High reasoning cited 173 unique domains compared with 127 under minimal reasoning

Of the 173 domains cited under high reasoning, 99 did not appear at all under minimal reasoning.

Semrush also found a shift in source type: Reddit’s share declined from approximately 15% to 7%, while official, government and academic sources gained ground.

AI search implication: More complex prompts can open visibility opportunities for specialist, institutional and primary sources that do not appear in faster, shallower responses.

Source: Semrush, June 2026.

46. AI-cited content was 25.7% fresher than content appearing in organic Google results

Ahrefs compared millions of AI citations with conventional organic search results across seven AI platforms.

The research also found a 13.1% preference for more recently updated content.

This is an observed relationship rather than proof that freshness operates as one independent universal AI ranking factor.

AI search implication: Pages containing time-sensitive evidence, recommendations or statistics need substantive maintenance if they are to remain strong citation candidates.

Source: Ahrefs, AI citation freshness research, 2025–2026.

47. ChatGPT cited URLs hundreds of days newer than conventional Google results

Among the AI systems examined, ChatGPT showed one of the strongest skews towards newer source material.

That does not justify changing publication dates without updating the underlying content.

AI search implication: Research pages, statistics resources, comparisons and fast-changing commercial content should be maintained as living assets rather than published once and abandoned.

Source: Ahrefs, analysis of approximately 17 million AI citations.

48. ChatGPT outbound referral traffic increased 206% year on year during 2025

Semrush analysed more than one billion lines of US clickstream data across 17 months.

The same study found:

• ChatGPT referred traffic to approximately 71,000 domains per month in October 2024.
• This peaked at around 260,000 domains in October 2025.
• Approximately 170,000 domains were receiving referrals by February 2026.
• ChatGPT activated web search on 34.5% of queries in February 2026.

AI search implication: Direct traffic from AI systems remains comparatively small but is expanding into a meaningful referral channel for a growing number of websites.

Source: Semrush, ChatGPT Traffic Analysis: Insights From 17 Months of Clickstream Data, April 2026.

49. Google still sent approximately 190 times more website traffic than ChatGPT

Ahrefs compared referral traffic across approximately 76,000 websites.

The same analysis estimated that around 65% of ChatGPT usage could reasonably qualify as search or search-like activity under its methodology.

That equated to approximately 12% of Google’s search volume for tasks people might traditionally have performed through a search engine.

AI search implication: AI-search usage can become significant long before AI referral traffic approaches Google-scale website traffic.

Source: Ahrefs, ChatGPT Has 12% of Google’s Search Volume but Google Sends 190× More Traffic to Websites, February 2026.

50. Google AI Overviews now exceeds 2.5 billion monthly active users

Google reported in 2026 that:

• AI Overviews exceeded 2.5 billion monthly active users.
• AI Mode exceeded 1 billion monthly active users within approximately one year of launch.
• AI Mode query volume had more than doubled every quarter since launch.
• Overall Google Search query volume reached an all-time high.

These are global first-party Google figures rather than UK user totals.

AI search implication: The largest deployment of generative search is increasingly happening inside the world’s existing dominant search engine rather than only through standalone chatbots.

Source: Google, I/O and Search ecosystem updates, May–August 2026.

CGO Media Analysis — What the 50 AI Search Statistics Tell Us About UK Search in 2026

Taken together, the 50 statistics show that UK search is entering a hybrid phase.

Google remains dominant, but AI-generated answers are increasingly layered into the search journey through Google AI Overviews, AI Mode, ChatGPT and other assistants.

The result is not a simple replacement of one search engine by another.

It is a redistribution of discovery across ranked results, generated answers, citations, brand mentions, recommendations and increasingly fragmented user journeys.

1. Google Is Still the Centre of UK Search

The evidence does not support the idea that Google has already been displaced.

Google Search continues to reach around four in five UK adults and still handles search volume measured in billions of queries per month.

For most organisations, conventional Google visibility therefore remains commercially essential.

2. AI Search Is Growing Through Two Routes

The transition is happening in two different ways:

• Users actively choosing ChatGPT, Gemini, Copilot or other AI assistants.
• Google inserting AI-generated answers directly into mainstream search.

This is strategically important because the second route does not require consumers to change platform at all.

3. Younger UK Audiences Are Further Ahead

AI adoption among younger UK adults is substantially above the national average.

That means the commercial importance of AI search varies by customer demographic.

Organisations targeting younger professionals, students or digitally intensive consumers may experience the shift earlier and more strongly than businesses serving older audiences.

4. ChatGPT Has Become a Major UK Discovery Platform

ChatGPT’s UK audience increased rapidly within one year and is now measured in tens of millions of visits and millions of users.

Its importance therefore extends beyond novelty or technical experimentation.

For many categories, it now functions as an information-retrieval and recommendation environment in its own right.

5. AI Search Is Additive Before It Is Fully Substitutive

The strongest current model is not:

Google OR ChatGPT

It is:

Google + ChatGPT + AI Overviews + Follow-Up Search + Direct Verification

Users can move between platforms within one research journey.

6. AI Search Is Especially Strong for Research-Led Queries

The referral-category data shows stronger relative representation in:

• Science and education.
• News and media.
• Shopping.
• Health.
• Lifestyle.

These are categories where users often need explanation, comparison, synthesis or follow-up questions.

That makes AI particularly well suited to complex research behaviour.

7. Search Journeys Are Becoming Less Linear

Traditional SEO often assumes a relatively simple funnel:

AI search introduces more possible pathways:

8. Rankings Alone Are Becoming a Less Complete Performance Metric

AI Overviews can reduce click-through rates even when the ranking remains unchanged.

That means a stable number-one position can produce less traffic than it did previously.

SEO reporting therefore needs to separate:

• Ranking.
• Impressions.
• Click-through rate.
• AI Overview presence.
• Citation visibility.

9. Zero-Click Search Is Becoming More Economically Important

Generated answers can satisfy users without requiring a website visit.

This creates a distinction between:

• Visibility.
• Influence.
• Traffic.

An organisation can influence a user inside an AI answer without receiving a direct click.

10. Citation Visibility Is Not the Same as Brand Visibility

One of the clearest findings in the dataset is that AI systems can cite a website without naming the brand.

They can also mention a brand without linking to it.

This means future reporting should distinguish:

• Citations.
• Brand mentions.
• Recommendations.
• Referral clicks.

11. AI Platforms Behave Differently

ChatGPT, Gemini, Google AI Overviews and AI Mode do not produce identical relationships between citations and brand mentions.

This makes a universal “AI ranking” concept too simplistic.

Performance should be measured separately by platform.

12. Comparative Queries Are More Commercially Important for Brand Visibility

Comparative prompts produce materially more brand mentions than purely informational queries.

