AI Search Traffic Statistics UK 2026

Section 3 — Zero-Click Search, Open-Web Traffic & Search Journey Statistics

AI Search Traffic needs to be understood within a much longer shift toward zero-click discovery.

Google had already been answering increasing numbers of searches directly through maps, featured snippets, knowledge panels, shopping modules and other search features before generative AI became widely available.

AI Overviews and AI Mode extend that process by allowing users to receive much more complete answers without necessarily visiting an external website.

The result is an increasingly important distinction between:

Search Visibility

and:

Search Referral Traffic.

The statistics below examine how much Google activity now ends without a click, how much traffic reaches the open web and how AI-native Google experiences compare with conventional search.

Data classification: Statistics 21–25 use US and European clickstream benchmarks from Similarweb, Datos and SparkToro. Statistics 26–30 use large-scale US/international behavioural datasets from Semrush and Datos. These figures are not UK-specific and are therefore presented as comparative benchmarks.

21. 68% of US Google searches ended without any click in early 2026

US Clickstream Benchmark. SparkToro reported analysis of Similarweb clickstream data showing that 68% of US Google searches ended without a click during the first four months of 2026.

This means roughly two thirds of measured Google searches did not result in the user clicking:

• an open-web website;
• a Google-owned property;
• or a paid search result.

The statistic illustrates how far Search Visibility has separated from website traffic.

A search can still expose a user to:

• a brand;
• a business;
• a product;
• a source;
• or a recommendation

without producing an attributable visit.

AI traffic implication: Search Traffic should increasingly be treated as only one outcome of Search Visibility.

The broader model is:

Search Exposure → Possible Click / Brand Recognition / Further Search / Later Visit

rather than assuming every successful search appearance should immediately produce a website session.

Source:

SparkToro — The Click Is Now Optional, 2026

22. 59.7% of European Union Google searches ended without a click in the 2024 Datos study

European Union Benchmark. SparkToro and Datos analysed large-scale clickstream behaviour and found that 59.7% of Google searches in the European Union ended without a click in 2024.

Zero-click searches included situations where users:

• ended the browsing session;
• changed or refined the query;
• or otherwise did not click a result from the search page.

The finding is particularly useful for UK organisations because it provides a geographically closer comparison than US-only search data, although it should not be described as UK behaviour.

AI traffic implication: High zero-click behaviour was already established in European search before the later expansion of AI Overviews and AI Mode.

Generative AI therefore accelerates an existing structural trend rather than creating zero-click search from nothing.

Source:

SparkToro & Datos — 2024 Zero-Click Search Study

23. 58.5% of US Google searches ended without a click in the same 2024 study

US Benchmark. The SparkToro and Datos study found that 58.5% of Google searches in the United States ended without a click.

The US and EU figures were therefore remarkably similar:

• EU — 59.7%
• US — 58.5%

This suggests that zero-click behaviour was not confined to one regulatory or cultural search environment.

Instead, it reflected broader changes in how Google Search itself operates.

AI traffic implication: Businesses should avoid interpreting falling organic referral rates as evidence that search itself is necessarily shrinking.

Search activity can remain strong while a smaller proportion of that activity leaves the search platform.

Source:

SparkToro & Datos

24. Only 374 open-web clicks were generated for every 1,000 EU Google searches

European Union Benchmark. SparkToro and Datos calculated that for every 1,000 Google searches in the European Union, approximately 374 clicks reached the open web.

The open web was defined as destinations that were neither:

• Google-owned properties;
• nor websites receiving a paid Google advertising click.

Put differently, fewer than four open-web visits were produced for every ten Google searches in the dataset.

This matters because publishers and businesses frequently interpret search volume as though it were equivalent to available website traffic.

It is not.

AI traffic implication: Search-demand forecasts should distinguish between:

Total Search Volume

and:

Realistically Available External Traffic.

The second figure can be substantially smaller.

Source:

SparkToro & Datos — Open-Web Search Traffic Study

25. Only 360 open-web clicks were generated for every 1,000 US Google searches

US Benchmark. The equivalent US figure was even lower.

For every 1,000 Google searches in the United States, approximately 360 clicks reached the open web.

This means roughly 640 of every 1,000 searches did not produce an open-web click within the study’s measurement model.

Some users:

• ended their search;
• searched again;
• clicked Google-owned services;
• or interacted with paid results.

AI traffic implication: Organic Search Traffic represents only part of total Search Demand.

As generative answers expand, businesses should increasingly ask:

“What proportion of the searches in our market can realistically produce an external visit?”

rather than estimating traffic solely from headline keyword volume.

Source:

SparkToro & Datos

26. Around 22% of Google searches in the Datos study resulted in another search

US and European Behavioural Benchmark. SparkToro reported that approximately 22% of searches resulted in the user conducting another search.

This is an important form of zero-click behaviour because not every search without a click represents a satisfied user.

The searcher may instead:

• refine the wording;
• add more detail;
• change the question;
• or begin a related search.

This means zero-click behaviour can represent several very different outcomes:

Satisfied Without Clicking

or:

Search Continued Through Another Query.

AI traffic implication: Search Journey analysis should consider sequences of searches rather than treating each query as an isolated event.

AI systems make this even more relevant because conversational interfaces encourage users to refine questions repeatedly within the same session.

Source:

SparkToro & Datos

27. Around 20% of Google searches that generated at least one click produced more than one click

US and European Behavioural Benchmark. SparkToro reported that approximately 20% of Google searches producing at least one click resulted in clicks on multiple results.

This demonstrates that search journeys are not always winner-takes-all.

A user may visit:

• several retailers;
• multiple professional-service providers;
• different publishers;
• or several product pages

before deciding which source to trust or which provider to select.

AI traffic implication: Being included in the consideration set can still create commercial value even when another website receives the first click.

AI Search may compress this multi-site research process by performing more comparison before the external visit occurs.

The eventual AI-referred user may therefore arrive considerably later in the decision journey.

Source:

SparkToro & Datos — Search Click Behaviour

28. 92–94% of Google AI Mode sessions produced no external-domain visit in Semrush’s study

US AI Search Benchmark. Semrush analysed clickstream data covering almost 69 million Google Search sessions between May and July 2025.

It found that approximately 92–94% of Google AI Mode sessions were zero-click with respect to external websites.

AI Mode represents a more conversational form of Google Search in which the system can:

• answer complex questions;
• conduct multi-step research;
• compare information;
• and provide cited responses.

The extremely high zero-click rate therefore demonstrates how dramatically traffic economics can change when the answer interface becomes more complete.

AI traffic implication: AI Mode should not be evaluated primarily as a referral channel.

Its value may increasingly involve:

• Brand Visibility;
• citation;
• Recommendation Visibility;
• and later-stage customer discovery.

Source:

Semrush — Google AI Mode’s Early Adoption and SEO Impact

29. Only 6–8% of Google AI Mode sessions led users to an external domain

US AI Search Benchmark. Within the same Semrush clickstream analysis, only approximately 6–8% of AI Mode sessions resulted in the user visiting an external domain.

This means that direct referral traffic represented a small minority of AI Mode interactions.

The statistic highlights the growing difference between:

being visible within AI Search

and:

receiving AI Search traffic.

A business may appear within an AI-generated answer without generating an immediate website visit.

AI traffic implication: AI Search reporting should therefore use two separate KPI families:

• Visibility metrics — mentions, citations, recommendations, source inclusion;
• Traffic metrics — sessions, landing pages, conversions and revenue from AI referrals.

Combining the two can hide important changes in performance.

Source:

Semrush — AI Mode Clickstream Study

30. In a controlled Semrush comparison, zero-click rates fell from 38.1% to 36.2% after AI Overviews appeared for the same keywords

International Search Benchmark. Semrush and Datos tracked a group of keywords that did not trigger AI Overviews in May 2025 but did trigger them by October.

For that same set of keywords, the measured zero-click rate changed from:

• 38.1% before an AI Overview appeared
• to 36.2% after the AI Overview appeared.

This finding is important because it introduces necessary nuance into the wider discussion about AI-generated answers.

AI Overviews are strongly associated with search categories that already have high zero-click behaviour, but that does not mean every AI Overview directly causes fewer clicks in every dataset.

Different studies can produce different results because they measure:

• different query sets;
• different time periods;
• different devices;
• different countries;
• and different definitions of a click.

AI traffic implication: Businesses should avoid assuming a universal traffic-loss percentage.

The correct approach is to measure the organisation’s own:

Query → AI Feature → Impression → Click → Conversion

data over time.

Market-level studies provide useful benchmarks, but first-party performance remains the most relevant evidence.

Source:

Semrush — What to Do About AI Overviews Traffic Loss

Section 4 — AI Referral Traffic Growth Statistics

Zero-click behaviour does not mean AI systems generate no traffic.

At the same time that generative platforms are answering more questions directly, users are also clicking through from AI systems to:

• retailers;
• travel companies;
• financial-services providers;
• technology businesses;
• publishers;
• research organisations;
• and other websites.

This creates a new acquisition channel:

AI Referral Traffic.

The absolute share of web traffic generated by AI remains considerably smaller than traditional organic search, but the growth rates recorded across several industries are substantial.

The following statistics examine how quickly generative AI referrals expanded during 2025 and early 2026.

Data classification: Statistics 31–38 use US industry-level data from Adobe Analytics. Statistics 39–40 use US clickstream data from Semrush and Datos. These should be treated as international benchmarks rather than UK-specific traffic measurements.

31. AI referral traffic to US retail websites increased 693.4% during the 2025 holiday season

US Ecommerce Benchmark. Adobe Analytics reported that traffic from generative AI platforms to US retail websites increased by 693.4% year on year during the November–December 2025 holiday season.

The analysis is based on Adobe’s large ecommerce dataset covering more than one trillion visits to US retail websites.

The increase demonstrates how rapidly generative AI has developed as a Product Discovery and retailer-referral channel.

Consumers can increasingly ask AI systems to:

• identify suitable products;
• compare alternatives;
• find retailers;
• research specifications;
• compare prices;
• and evaluate purchasing options.