This means businesses need visibility across both:

• Informational research.
• Commercial comparison.

The two stages produce different outcomes.

13. Conventional Google Rankings Do Not Fully Explain AI Citations

Only a minority of AI-cited URLs overlap with Google’s top ten for the original prompt.

Even Google AI Overviews often cite pages that are not in the top ten for the same query.

This indicates another source-selection stage beyond conventional organic ranking.

14. Query Fan-Out Changes the Competitive Set

AI systems can break one user prompt into multiple related searches.

A page can therefore be retrieved because it answers one generated subquestion rather than because it ranks for the user’s exact wording.

This makes topic architecture increasingly valuable.

15. Domain-Level Authority Can Matter Beyond One Ranking URL

ChatGPT’s stronger overlap at domain level than exact-URL level suggests that a trusted domain can compete through several pages.

This supports a broader content model:

One Topic → Multiple Specialist Pages → Multiple AI Retrieval Opportunities

16. Reasoning Depth Changes Source Selection

Higher reasoning produces more searches, more citations and a broader set of sources.

The same prompt can therefore expose different websites depending on how deeply the model researches the question.

AI visibility is not fixed.

17. AI Search Visibility Is Probabilistic

Only around one quarter of cited domains remained consistent between minimal- and high-reasoning ChatGPT tests.

This means one test response is insufficient to establish durable AI visibility.

Repeated measurements are necessary.

18. Freshness Is Increasingly Important

AI systems frequently cite newer information than conventional organic results.

This does not mean every page needs constant rewriting.

It means content containing changing information should remain current, especially:

• Statistics.
• Research.
• Prices.
• Comparisons.
• Product information.
• Market analysis.

19. AI Referral Traffic Is Growing From a Small Base

ChatGPT referral traffic is growing rapidly.

But conventional Google traffic remains vastly larger.

This creates a common measurement mistake: interpreting high percentage growth as equivalent scale.

Both growth rate and absolute traffic share need to be reported.

20. Google Itself Is Becoming an AI Search Platform

Google AI Overviews and AI Mode now operate at enormous global scale.

This means AI search growth does not depend entirely on standalone assistants.

The most important transition may happen inside the search engine consumers already use.

The Strategic Shift

Traditional SEO asked:

Can this page rank and generate clicks?

AI search adds:

Can this organisation become one of the sources an AI system repeatedly retrieves, cites, mentions and recommends?

What UK Businesses Should Do in 2026 to Improve AI Search Visibility

The 50 statistics show that AI search visibility cannot be built through one isolated optimisation.

Organisations need strong conventional search foundations while also increasing the likelihood that their information is retrieved, cited, associated with the brand and used in recommendations across generative systems.

1. Protect Conventional SEO Foundations

AI search does not remove the need for:

• Crawlability.
• Indexation.
• Canonicalisation.
• Internal linking.
• Mobile performance.
• Page speed.
• Search-friendly site architecture.

Search engines and generative systems still depend heavily on accessible web content.

2. Build Topic Architecture Rather Than Isolated Keyword Pages

Query fan-out means one user question can become many related searches behind the scenes.

Businesses should therefore build connected content covering:

• Core concepts.
• Definitions.
• Comparisons.
• Alternatives.
• Pricing.
• Implementation.
• Use cases.
• Limitations.
• Industry applications.

This creates multiple possible routes into the retrieval process.

3. Answer the Subquestions Behind the Main Query

A generative system may retrieve a page because it answers one subproblem rather than the original user question.

Businesses should map the wider research journey around priority commercial topics.

4. Make Page Titles and URLs Highly Specific

Clear titles and descriptive URLs help both users and retrieval systems understand exactly what a page covers.

Avoid vague headings built around marketing language alone.

Prefer:

/ai-search-statistics-uk-2026/

over structures that provide little semantic context.

5. Publish Original Research

Original evidence creates a stronger reason for an AI system to cite the organisation directly.

Useful formats include:

• Industry studies.
• Surveys.
• Datasets.
• Statistics.
• Experiments.
• Benchmarks.
• Case studies.

If every statistic on a page originates elsewhere, the original publisher remains the stronger citation candidate.

6. Make Research Provenance Clear

Original research should explain:

• Who conducted it.
• When it was conducted.
• The sample size.
• The methodology.
• The geography.
• The limitations.
• The publication or update date.

This improves usefulness for journalists, researchers and machine systems alike.

7. Keep Important Evidence Current

AI citation research shows a strong skew towards relatively recent information.

Businesses should regularly review:

• Statistics pages.
• Comparison pages.
• Pricing content.
• Product information.
• Industry research.

Updating should improve the substance, not merely change the visible date.

8. Create Comparison Content

Comparative queries generate significantly more brand mentions than purely informational queries.

Useful content can include:

• Product A vs Product B.
• Agency comparisons.
• Alternative providers.
• Pricing comparisons.
• Feature comparisons.
• Suitability by business type.

The content should be transparent and genuinely useful rather than disguised advertising.

9. Strengthen Brand-Entity Clarity

Citation and brand mention are separate outcomes.

Businesses should make their identity consistently clear across:

• Website.
• Author profiles.
• Research-team pages.
• Press pages.
• External profiles.
• Directories.

The objective is to make the organisation behind the information unambiguous.

10. Build External Corroboration

Generative systems retrieve information from across the web.

Independent evidence can strengthen the wider authority environment around the organisation.

Relevant sources include:

• Trade publications.
• National or regional media.
• Professional associations.
• Academic repositories.
• Research platforms.
• Industry directories.

11. Use Digital PR to Distribute Original Evidence

Digital PR becomes more useful when it carries information others have a reason to cite.

Strong assets include:

• Original statistics.
• Market data.
• Research findings.
• Expert commentary.
• New datasets.

The aim is wider independent recognition, not simply link acquisition.

12. Build Content for Informational and Commercial Intent

Informational content can generate citations while comparative content can generate more explicit brand mentions.

Businesses therefore need both.

13. Optimise for More Than One AI Platform

ChatGPT, Gemini, Google AI Overviews, AI Mode and Perplexity can use very different source sets.

Where commercially relevant, monitor each separately.

Do not assume success in ChatGPT automatically means success in Gemini or Google AI Search.

14. Measure AI Overview Exposure

For important informational queries, track whether Google displays an AI Overview.

This helps explain situations where:

• Rankings remain stable.
• Impressions remain high.
• Click-through rate falls.

Traffic loss should not automatically be interpreted as a ranking problem.

15. Measure Citations and Brand Mentions Separately

Every AI visibility dashboard should distinguish between:

• Source citation.
• Brand mention.
• Recommendation inclusion.
• Referral visit.

These outcomes are related, but they are not interchangeable.

16. Repeat Prompt Testing

AI search visibility is probabilistic.