A proportion of those journeys now end with a click to a retailer.

AI traffic implication: Ecommerce organisations should establish dedicated reporting for AI referrals before the channel becomes large enough to materially affect overall acquisition reporting.

The relevant baseline should include:

• AI referral sessions;
• AI landing pages;
• products viewed;
• Add-to-Cart behaviour;
• transactions;
• and revenue.

Source:

Adobe Digital Insights — AI-Driven Traffic Surges Across Industries

32. AI-driven retail traffic increased 769% year on year in November 2025

US Ecommerce Benchmark. Adobe reported that generative AI referral traffic to retail websites increased by 769% year on year during November 2025.

This was even higher than the overall November–December holiday-season growth rate.

November includes major purchasing periods such as:

• Black Friday;
• Cyber Monday;
• early Christmas shopping;
• and other promotional events.

The result suggests that AI-assisted discovery is beginning to participate in high-intent shopping periods rather than being restricted to general informational research.

AI traffic implication: AI referral performance should be analysed by season and purchasing event.

A retailer may experience disproportionately high AI traffic during periods when users are actively comparing:

• products;
• offers;
• gift ideas;
• and retailers.

Source:

Adobe Digital Insights

33. AI-driven retail traffic increased 673% year on year in December 2025

US Ecommerce Benchmark. Adobe reported a further 673% year-on-year increase in AI-driven retail traffic during December 2025.

The continued growth after November indicates that AI referrals were not driven by a single isolated promotional event.

Instead, they remained substantially above the previous year’s level throughout the core holiday-shopping period.

AI traffic implication: Retail organisations should monitor whether AI traffic follows:

• seasonal demand;
• product launches;
• sales periods;
• or specific types of commercial query.

This can help distinguish general platform growth from commercially meaningful AI-assisted Product Discovery.

Source:

Adobe — AI Referral Traffic Research

34. AI referral traffic to travel websites increased 539% during the 2025 holiday season

US Travel Benchmark. Adobe reported that AI-driven referral traffic to travel websites increased by 539% year on year during the 2025 holiday period.

Travel is particularly well suited to conversational discovery because consumers frequently need help comparing:

• destinations;
• hotels;
• flights;
• routes;
• activities;
• budgets;
• and itineraries.

A single AI conversation can combine several stages of the traditional travel-search process.

For example:

Destination Research → Hotel Comparison → Itinerary Planning → Booking Site Referral

AI traffic implication: Travel businesses should consider whether their content is sufficiently clear and authoritative to support both recommendation inclusion and eventual referral traffic.

Source:

Adobe Digital Insights — AI-Sourced Traffic Insights 2025

35. AI referral traffic to financial-services websites increased 266%

US Financial Services Benchmark. Adobe reported that generative AI traffic to financial-services websites increased by 266% year on year during the 2025 holiday period.

Financial-services discovery can involve complex research relating to:

• banking products;
• loans;
• credit;
• investments;
• insurance;
• and financial guidance.

Generative AI is particularly suited to explaining differences between products and simplifying complex terminology before a user visits a provider.

AI traffic implication: Financial organisations should monitor AI referral traffic separately from conventional organic traffic because the user may arrive after completing a substantial amount of preliminary research inside the AI platform.

That can create a different visitor profile from a conventional search click.

Source:

Adobe Digital Insights

36. AI-driven traffic to banking websites increased 344%

US Banking Benchmark. Within Adobe’s sector analysis, AI-driven traffic to banking websites increased by approximately 344% year on year.

Banking represents a particularly important category because consumers may use AI to research:

• accounts;
• fees;
• interest rates;
• credit products;
• banking features;
• and provider differences.

The result indicates that AI-assisted discovery is expanding beyond relatively low-risk informational searches into sectors involving significant financial decisions.

AI traffic implication: Trust, factual accuracy and source authority become particularly important when AI-driven traffic relates to high-stakes financial products.

Organisations should ensure that:

• product information is current;
• fees are clear;
• eligibility information is accurate;
• and external descriptions of the organisation are consistent.

Source:

Adobe — AI-Sourced Traffic Insights

37. AI referral traffic to technology and software websites increased 120%

US Technology Benchmark. Adobe reported that generative AI referral traffic to technology and software websites increased by 120% year on year.

The percentage increase was lower than retail, travel and financial services, but Adobe found that technology and software had the highest overall AI-driven visit share among the industries measured.

This distinction is important.

Growth percentage and current market penetration are not the same thing.

A sector can show a lower percentage growth rate because AI referral traffic was already more developed at the beginning of the measurement period.

AI traffic implication: Businesses should track both:

• AI traffic growth rate;
• and AI traffic share.

A rapidly growing channel can still be economically small, while a slower-growing channel may already represent a meaningful proportion of discovery.

Source:

Adobe Digital Insights — 2025 AI Traffic Report

38. AI referral traffic to media and entertainment websites increased 92%

US Media Benchmark. Adobe reported that AI-driven traffic to media and entertainment websites increased by 92% year on year.

Although this was the lowest growth rate among the major industries highlighted by Adobe, the category still experienced substantial expansion.

Media websites can receive AI referrals when users seek:

• news;
• reviews;
• entertainment information;
• background research;
• or primary-source reporting.

However, publishers also face a particular challenge because generative systems can summarise their information without requiring a website visit.

AI traffic implication: Media organisations may simultaneously experience:

• new AI referral opportunities;
• and reduced conventional search traffic through AI-generated summaries.

This makes citation and referral measurement especially important for publishing businesses.

Source:

Adobe Digital Insights

39. ChatGPT outbound referral traffic increased 206% year on year

US Clickstream Benchmark. Semrush analysed more than 1 billion lines of US clickstream data covering desktop and mobile activity between October 2024 and February 2026.

The study found that outbound referral traffic from ChatGPT increased by 206% between January 2025 and January 2026.

This is important because ChatGPT’s overall traffic had begun to plateau toward the end of 2025.

Despite that stabilisation in platform visits, ChatGPT continued sending substantially more users out to the wider web.

This suggests that:

platform usage growth

and:

referral traffic growth

do not necessarily move at the same rate.

AI traffic implication: Businesses should measure outbound AI referrals directly rather than assuming they can infer referral opportunity from total ChatGPT usage alone.

A platform can become more effective at sending visitors to websites even when its own overall audience growth begins to slow.

Source:

Semrush — ChatGPT Traffic Analysis: 17 Months of Clickstream Data

40. The number of domains receiving ChatGPT referrals peaked at around 260,000 per month

US Clickstream Benchmark. Semrush found that the number of unique domains receiving ChatGPT referral traffic expanded significantly during its 17-month study.

In October 2024, ChatGPT referred traffic to approximately 71,000 unique domains per month within the sample.

That number reached a peak of approximately 260,000 domains in October 2025.

By February 2026, approximately 170,000 unique domains per month were still receiving ChatGPT referrals.

The peak represented more than 3.5 times the number of domains receiving referrals at the start of the period.

This demonstrates that AI referral traffic is becoming distributed across a much broader section of the web.

It is not limited only to a handful of major publishers or technology platforms.

AI traffic implication: The opportunity to receive AI referral traffic is expanding across more organisations and industries.

However, distribution remains uneven.

Semrush separately found that more than 30% of ChatGPT referral traffic was still concentrated among only ten domains by February 2026.

Businesses therefore need both:

AI Citation Eligibility

and:

enough authority to compete for referral visibility.

Source:

Semrush — ChatGPT Referral Traffic Study

Section 5 — AI Referral Engagement & Visitor Quality Statistics

Rapid traffic growth does not automatically mean that AI referrals are commercially valuable.

A new acquisition channel becomes more important when the visitors it generates also demonstrate meaningful engagement after they reach the website.

Several large-scale datasets now indicate that AI-referred visitors can behave differently from visitors arriving through established digital channels.

In retail in particular, AI-referred users have increasingly shown:

• lower bounce rates;
• longer sessions;
• more page views;
• and stronger engagement.

This may reflect the nature of conversational discovery.

An AI assistant can help a user:

• clarify a requirement;
• compare alternatives;
• understand a product;
• and narrow a shortlist

before the user reaches the external website.

The resulting visitor may therefore arrive with greater context and stronger intent than a user making an exploratory click from a conventional search result.

Data classification: Statistics 41–49 use US retail engagement data from Adobe Digital Insights. Statistic 50 uses Similarweb desktop clickstream data covering ChatGPT referrals before and after a major May 2026 interface change.

41. AI-referred retail visitors had a 23% lower bounce rate by February 2025

US Ecommerce Benchmark. Adobe reported that by February 2025, visitors arriving at US retail websites from generative AI sources had a 23% lower bounce rate than visitors from other traffic sources.

The finding was based on Adobe Analytics data covering extremely large volumes of US ecommerce activity.

Lower bounce does not automatically mean higher revenue, but it indicates that AI-referred visitors were less likely to arrive and leave without further engagement.

This can be consistent with a user who has already completed part of the research process before clicking.

AI traffic implication: AI referral traffic should not be evaluated only by volume.

Businesses should compare AI traffic with other channels using:

• bounce rate;
• engagement rate;
• time on site;
• pages viewed;
• conversion;
• and revenue per visit.

Source:

Adobe Digital Insights — The Explosive Rise of Generative AI Referral Traffic

42. AI-referred retail visitors viewed 12% more pages per visit by February 2025

US Ecommerce Benchmark. Adobe reported that visitors arriving from generative AI sources viewed 12% more pages per visit than traffic from other sources by February 2025.

Additional page depth can indicate that users are:

• exploring more products;
• reading more information;
• comparing alternatives;
• or moving further through the purchasing journey.

The result is particularly interesting because AI assistants can conduct part of the research process before the click occurs.

The user may arrive already knowing:

• what type of product they need;
• which features matter;
• which brands are relevant;
• or why the retailer was recommended.

AI traffic implication: AI referral landing pages should make it easy for users to continue a research journey rather than forcing them to restart it.