The same prompt can produce different citations depending on:

• Model version.
• Reasoning depth.
• Search context.
• Source availability.

One screenshot should never be treated as a durable ranking.

17. Monitor Complex Buyer-Journey Prompts

Higher-reasoning queries trigger more fan-out searches and source citations.

Businesses should therefore monitor prompts such as:

• Which provider is best for my requirements?
• What are the alternatives?
• Which option offers the best fit?
• How do these providers compare?

These prompts may be more commercially valuable than basic informational questions.

18. Measure AI Referral Traffic — But Do Not Use It as the Only KPI

ChatGPT referral traffic is growing quickly, but it remains far smaller than Google traffic.

Also monitor:

• Branded search.
• Direct traffic.
• Assisted conversions.
• CRM lead sources.
• Sales conversations mentioning AI tools.

AI influence can occur without a direct referral click.

19. Do Not Chase Unsupported AI SEO Shortcuts

Businesses should prioritise evidence-led optimisation over speculative tactics.

The strongest current areas remain:

• Technical discoverability.
• Topical depth.
• Semantic relevance.
• Original evidence.
• Freshness.
• External authority.

Any new technique should be tested against measurable outcomes rather than accepted because it is described as “AI SEO”.

20. Measure Commercial Impact

The purpose of AI search visibility is not simply to appear in an answer.

Businesses should connect AI visibility with:

• Lead volume.
• Qualified enquiries.
• Pipeline.
• Transactions.
• Revenue.
• Customer value.

The Practical Priority

AI Search Measurement Framework — What UK Businesses Should Track Each Month

AI search performance cannot be measured accurately through one visibility score.

A business may retain strong Google rankings while losing click-through rate to AI Overviews. It may be cited frequently in ChatGPT without being named. It may receive brand mentions without a direct website link.

Each outcome needs to be tracked separately before the commercial effect can be understood.

1. Google Organic Rankings

Continue tracking conventional rankings for priority commercial and informational queries.

Measure:

• Average position.
• Top-three rankings.
• Top-ten rankings.
• New ranking pages.
• Lost ranking pages.

Traditional search visibility remains the foundation against which AI-search changes can be interpreted.

2. Google Search Impressions

Track impressions independently from clicks.

A page may remain highly visible in Google while attracting fewer visits if AI Overviews answer more of the query directly.

3. Organic Click-Through Rate

CTR is increasingly important because AI Overviews can change the value of the same ranking position.

Track CTR by:

• Query.
• Page.
• Position.
• AI Overview presence.

A traffic decline with stable rankings can indicate search-interface change rather than an SEO ranking problem.

4. AI Overview Presence

For important informational queries, record whether a Google AI Overview appears.

Track:

• Percentage of monitored queries triggering AI Overviews.
• New AI Overview triggers.
• Queries where an AI Overview disappears.

5. AI Overview Citation Visibility

Track whether the organisation’s pages appear as sources inside Google’s AI-generated answers.

Measure:

• Total cited queries.
• Unique cited pages.
• Citation frequency.
• Repeat citation rate.

6. ChatGPT Citation Frequency

Track whether ChatGPT uses the organisation as a cited source across a stable benchmark prompt set.

Record:

• Total citations.
• Unique prompts producing citations.
• Unique pages cited.
• Month-on-month change.

7. Brand Mention Rate

A citation does not guarantee that the brand is named.

Track how often the organisation appears explicitly within generated answers.

Separate:

• Brand mention with citation.
• Brand mention without citation.
• Citation without brand mention.

8. Recommendation Inclusion

Commercial prompts should be measured differently from purely informational questions.

Track whether the business is:

• Named as an option.
• Included in a shortlist.
• Recommended.
• Compared with competitors.

9. Prompt Coverage

Create a fixed benchmark covering the stages of the buyer journey.

Include:

• Informational prompts.
• Problem-led prompts.
• Comparison prompts.
• Alternative-provider prompts.
• Provider-selection prompts.
• Brand prompts.

Re-run the same core prompt set so changes can be measured consistently.

10. Cross-Platform AI Visibility

Do not combine all AI platforms before examining them individually.

Where relevant, monitor:

• Google AI Overviews.
• Google AI Mode.
• ChatGPT.
• Gemini.
• Perplexity.
• Copilot.

11. Citation Consistency

AI answers vary.

Repeat important prompts and classify citations as:

• Consistent.
• Occasional.
• One-off.

This produces a more reliable measure than a single successful appearance.

12. Cited Page Distribution

Track which page types receive generative visibility.

Examples:

• Research papers.
• Statistics pages.
• Service pages.
• Comparison pages.
• Guides.
• Frameworks.
• Case studies.

This can reveal which content formats AI systems consistently prefer for different query types.

13. Citation Concentration

Determine whether most AI visibility depends on a very small number of URLs.

Track:

• Total cited URLs.
• Top-cited URL share.
• Top-five cited URL share.
• Number of topic clusters receiving citations.

14. Competitor Citation Share

Run the same prompt set for major competitors.

Measure:

• Share of citations.
• Share of brand mentions.
• Share of recommendations.
• Platform-specific strengths.

15. Content Freshness of Cited Pages

For pages appearing frequently in AI answers, track:

• Original publication date.
• Last substantive update.
• Age of key statistics.
• Age of supporting evidence.

This helps identify whether citation decline may be connected with aging information.

16. AI Referral Traffic

Track referral sessions from AI platforms where technically identifiable.

Measure:

• Sessions.
• Landing pages.
• Engagement.
• Conversions.
• Revenue.

Keep this metric separate from citation visibility because many AI exposures do not produce a click.

17. Branded Search Demand

AI recommendations can influence later behaviour.

Track changes in:

• Brand-only searches.
• Brand + service searches.
• Brand + review searches.
• Direct website traffic.

These movements should be treated as supporting evidence rather than automatically attributed to AI.

18. Assisted Conversions

AI search can influence a customer without generating the final click.

Where possible, connect AI exposure with:

• CRM source notes.
• Lead forms.
• Sales conversations.
• Customer surveys.
• Attribution modelling.

19. Leads and Revenue

The final objective is commercial performance.

Track:

• Qualified leads.
• Pipeline.
• Transactions.
• Revenue.
• Customer value.

20. Search-to-Commercial-Outcome Funnel

Recommended Monthly Reporting Table

The Measurement Rule

AI Search Trends UK 2026–2027 — How Google, ChatGPT, AI Overviews & Generative Discovery Are Evolving

The 50 statistics in this report point towards a search environment that is becoming more conversational, more fragmented and increasingly capable of answering questions before users visit a website.

The following trends are evidence-led directions rather than guaranteed forecasts.

They are based on current UK adoption data, Google Search behaviour, ChatGPT growth, AI Overview click studies, citation research, query fan-out and first-party platform data.