Relevant pages should provide clear access to:

• Product Information;
• comparisons;
• pricing;
• reviews;
• alternatives;
• and conversion actions.

Source:

Adobe Digital Insights

43. AI-referred retail visits lasted 41% longer by February 2025

US Ecommerce Benchmark. Adobe found that visits originating from AI sources lasted 41% longer than visits from non-AI traffic sources by February 2025.

Longer sessions can reflect stronger alignment between:

• the user’s requirement;
• the recommendation;
• and the content encountered after the click.

This supports a potential AI Search journey in which:

AI Research → Qualified Recommendation → Relevant Landing Page → Deeper Engagement

rather than:

Broad Search → Exploratory Click → Immediate Exit

AI traffic implication: Referral quality should be assessed alongside referral volume.

A smaller traffic source can still become strategically important if its visitors engage substantially more deeply.

Source:

Adobe — Generative AI Referral Traffic

44. AI-referred shoppers were 13.6% more engaged than non-AI visitors by October 2025

US Ecommerce Benchmark. Adobe’s October 2025 retail data reported that shoppers arriving from generative AI sources were 13.6% more engaged than visitors from non-AI channels.

The finding shows that the engagement advantage identified earlier in 2025 had not disappeared as AI referral traffic expanded.

Instead, stronger visitor quality remained visible as the channel matured.

AI traffic implication: Growth in an acquisition channel does not necessarily have to reduce traffic quality.

Organisations should monitor whether AI referral expansion is accompanied by:

• stable engagement;
• improving engagement;
• or declining engagement.

That distinction can help determine whether AI traffic growth reflects qualified discovery or simply greater platform exposure.

Source:

Adobe Digital Insights — October 2025 Retail Data

45. AI-referred retail visitors had a 31% lower bounce rate by October 2025

US Ecommerce Benchmark. Adobe reported that by October 2025, visitors arriving from generative AI sources were 31% less likely to leave immediately than visitors from other channels.

Earlier in the year the reported advantage had been 23%.

The widening gap suggests that AI recommendations may have become better aligned with user intent as:

• AI models improved;
• search functionality developed;
• and consumers became more accustomed to using AI for shopping research.

AI traffic implication: Changes in AI traffic quality should be tracked longitudinally.

Historical benchmarks can become obsolete quickly as both models and consumer behaviour evolve.

Source:

Adobe Digital Insights

46. AI-referred retail visitors spent 44% longer on site in October 2025

US Ecommerce Benchmark. Adobe reported that AI-referred shoppers spent 44% more time on retail websites than non-AI visitors during its October 2025 measurement.

This represented a slight increase from the 41% advantage Adobe had reported in February 2025.

The continued difference suggests that deeper engagement was not merely an early-adopter phenomenon.

AI traffic implication: AI visitors may warrant separate behavioural benchmarks rather than being grouped with:

• organic search;
• direct;
• social;
• email;
• or referral traffic generally.

A separate AI segment makes it easier to identify whether these users behave differently after arrival.

Source:

Adobe — Holiday Shopping Actuals

47. AI-referred shoppers viewed 12% more pages per visit in October 2025

US Ecommerce Benchmark. Adobe’s October 2025 data also found that AI-referred shoppers viewed 12% more pages per visit than visitors from non-AI sources.

The result was broadly consistent with Adobe’s February 2025 measurement.

This stability is useful because it suggests that additional page depth persisted even as the total volume of AI referrals increased significantly.

AI traffic implication: Website teams should identify which pages AI users visit after landing and whether those journeys lead naturally toward:

• Product Categories;
• comparisons;
• pricing;
• trust information;
• checkout;
• or lead generation.

AI referral optimisation therefore extends beyond simply earning the citation or click.

Source:

Adobe Digital Insights

48. AI-referred shoppers were 33% less likely to leave immediately during the 2025 holiday season

US Ecommerce Benchmark. Adobe reported that during the 2025 holiday season, shoppers arriving from generative AI assistants were 33% less likely to leave a retail website immediately than visitors from non-AI traffic sources.

The bounce-rate advantage had increased from earlier periods:

• 23% lower in February 2025;
• 31% lower in October 2025;
• 33% lower during the holiday season.

This progression suggests that the quality of AI referral matching improved during 2025 within Adobe’s retail dataset.

AI traffic implication: One potential value of conversational discovery is that the AI system may function as an initial qualification layer.

The user can clarify their requirement before being sent to the retailer.

That can potentially create:

Fewer Irrelevant Clicks → More Qualified Visits

although the extent of this effect will vary by industry and platform.

Source:

Adobe Digital Insights — AI Traffic Surges Across Industries

49. AI-referred shoppers spent 45% more time on site and viewed 13% more pages during the 2025 holiday season

US Ecommerce Benchmark. Adobe found that during the 2025 holiday period, AI-referred retail visitors:

• spent 45% more time on the website;
• and viewed 13% more pages per visit

than visitors from other traffic sources.

The two measures reinforce the lower bounce-rate findings and suggest that AI-referred users were not simply arriving in greater numbers.

They were also exploring retail websites more deeply.