1. Google Will Remain Central While Becoming More AI-Led

The evidence does not suggest that Google is disappearing.

Instead, Google is integrating generative AI more deeply into the search experience itself.

AI Overviews now operates at global scale, while AI Mode has grown rapidly and is being integrated more closely with conventional Search.

For UK businesses, the likely direction is therefore not “SEO or AI search”.

It is increasingly:

SEO + AI Search Inside the Same Google Ecosystem

2. Search Will Become More Conversational

Google is explicitly moving Search towards a more continuous conversational experience.

Users can increasingly move from:

• Initial query.
• AI Overview.
• Follow-up question.
• AI Mode.
• External source.

without treating each interaction as a separate search session.

3. AI Search Adoption Will Continue to Rise Among Younger UK Audiences

UK AI adoption is already substantially higher among 16–34-year-olds than among the population overall.

These cohorts are likely to normalise conversational search more quickly across:

• Education.
• Shopping.
• Travel.
• Work.
• Technology.
• Financial research.

The commercial impact will therefore appear earlier in some sectors than others.

4. ChatGPT Will Continue to Function as a Parallel Search Interface

ChatGPT has already developed a large direct UK audience.

Its strongest role is likely to remain research-led discovery involving:

• Explanation.
• Comparison.
• Recommendation.
• Planning.
• Follow-up questions.

It does not need to replace Google completely to materially affect the research journey.

5. Search Journeys Will Become More Multi-Platform

Users increasingly have several ways to solve the same information need.

A future journey may look like:

Visibility therefore needs to extend across the journey rather than one platform alone.

6. AI Overviews Will Continue to Pressure Informational CTR

Current click studies consistently show lower organic click-through when an AI-generated summary appears.

This means informational content can remain visible while producing fewer direct visits.

Through 2027, businesses should expect increasing separation between:

• Search visibility.
• Website traffic.

7. Rankings Will Remain Important but Become Less Sufficient

Traditional rankings still influence discoverability.

But the evidence shows that AI systems frequently cite pages outside the organic top ten for the original query.

The performance question increasingly becomes:

Can the source rank, be retrieved and survive generative source selection?

8. Citation Visibility Will Become a Standard Search KPI

As generated answers become more common, organisations will increasingly track whether their content is used as evidence.

Citation metrics are likely to sit alongside:

• Rankings.
• Impressions.
• Traffic.
• Backlinks.

But citation volume alone will not be sufficient.

9. Brand Mentions Will Become a Separate KPI

Current research shows that a source can be cited without the brand being named.

This makes explicit brand recognition increasingly important.

Search reporting will likely distinguish:

• Source citation.
• Brand mention.
• Recommendation.

10. Recommendation Visibility Will Become More Commercially Important

Commercial AI prompts increasingly ask:

• Which provider should I choose?
• What is the best option?
• What are the alternatives?
• Which company suits my requirements?

Being cited as an informational source and being recommended as a provider are fundamentally different outcomes.

The latter is likely to become one of the most important AI-search KPIs for commercial organisations.

11. Query Fan-Out Will Increase the Importance of Topic Architecture

Deeper reasoning can trigger many related web searches behind one prompt.

This means organisations increasingly need content covering the wider question space around a topic.

The strategic model moves away from:

One Keyword → One Page

towards:

One Topic → Multiple Connected Evidence Assets

12. Complex Queries Will Create More Source Competition

Higher-reasoning AI responses use more searches and more sources.

This creates more citation opportunities but also increases the number of competing domains.

Businesses targeting complex buyer journeys will therefore need stronger evidence than businesses answering simple factual questions.

13. AI Visibility Will Remain Probabilistic

The same prompt can produce different sources depending on reasoning depth, model configuration and retrieval conditions.

This means future AI-search reporting will need:

• Repeated prompts.
• Historical tracking.
• Consistency scores.
• Platform-specific measurement.

14. Original Research Will Become More Valuable

AI systems need evidence from which to construct answers.

Organisations that create primary information can become the source rather than merely summarising somebody else’s work.

Strategically valuable formats include:

• Statistics.
• Surveys.
• Datasets.
• Experiments.
• Research papers.
• Benchmarks.

15. Content Maintenance Will Become Part of AI Search Strategy

Current citation research shows a measurable bias towards relatively recent and recently updated content.

Through 2027, businesses should expect more value from systematic updates to:

• Statistics.
• Research.
• Pricing.
• Product information.
• Comparison content.
• Industry analysis.

Freshness should come from substantive updating, not cosmetic date changes.

16. AI Referral Traffic Will Grow Faster Than Its Absolute Share

ChatGPT referral traffic is already growing at triple-digit rates.

But Google remains vastly larger as a source of website traffic.

Businesses therefore need to avoid confusing:

• Fast growth.
• Large scale.

AI referrals can become strategically important before they become numerically dominant.

17. Zero-Click Influence Will Become Harder to Measure

An AI-generated answer can influence a consumer without creating a direct website visit.

A user may:

• See the brand in an AI recommendation.
• Remember it.
• Search the brand later.
• Visit directly.
• Convert through another channel.

Attribution systems will increasingly need to consider this indirect influence.

18. Search Console and Analytics Will Evolve Around AI Search

Google has already begun introducing new reporting and controls for AI-powered Search.

This points towards more mature measurement of generative search performance over time.

Businesses should expect AI Search reporting gradually to become less dependent on manual prompt testing.

19. SEO and GEO Will Continue to Converge

The strongest assets for conventional SEO are also useful for generative search:

• Crawlable pages.
• Clear topical structure.
• Original evidence.
• Authority.
• Fresh information.
• Strong entities.

The difference increasingly lies in the output being measured.

SEO measures rankings and traffic.

GEO additionally measures retrieval, citation, mention and recommendation.

20. Search Will Become a Multi-Surface Visibility Problem

The strongest search strategies through 2027 will need to consider several simultaneous outcomes.

What This Means for UK Organisations

The transition does not justify abandoning established SEO.

It requires expanding the definition of search visibility.

Research Methodology & Limitations

This report brings together 50 measurable data points relating to UK AI adoption, ChatGPT usage, Google Search, AI Overviews, search behaviour, click-through rates, AI citations, brand mentions, source selection, query fan-out, content freshness and AI referral traffic.

The objective is not to imply that one dataset can describe the entire AI-search environment.

Instead, the research combines UK-specific evidence with clearly labelled US, international and first-party platform benchmarks where equivalent UK data is not yet available.

1. Research Scope

The 50 statistics are organised around five main areas:

1. UK AI adoption and ChatGPT growth.
2. Changing search behaviour and information discovery.
3. AI Overviews, click-through rates and zero-click search.
4. AI citations, brand mentions and source selection.
5. Reasoning depth, freshness, referrals and AI-search scale.

These areas reflect the main stages through which AI search can influence the user journey.

2. Source Hierarchy

Sources were prioritised according to transparency, methodological detail, dataset scale and proximity to the underlying evidence.

The preferred hierarchy was:

1. UK regulatory and audience-measurement evidence — including Ofcom and Ipsos iris.
2. First-party platform data — including Google.
3. Independent behavioural research — including Pew Research Center.
4. Large-scale specialist search research — including Ahrefs and Semrush.
5. CGO Media calculations derived directly from published figures.

3. UK Evidence Is Prioritised Wherever Available

UK-specific evidence is used for areas including:

• AI-tool adoption.
• ChatGPT audience reach.
• Google Search usage.
• UK information-search behaviour.
• UK referral-category comparisons.

These figures provide the UK-specific foundation for the report.

4. International Benchmarks Are Explicitly Labelled

Detailed research into AI citations, source selection, click-through rates and reasoning behaviour is often available at international or US level before equivalent UK research is published.

Where such findings are used, they are labelled as:

• US Behavioural Benchmark.
• International SEO Benchmark.
• Multi-Platform Benchmark.
• First-Party Global Platform Data.

They should not be converted into UK population estimates.

5. AI Tool Usage Is Not the Same as AI Search Usage

A person may use ChatGPT, Gemini or Copilot for:

• Writing.
• Summarisation.
• Coding.
• Brainstorming.
• Research.
• Search.

Therefore, statistics measuring AI-tool adoption should not automatically be interpreted as AI-search adoption.

6. Website or App Reach Is Not the Same as Search Volume

Audience measurement showing how many people visited ChatGPT or another AI platform does not reveal how many search-like queries they performed.

Reach, visits and search volume are separate measures.

7. Visits Are Not Unique Users

One person can generate many website or app visits during a month.

Traffic figures such as ChatGPT’s UK visit totals therefore describe usage intensity rather than unique audience size.

8. Google Search Volume and ChatGPT Usage Are Not Directly Equivalent

Google searches and ChatGPT prompts are different interaction types.

A ChatGPT conversation can contain multiple follow-up questions within one session.

Comparisons between the two platforms should therefore remain approximate and methodology-dependent.

9. AI Overview Trigger Rates Vary by Query Type

AI Overviews are not distributed evenly across all searches.

They are more common for:

• Longer queries.
• Question-based searches.
• Informational searches.
• Full-sentence searches.

Any overall prevalence figure should therefore be interpreted in relation to the underlying query mix.

10. AI Overview Prevalence Changes Over Time

Google continues to change when and how AI Overviews appear.

A study conducted in March 2025 may therefore measure a different environment from one conducted in late 2026.

All AI Overview statistics should retain their measurement date.

11. CTR Studies Are Sensitive to Query Mix and Device

Click-through rate varies by:

• Ranking position.
• Device.
• Intent.
• SERP layout.
• Query length.

CTR-impact studies should therefore be interpreted as measured relationships within the study sample rather than universal guaranteed losses.

12. AI Overview Correlation Does Not Prove Every Traffic Decline Was Caused by AI

Search CTR can change because of:

• AI Overviews.
• Featured snippets.
• Video results.
• Shopping units.
• Other SERP features.
• Changes in user behaviour.

Strong studies attempt to control for these factors, but causal claims should remain limited to what the methodology supports.

13. Citations and Brand Mentions Are Different Outcomes

A source can be cited without the brand being named.

A brand can also be mentioned without receiving a source citation.

The report therefore treats:

• Citation.
• Brand mention.
• Recommendation.

as separate forms of visibility.

14. Recommendation Is Not the Same as Mention

A brand may appear neutrally within an answer without being endorsed or recommended.

Recommendation tracking therefore requires a separate definition and measurement process.

15. Search-Ranking Overlap Does Not Prove Ranking Causes Citation

A page ranking strongly in Google may also be cited by an AI system.

That does not prove the ranking itself caused the citation.

Both may reflect shared underlying factors such as:

• Authority.
• Relevance.
• Topical depth.
• Freshness.

16. Domain Overlap and Exact-URL Overlap Must Be Separated

An AI system may select a different page from the same domain that ranks in Google.

This means domain-level agreement can be much higher than exact-URL agreement.

The two metrics should not be reported as though they measure the same behaviour.

17. Query Fan-Out Is Only Partly Observable

AI systems can generate multiple related searches behind one prompt.

Researchers may observe some derived searches, but not necessarily every internal retrieval step.

Source-selection studies therefore reveal patterns without fully reconstructing the entire internal process.

18. Reasoning Mode Can Change the Source Set

The same prompt can produce different citations depending on reasoning depth.

This makes AI visibility probabilistic rather than fixed.

Repeated testing is therefore more reliable than one-off screenshots.

19. Model Versions Change

ChatGPT, Gemini, Copilot and other systems are updated frequently.

Changes can affect:

• Retrieval.
• Reasoning.
• Citation behaviour.
• Source selection.
• Recommendation output.

AI-search research therefore has a shorter shelf life than many conventional SEO studies.

20. Freshness Correlation Does Not Prove a Standalone Freshness Factor

AI systems often cite newer information.

However, recently updated content may also be:

• More accurate.
• Better maintained.
• More relevant.
• More comprehensive.

Freshness should therefore be treated as an observed relationship rather than a proven universal ranking factor.

21. Referral Traffic Does Not Capture All AI Influence

A user can encounter a brand in an AI answer without clicking its website.

They may later:

• Search for the brand.
• Visit directly.
• Convert through another channel.

Referral traffic therefore under-measures some potential AI influence.

22. Rapid Referral Growth Can Start From a Small Base

Triple-digit growth in AI referral traffic does not mean AI sends more visits than Google.

Growth rate and absolute scale should always be shown together.

23. First-Party Platform Data Requires Context

Figures published by Google about AI Overview or AI Mode usage provide valuable evidence about scale.

They are first-party platform metrics and should not be treated as independent audience research.

24. Search-Like AI Usage Requires a Definition

Some studies estimate the proportion of ChatGPT activity that resembles conventional search.

The result depends on how “search” is defined.

Comparisons with Google search volume should therefore retain the original methodology.

25. Commercial Impact Requires First-Party Business Data

External research can describe how AI search behaves.

It cannot determine the financial value of AI visibility for an individual organisation.

Businesses should combine external benchmarks with:

• Search Console.
• Analytics.
• AI referral data.
• CRM data.
• Lead data.
• Revenue.

Research Updates

CGO Media treats this page as a living research resource.

Individual statistics may be updated or replaced when:

• Stronger UK-specific evidence becomes available.
• Google changes AI Overview or AI Mode behaviour.
• ChatGPT or other models materially change retrieval or citation behaviour.
• Larger datasets improve confidence.
• Older benchmarks become unrepresentative.
• A material factual or methodological issue is identified.