Adobe separately reported that 64% of consumers using AI for shopping were satisfied with the links AI assistants provided, while more than 55% reported clicking links supplied by AI assistants.

AI traffic implication: AI referral quality appears increasingly connected with recommendation relevance.

For businesses, the goal should therefore be more specific than simply increasing AI mentions.

The stronger objective is:

Relevant AI Recommendation → Appropriate Landing Page → High-Quality Engagement

Source:

Adobe Digital Insights — AI-Sourced Traffic Insights 2025

50. ChatGPT-referred page views increased 24% after its May 2026 brand-link update

International / Similarweb Clickstream Benchmark. Similarweb analysed ChatGPT referral behaviour before and after an interface change introduced on 7 May 2026.

The update made brand names and links more prominent inside ChatGPT answers.

Similarweb found that average page views per ChatGPT-referred visit increased from:

• 3.8 pages before the update
• to 4.7 pages afterwards.

That represented a 24% increase in page depth.

Average time on site also increased from:

• 3.5 minutes
• to 3.9 minutes,

an improvement of approximately 11%.

The dataset covered desktop clickstream activity from 30 April to 20 May 2026.

The result is particularly important because it demonstrates that AI referral behaviour can change significantly when the AI platform changes how prominently external brands and links are displayed.

AI traffic implication: AI referral traffic is partly dependent on platform design.

Businesses therefore need to recognise that traffic can change because of:

• their own visibility;
• model behaviour;
• citation selection;
• and changes to the AI interface itself.

The measurement model becomes:

AI Visibility × Link Prominence × User Intent × Landing-Page Relevance = Referral Engagement

Source:

Similarweb — ChatGPT Referral Traffic Near Triples Overnight

Section 6 — AI Referral Conversion, Revenue & Commercial Value Statistics

Engagement becomes commercially meaningful when AI-referred visitors begin converting into customers.

Adobe’s retail data shows one of the clearest developments in AI Search Traffic during 2025.

At the beginning of the year, AI-referred visitors converted substantially less often than users arriving through established digital channels.

Over the following months, that gap narrowed, disappeared and eventually reversed.

By the 2025 holiday shopping season, Adobe reported that AI-referred retail visitors were converting more frequently than visitors from other traffic sources.

This represents an important shift in AI Traffic economics.

The development can be summarised as:

Research Channel → Qualified Referral Channel → Commercial Acquisition Channel

Data classification: Statistics 51–60 use Adobe Analytics data covering US retail websites. Adobe’s underlying commerce dataset covers more than one trillion visits to US retail websites. These figures should therefore be treated as large-scale US ecommerce benchmarks rather than UK-specific conversion statistics.

51. AI-referred retail visitors were 49% less likely to convert in January 2025

US Ecommerce Benchmark. Adobe reported that in January 2025, traffic arriving from generative AI sources was 49% less likely to convert than traffic from non-AI sources.

This demonstrates how immature AI referral traffic still was at the beginning of 2025.

Consumers were increasingly using AI to:

• research products;
• understand features;
• compare alternatives;
• and identify retailers

but were still less likely to complete a transaction immediately after the AI referral.

The behaviour is consistent with AI operating primarily within the research and consideration stage.

AI traffic implication: Early AI referral performance should not be judged only by immediate transactions.

A user may conduct research through AI and later return through:

• Google;
• direct navigation;
• email;
• or another referral source.

That makes multi-touch attribution particularly important when measuring AI Search.

Source:

Adobe Digital Insights — Generative AI-Powered Shopping

52. The AI retail conversion gap narrowed to 38% by April 2025

US Ecommerce Benchmark. Adobe reported that by April 2025, generative AI referral traffic was 38% less likely to convert than non-AI traffic.

This represented a material improvement from the 49% gap measured in January.

Within only three months, AI-referred traffic had moved considerably closer to conventional digital channels in commercial performance.

This suggests that both:

• consumer familiarity with AI shopping;
• and the relevance of AI-generated recommendations

were improving.

AI traffic implication: AI referral benchmarks can change quickly.

Businesses should therefore avoid setting permanent expectations based on one early-stage conversion measurement.

Source:

Adobe Digital Insights

53. AI-referred shoppers were only 23% less likely to convert by July 2025

US Ecommerce Benchmark. Adobe found that by July 2025, generative AI traffic was only 23% less likely to convert than non-AI traffic.

The progression during the year was:

• January 2025 — 49% lower conversion
• April 2025 — 38% lower
• July 2025 — 23% lower

The gap therefore narrowed by more than half between January and July.

This indicates that consumers were becoming increasingly comfortable progressing from AI-assisted Product Research directly into a transaction.

AI traffic implication: AI referral traffic should be monitored as a developing acquisition channel rather than permanently classified as upper-funnel research traffic.

The commercial behaviour of AI users is changing.

Source:

Adobe — AI-Powered Shopping Research

54. By August 2025, AI referral traffic was only 9% less likely to convert

US Ecommerce Benchmark. Adobe reported that in August 2025, shoppers arriving from generative AI sources were only 9% less likely to convert than visitors from non-AI channels.