For the wider CGO Media approach to evidence selection, calculations, limitations and research updates, see the CGO Media Research Methodology.

Frequently Asked Questions — AI Search Statistics UK 2026

The following questions address the main issues surrounding AI search in the UK, including ChatGPT, Google AI Overviews, AI Mode, citations, brand mentions, zero-click search, referral traffic and the relationship between SEO and generative search.

What is AI search?

AI search refers to information discovery using generative AI systems that can interpret a query, retrieve information, synthesise evidence and produce a generated answer.

Examples include:

• Google AI Overviews.
• Google AI Mode.
• ChatGPT with web search.
• Gemini.
• Perplexity.
• Microsoft Copilot.

Is AI search replacing Google in the UK?

Not at present.

Google Search still reaches around four in five UK adults and processes search volume measured in billions of UK queries each month.

The stronger interpretation is that AI search is creating additional discovery routes while Google itself becomes more AI-driven.

How many UK adults use AI tools?

Ofcom’s 2026 research found that 54% of UK adults use AI tools such as ChatGPT, Copilot or Gemini.

Usage is substantially higher among younger adults, reaching 79% among 16–24-year-olds and 74% among 25–34-year-olds.

These figures measure broader AI-tool usage rather than AI search alone.

How popular is ChatGPT in the UK?

Ipsos iris data cited by Ofcom found that ChatGPT reached approximately 13 million UK online adults in June 2025, equal to around 26% of the measured online adult population.

Its audience had grown sharply from the previous year.

What are Google AI Overviews?

Google AI Overviews are generated summaries that appear directly within Google Search for some queries.

They can:

• Summarise information.
• Combine several sources.
• Display citations.
• Reduce the need for users to open several conventional results.

How often do AI Overviews appear?

The answer depends on the country, query type and measurement period.

Ofcom has reported AI Overviews appearing on roughly 30% of Google searches in the UK search environment, while separate behavioural studies have measured different trigger rates on specific query sets.

Longer informational and question-based searches are generally more likely to trigger generated summaries.

Do AI Overviews reduce website clicks?

Current large-scale studies indicate that they can.

Pew found traditional-result clicks were lower when an AI summary appeared, while Ahrefs measured substantial CTR reductions across Google’s organic top ten for informational keywords triggering AI Overviews.

The exact effect varies by query, ranking position and search environment.

Does ranking number one still matter when an AI Overview appears?

Yes, but the traffic value of that ranking can change.

A page may retain position one while receiving fewer clicks because the user encounters a generated answer first.

Ranking position and click-through rate should therefore be monitored separately.

What is zero-click AI search?

Zero-click AI search occurs when the user’s information need is substantially satisfied by the generated answer without a visit to an external website.

This can still create:

• Brand exposure.
• Source visibility.
• Future branded searches.
• Commercial influence.

But it may not create a directly measurable referral session.

What is an AI citation?

An AI citation is a visible reference or link to a source used in a generated answer.

Depending on the platform, it can appear as:

• An inline citation.
• A linked source card.
• A reference panel.
• A source list.

Is an AI citation the same as a brand mention?

No.

An AI system can cite a webpage without naming the brand associated with it.

It can also mention a brand without linking to the brand’s website.

Citation and brand mention should therefore be tracked independently.

What is a ghost citation?

A ghost citation occurs when an AI system cites a source but does not explicitly mention the associated brand in the answer.

This creates source visibility without equivalent brand recognition.

Does a page need to rank in Google’s top ten to be cited by AI?

No.

Current citation research shows many AI-cited URLs do not rank in the top ten for the original user prompt.

AI systems can generate related searches and retrieve the page through a different subquery.

Does SEO still matter for AI search?

Yes.

AI systems continue to depend heavily on accessible web content, search infrastructure and discoverable sources.

SEO supports:

• Crawlability.
• Indexation.
• Topic architecture.
• Relevance.
• Authority.

Generative search adds another source-selection layer rather than making these foundations irrelevant.

What is query fan-out?

Query fan-out occurs when an AI system turns one user question into several related searches or subqueries.

For example, a request to identify the best software platform could trigger searches for:

• Pricing.
• Features.
• Alternatives.
• Integrations.
• User experience.
• Security.

Different sources may then be retrieved for each subtopic.

Why does query fan-out matter for SEO?

It means a page does not necessarily need to rank for the user’s exact original wording to become a citation candidate.

A strong website can be retrieved through one of the underlying subqueries.

This increases the importance of comprehensive topic architecture.

Does deeper AI reasoning change citations?

Yes.

Current ChatGPT research shows that deeper reasoning can increase:

• Web searches.
• Source counts.
• Citation rates.
• Source diversity.

It can also materially change which domains are selected.

Does fresh content perform better in AI search?

Several citation studies show AI systems disproportionately citing newer or recently updated information.

However, this does not prove that publication date alone is an independent ranking factor.

Content should be updated because the underlying information has changed, not merely to create a new timestamp.

Does original research improve AI visibility?

Original research can increase the likelihood that an organisation becomes the primary source for a fact or finding.

Useful assets include:

• Statistics.
• Surveys.
• Datasets.
• Experiments.
• Benchmarks.
• Case studies.

It does not guarantee citation, but it gives AI systems a stronger reason to attribute evidence directly to the organisation.

Do comparison pages matter for AI search?

Yes.

Current research shows comparative queries generate significantly more explicit brand mentions than purely informational queries.

Comparison content can therefore be especially important for provider selection and recommendation visibility.

Does ChatGPT send website traffic?

Yes.

ChatGPT referral traffic is growing rapidly and an increasing number of domains receive traffic from the platform.

However, Google still sends vastly more website traffic overall.

Can AI search create value without a website click?

Yes.

A user can encounter an organisation within an AI-generated answer and later:

• Search the brand.
• Visit directly.
• Return through Google.
• Contact the business through another channel.

This creates attribution challenges because the AI interaction may influence the decision without appearing as the final referral source.

Should businesses optimise separately for ChatGPT, Gemini and Google AI Search?

They should measure them separately, but the underlying optimisation strategy should not become fragmented into completely different systems.

The strongest shared foundations are:

• Technical accessibility.
• Topical relevance.
• Original evidence.
• Brand clarity.
• External authority.
• Fresh information.

How should businesses measure AI search visibility?

A useful framework should track:

• Google rankings.
• AI Overview presence.
• Citations.
• Brand mentions.
• Recommendation inclusion.
• Cross-platform consistency.
• AI referral traffic.
• Commercial outcomes.

These metrics should remain separate before being summarised into any broader visibility model.

How often should AI visibility be checked?

Monthly reporting is a sensible baseline for most commercial programmes.

High-value prompts or rapidly changing sectors may require more frequent monitoring.

Because AI outputs vary, important prompts should also be repeated rather than tested once.

Is GEO replacing SEO?

No.

GEO extends the measurement and optimisation problem beyond rankings and clicks.

A useful distinction is:

The two disciplines increasingly share the same technical and authority foundations.

What should UK businesses prioritise first?

A practical priority order is:

Conclusion — Final Research Findings from the 50 AI Search Statistics UK 2026

The 50 statistics in this report show that AI search is no longer a peripheral development in the UK search market.

Generative search now operates through two powerful routes: standalone AI systems such as ChatGPT and Gemini, and AI-generated experiences embedded directly inside Google Search.

The result is not the disappearance of traditional search. It is the emergence of a broader search ecosystem in which rankings, generated answers, citations, brand mentions, recommendations and zero-click influence increasingly operate together.

Google Remains the Foundation of UK Search

The UK evidence does not support the idea that conventional search has already been replaced.

Google continues to reach a large majority of UK adults and process search volume measured in billions of queries.

For most organisations, ranking visibility and organic traffic therefore remain fundamental.

But Google Is Becoming an AI Search Platform

The distinction between traditional search and AI search is becoming less clear.

AI Overviews and AI Mode increasingly place generated answers directly inside Google’s existing search ecosystem.

Users do not necessarily need to make a conscious decision to switch to an AI assistant in order to experience generative search.

ChatGPT Has Become a Significant UK Discovery Environment

ChatGPT’s UK audience has grown rapidly and now represents a meaningful information-discovery channel.

Its importance is particularly visible in research-led categories such as:

• Technology.
• Education.
• News.
• Shopping.
• Health.

This does not mean ChatGPT has overtaken Google.

It means UK businesses can no longer evaluate search visibility through Google alone.

Search Is Becoming a Multi-Platform Journey

The modern search journey can now move between several environments.

This makes search visibility less linear and attribution more difficult.

AI Overviews Are Changing the Economics of Organic Rankings

One of the strongest findings in the research is that generated summaries can materially reduce organic click-through rates.

A business may retain its search ranking while receiving less website traffic because the user encounters an answer before reaching the conventional result.

This makes ranking position alone a less complete measure of performance.

Zero-Click Influence Is Becoming More Important

AI search can create influence without creating an immediate visit.

Users may:

• See a source cited.
• Encounter a brand recommendation.
• Remember the organisation.
• Search the brand later.
• Convert through another channel.

This creates a growing attribution gap between search influence and measurable referral traffic.

Citation Does Not Equal Brand Visibility

The citation research demonstrates that an organisation can supply information to an AI-generated answer without receiving equivalent brand recognition.

Future search reporting therefore needs to distinguish:

• Source citation.
• Brand mention.
• Recommendation.
• Referral click.

These are different outcomes with different commercial implications.

AI Platforms Have Different Source Behaviours

ChatGPT, Gemini, Perplexity, AI Overviews and AI Mode do not select or present sources in identical ways.

One platform may cite an organisation frequently while another mentions the brand more often.

There is therefore no single universal AI ranking.

Traditional Rankings Do Not Fully Explain AI Citations

AI citation datasets show substantial differences between conventional Google rankings and the sources selected for generated answers.

This means AI search adds another stage:

Query Fan-Out Changes What Businesses Compete For

Generative systems can break one user question into many secondary searches.

This makes topic architecture increasingly important.

Businesses benefit from covering the full information space around a subject rather than depending on one page targeting one keyword.

Reasoning Depth Changes Source Selection

Higher-reasoning AI responses generate more searches, more citations and more source diversity.

They can also select materially different domains from the same prompt.

This makes AI visibility probabilistic rather than fixed.

AI Search Measurement Needs Repetition

One prompt, one answer and one screenshot are not enough to establish durable visibility.

Reliable AI-search measurement needs:

• Repeated prompt testing.
• Platform-specific tracking.
• Historical comparison.
• Consistent prompt sets.

Original Research Becomes More Valuable in an AI Search Environment

Generative systems need evidence from which to construct answers.

Organisations producing primary information can become citation sources rather than merely republishing somebody else’s findings.

This increases the strategic value of:

• Statistics.
• Surveys.
• Datasets.
• Research papers.
• Experiments.
• Benchmarks.

Freshness Is Becoming Part of Source Competition

AI citation research shows a consistent preference for relatively recent information in many search environments.

For changing topics, content maintenance therefore becomes part of visibility strategy.

High-priority pages should be reviewed when:

• Statistics change.
• Prices change.
• Products change.
• Research changes.
• Market conditions change.

AI Referral Traffic Is Growing — But Google Still Dominates

ChatGPT referral traffic is growing rapidly from a relatively small base.

Google continues to send vastly more website traffic.

This means businesses should not replace conventional SEO investment simply because AI referral growth rates look dramatic.

SEO and GEO Are Converging

The strongest foundations for AI visibility are closely related to strong search foundations:

• Technical accessibility.
• Topical relevance.
• Strong site architecture.
• Original evidence.
• Authority.
• Fresh information.

The difference increasingly lies in what is measured after discovery.

Traditional SEO measures rankings and clicks.

GEO additionally measures retrieval, citation, brand mention and recommendation.

Search Performance Needs a Broader Measurement Model

The emerging search environment requires organisations to measure both conventional and generative visibility.

Final CGO Media Research Position

Traditional search optimisation asked:

Can this page rank and earn the click?

AI search adds a second strategic question:

Can this organisation become one of the sources generative systems repeatedly retrieve, cite, mention and recommend?

Research Usage, Citation & Press

The AI Search Statistics UK 2026 research is intended to support journalists, researchers, academics, businesses, agencies and organisations examining how generative AI is changing search behaviour, discovery, citations, website traffic and brand visibility.

Statistics, findings and CGO Media analysis from this report may be referenced provided that the original geography, platform, methodology and evidence classification are preserved.

How to Cite This Research

A suggested citation format is:

Where an individual statistic is quoted, the original study identified beneath that statistic should also be consulted and referenced where appropriate.

Journalists & Media

Journalists may quote individual statistics, findings and CGO Media analysis from this report in editorial coverage.

Where possible, please:

• Credit CGO Media.
• Link to the original research page.
• Retain whether the statistic is UK-specific, US, international or global.
• Preserve the named platform.
• Preserve the original measurement date.
• Distinguish AI-tool usage from AI-search usage.
• Distinguish citations from brand mentions.
• Distinguish measured data from CGO Media interpretation.

For press enquiries, methodology questions, expert commentary or supporting material, visit:

CGO Media Press & Media

Researchers & Academics

Researchers may reference this evidence synthesis provided that the original evidence type, geography and platform remain identifiable.