The remaining conversion gap had therefore become relatively small.

The progression from January to August was substantial:

49% Behind → 38% Behind → 23% Behind → 9% Behind

within less than eight months.

AI traffic implication: Rapid improvement in conversion quality means historic AI-referral benchmarks can become obsolete within a matter of months.

Businesses should therefore use their own current analytics rather than relying exclusively on industry data from the previous year.

Source:

Adobe Digital Insights — October 2025 Retail Data

55. AI referral traffic converted 5% better than non-AI traffic in September 2025

US Ecommerce Benchmark. Adobe reported that September 2025 marked a significant turning point.

For the first time in its retail tracking, generative AI referral traffic converted 5% better than non-AI traffic.

This represented a reversal of the performance pattern seen earlier in the year.

AI traffic had moved from:

lower conversion

to:

higher conversion.

The result suggests that by late 2025, users arriving after AI-assisted research could be further advanced within the purchasing process.

AI traffic implication: AI referrals should no longer automatically be categorised as low-conversion informational traffic.

Their commercial value needs to be measured directly.

Source:

Adobe Digital Insights

56. AI-referred shoppers converted 16% better than other traffic sources in October 2025

US Ecommerce Benchmark. Adobe reported that by October 2025, generative AI visitors were 16% more likely to convert than visitors from other digital traffic sources.

This was the second consecutive month in which AI referral conversion outperformed non-AI traffic.

The direction of travel during 2025 was therefore remarkable:

• January — AI substantially behind;
• April — gap narrowing;
• July — gap narrowing further;
• August — near parity;
• September — AI ahead;
• October — AI further ahead.

AI traffic implication: Conversational discovery may increasingly create a form of pre-qualified traffic.

The user can complete substantial research before clicking, potentially producing a shorter final journey from referral to purchase.

Source:

Adobe — October 2025 Shopping Analysis

57. AI-referred visits generated 8% more revenue per session in October 2025

US Ecommerce Benchmark. Adobe reported that the stronger conversion performance in October 2025 translated directly into commercial value.

Visits originating from generative AI sources generated 8% more revenue per session than non-AI traffic.

This is especially important because traffic quality should ultimately be evaluated through economic outcomes rather than engagement alone.

Revenue per visit combines several commercial factors:

• conversion probability;
• order value;
• product mix;
• and transaction completion.

AI traffic implication: Revenue per AI-referred visitor may be a more meaningful KPI than AI traffic volume itself.

A smaller acquisition channel can become commercially significant if each visit produces greater economic value.

The relevant equation is:

AI Sessions × Revenue per Visit = AI-Referred Revenue

Source:

Adobe Digital Insights

58. AI referral traffic converted 31% better than other traffic sources during the 2025 holiday season

US Ecommerce Benchmark. Adobe’s final 2025 holiday-shopping analysis found that shoppers arriving from generative AI sources converted 31% more frequently than visitors from other traffic sources.

The measurement covered the period from:

1 November to 31 December 2025.

Adobe’s comparison included established digital channels such as:

• organic search;
• paid search;
• email;
• social;
• affiliates;
• and other referral traffic.

This represents one of the clearest indications that AI referral traffic had moved beyond experimental commercial relevance.

AI traffic implication: AI Search can now generate visitors whose immediate purchase propensity exceeds that of established digital traffic sources in some ecommerce datasets.

This makes AI referral conversion a meaningful acquisition KPI rather than an experimental metric.

Source:

Adobe Digital Insights — 2025 Holiday AI Traffic

59. AI-referred shoppers converted 54% better than other traffic on Thanksgiving 2025

US Ecommerce Benchmark. Adobe found that the conversion advantage became even stronger during important shopping events.

On Thanksgiving Day 2025, AI-referred retail traffic converted 54% better than non-AI traffic.

This is notable because high-intent shopping periods can involve significant:

• Product Comparison;
• offer research;
• deal discovery;
• and purchase urgency.

AI assistants are particularly suited to helping consumers navigate this complexity.

AI traffic implication: The commercial value of AI referrals may vary substantially according to:

• season;
• shopping event;
• product category;
• and purchase urgency.

Businesses should therefore avoid evaluating AI conversion only through annual averages.

Source:

Adobe Digital Insights

60. AI-referred shoppers converted 38% better than other traffic on Black Friday 2025

US Ecommerce Benchmark. On Black Friday 2025, Adobe reported that shoppers arriving through generative AI referrals converted 38% more frequently than visitors from other sources.

The result reinforces the Thanksgiving data and demonstrates that AI-referred users remained commercially strong during one of the most competitive online-shopping periods of the year.

Adobe also reported that AI-driven revenue per visit had improved dramatically during 2025.

From January to July alone, relative AI revenue per visit increased by 84%.

By the 2025 holiday period, Adobe reported a further substantial improvement in AI-driven revenue per visit.

The evidence therefore indicates that both:

conversion quality

and:

economic value per visit

were strengthening during the year.

AI traffic implication: AI Search Traffic should increasingly be evaluated through a complete commercial funnel:

AI Referral
→ Engagement
→ Conversion
→ Order Value
→ Revenue per Visit
→ Customer Value

This creates a much more meaningful measurement framework than referral sessions alone.