The report combines:

• UK regulatory and audience-measurement evidence.
• UK online-search behaviour research.
• US behavioural studies.
• International AI citation research.
• AI Overview CTR studies.
• Query fan-out and reasoning studies.
• AI referral-traffic analysis.
• First-party Google platform data.

The 50 statistics should therefore be interpreted as a structured evidence synthesis rather than one single statistical sample.

Businesses & Organisations

Businesses may use the findings for:

• SEO and GEO strategy.
• AI-search benchmarking.
• Content investment decisions.
• Research strategy.
• Digital PR planning.
• Board and management presentations.
• AI visibility reporting.
• Search attribution analysis.

External benchmarks should be combined with first-party business data wherever possible.

Useful internal metrics include:

• Google Search Console.
• Analytics.
• AI referral sessions.
• Prompt-monitoring data.
• CRM records.
• Leads.
• Pipeline.
• Revenue.

Using UK Adoption Statistics

UK-specific AI adoption figures should retain their exact scope.

The second version incorrectly converts broader AI-tool usage into AI-search usage.

Using ChatGPT Reach Data

Audience reach, visits and search volume should remain separate.

This does not mean 13 million people used ChatGPT specifically for web search during that month.

Using Google AI Overview Statistics

AI Overview prevalence varies by:

• Country.
• Query type.
• Intent.
• Measurement date.
• Dataset.

A trigger rate measured in one study should not be presented as a universal Google percentage.

Using Click-Through Rate Statistics

CTR findings should retain the relevant methodology.

Avoid converting this into:

“AI Overviews reduce every number-one ranking by 58%.”

Using Zero-Click Statistics

A zero-click search does not necessarily mean that the search produced no commercial value.

A generated answer can influence:

• Brand awareness.
• Future brand searches.
• Direct visits.
• Offline decisions.

Zero-click behaviour should therefore be separated from zero influence.

Using AI Citation Statistics

Citation rates should remain tied to the platform and query set in which they were measured.

Avoid presenting this as:

“ChatGPT cites sources 87% of the time.”

Using Brand-Mention Statistics

Brand mentions and citations are different outcomes.

For example:

This distinction is central to interpreting generative visibility correctly.

Using Search-Ranking Overlap Statistics

Low overlap between AI citations and Google’s top ten should not be interpreted as evidence that SEO no longer matters.

AI systems can:

• Generate fan-out queries.
• Retrieve pages through related searches.
• Select different pages from the same domain.
• Apply additional source filtering.

A responsible interpretation is that conventional ranking and generative citation selection are related but not identical processes.

Using Query Fan-Out Statistics

Measured fan-out behaviour should remain tied to the model and reasoning configuration studied.

The number of subqueries generated for one model configuration should not be presented as a fixed number applying to every generative search system.

Using Reasoning-Mode Statistics

Reasoning depth can materially change citations and source diversity.

This means one prompt output cannot be treated as a permanent ranking result.

Where AI visibility is commercially important, repeated testing is preferable.

Using Freshness Statistics

Research showing newer content being cited more often should not be converted into claims that changing a publication date will improve AI visibility.

The evidence supports substantive content maintenance rather than cosmetic timestamp changes.

Using AI Referral Traffic Statistics

Rapid growth percentages should be shown alongside absolute scale.

For example, ChatGPT referral traffic can grow by more than 200% while Google still sends vastly more website traffic overall.

Both facts are important.

Using First-Party Google Data

Google’s AI Overview and AI Mode audience figures provide important evidence about platform scale.

They should be identified as first-party global platform data rather than independent UK audience research.

Using CGO Media Calculations

Where CGO Media performs arithmetic using published source figures, the result should remain labelled:

CGO Media Calculation

These calculations do not alter the underlying research and should remain traceable to the source values.

Charts, Infographics & Visualisations

Charts, diagrams and infographics created from this report may be referenced with attribution to:

CGO Media Research Team — AI Search Statistics UK 2026

Where a visual reproduces third-party research, the original underlying source should remain identifiable.

Research Corrections & Updates

AI search changes rapidly.

CGO Media may update this report when:

• New UK-specific evidence becomes available.
• Google changes AI Overview or AI Mode behaviour.
• ChatGPT or other models materially change retrieval or citation behaviour.
• Larger studies improve the evidence base.
• Existing benchmarks become outdated.
• A material factual or methodological issue is identified.

Research Methodology

For the wider CGO Media approach to evidence classification, source selection, calculations, limitations and research updates, visit:

CGO Media Research Methodology

Sources & References

The following sources provide the primary evidence used throughout AI Search Statistics UK 2026.

CGO Media has prioritised UK regulatory and audience-measurement evidence, first-party platform data, independent behavioural research and large-scale specialist search studies with published methodologies.

Ofcom — Adults’ Media Use and Attitudes 2026

Ofcom — The Era of Answer Engines

Ofcom — Online Nation 2025

Pew Research Center — Google AI Summaries & Click Behaviour

Ahrefs — AI Overviews & Organic Click-Through Rate

Semrush — AI Citations & Brand Mentions

Ahrefs — AI Citations vs Google Rankings

Ahrefs — Google AI Overview Citation Selection

Ahrefs — ChatGPT Domain vs Exact-URL Overlap

Semrush — ChatGPT Reasoning & Query Fan-Out

Ahrefs — AI Citation Freshness

Semrush — ChatGPT Referral Traffic

Ahrefs — ChatGPT Search Volume vs Google Traffic

Google — AI Overviews & AI Mode Scale

CGO Media Calculations

Where CGO Media performs arithmetic using published source values, the result is labelled CGO Media Calculation.

Examples include:

• ChatGPT’s approximate UK audience growth between June 2024 and June 2025.
• Relative differences between ChatGPT and Google referral-category shares.
• Percentage differences derived directly from published source values.

These calculations do not create new underlying observations and should remain traceable to the original values.

How the Evidence Should Be Read

The studies in this report use different methodologies.

They include:

• UK audience measurement.
• Consumer surveys.
• Browser-behaviour tracking.
• Search-result analysis.
• Google Search Console data.
• AI citation databases.
• Prompt testing.
• Clickstream data.
• Web analytics.
• First-party platform reporting.

The 50 statistics should therefore be read as a structured evidence synthesis, not as though all figures came from one population or experiment.

Research Review & Update Policy

AI search is evolving rapidly.

CGO Media will review this report periodically and may amend or replace statistics when:

• Stronger UK-specific research becomes available.
• Google changes AI Overview or AI Mode behaviour.
• ChatGPT or other AI platforms materially change retrieval or citation systems.
• More representative datasets become available.
• Older benchmarks become unrepresentative.
• A factual or methodological issue is identified.

For further information about how CGO Media evaluates research evidence, calculations and limitations, see:

CGO Media Research Methodology