Source:

Adobe Digital Insights — AI-Driven Traffic Surges Across Industries

Section 7 — AI Referral Traffic by Industry & Sector Statistics

AI referral traffic is not developing at the same speed across every industry.

Some sectors already receive tens of millions of monthly visits from generative AI platforms, while others remain comparatively small but are growing extremely quickly.

This creates two different measurements:

Current AI Referral Volume

and:

AI Referral Growth.

The distinction matters because a sector with the highest percentage growth is not necessarily the sector receiving the most actual visitors.

For businesses evaluating AI Search Traffic, both measures are important.

Volume helps determine whether AI is commercially material today.

Growth helps identify where AI discovery may become commercially material next.

Data classification: Statistics 61–69 use Similarweb’s 2026 Generative AI Landscape research covering worldwide desktop and mobile web referrals between June 2025 and May 2026. Statistic 70 uses Adobe Digital Insights sector research. These are international benchmarks and should not be interpreted as UK-only traffic measurements.

61. AI platforms generated 770.7 million referral visits per month worldwide

Global AI Referral Benchmark. Similarweb’s 2026 Generative AI Landscape research estimated that AI platforms generated an average of 770.7 million referral visits to websites every month between June 2025 and May 2026.

That represented a year-on-year increase of 117.4%.

The measurement includes referral traffic from platforms such as:

• ChatGPT;
• Gemini;
• Claude;
• Perplexity;
• Grok;
• DeepSeek;
• and other generative AI services.

This provides useful context for the overall scale of AI referral traffic.

The channel remains considerably smaller than conventional search at global level, but hundreds of millions of external website visits are now being generated every month.

AI traffic implication: AI referrals have moved beyond experimental analytics noise and are becoming a distinct global traffic channel.

Businesses should begin reporting:

• AI referral share;
• AI referral growth;
• platform mix;
• and competitive AI traffic share.

Source:

Similarweb — AI Referral Traffic by Industry: 2026 Data

62. Marketplaces received 46.8 million AI referral visits per month

Global Industry Benchmark. Similarweb found that marketplaces were the largest sector for AI referral traffic, receiving an average of 46.8 million visits per month between June 2025 and May 2026.

The category includes major multi-product platforms such as:

• Amazon;
• Temu;
• and other large online marketplaces.

Marketplaces did not simply lead in volume.

AI referral traffic to the category also grew by 237.3% year on year.

Average monthly visits increased from approximately:

• 13.9 million in the previous annual period
• to 46.8 million.

AI traffic implication: AI shopping answers often need to send users somewhere to compare or purchase products.

Large marketplaces can benefit because they provide:

• broad inventories;
• multiple brands;
• product reviews;
• pricing;
• and transaction capability.

Retailers should therefore consider both their own AI visibility and their visibility within marketplaces AI systems frequently recommend.

Source:

Similarweb — Generative AI Landscape 2026

63. News websites received approximately 44.5 million AI referral visits per month

Global Industry Benchmark. Similarweb reported that news websites received approximately 44.5 million monthly AI referral visits.

News was effectively tied with travel as the second-largest sector measured.

AI referral traffic to news increased by approximately 115.6% year on year.

Large publishers are frequently used by AI systems because they provide:

• current information;
• breaking news;
• reporting;
• expert commentary;
• and primary-source journalism.

AI traffic implication: Publishing organisations face a dual AI effect.

Their material can be heavily used in generated answers, potentially reducing some conventional search clicks, while AI platforms can also create substantial referral traffic when users require the full article or source.

The commercial impact therefore depends on the balance between:

AI Content Consumption vs AI Referral Traffic.

Source:

Similarweb — AI Referral Traffic by Industry

64. Travel websites received approximately 44.5 million AI referral visits per month

Global Industry Benchmark. Travel also received approximately 44.5 million average monthly AI referral visits in Similarweb’s 2026 dataset.

The sector grew by approximately 115.6% year on year.

Travel is especially well suited to conversational AI because customer journeys frequently involve several connected questions:

• Where should I go?
• When should I travel?
• Which hotel is suitable?
• What flights are available?
• What activities should I book?
• How much will the trip cost?

An AI assistant can perform much of that research within one conversation before referring the user to an external website.

AI traffic implication: Travel AI referrals are already sufficiently large to deserve dedicated monthly reporting for airlines, hotel groups, booking platforms and destination organisations.

Source:

Similarweb — 2026 Generative AI Landscape

65. Finance received 19.1 million average monthly AI referral visits

Global Industry Benchmark. Similarweb reported that finance websites received approximately 19.1 million AI referral visits per month.

The sector recorded year-on-year growth of approximately 236.5%.

This means finance more than tripled its AI referral volume over the measured annual period.

Financial queries can involve complex decisions relating to:

• investments;
• insurance;
• banking;
• loans;
• mortgages;
• credit;
• and financial products.

AI systems can help simplify these subjects before directing users to specialist or provider websites.

AI traffic implication: AI referrals are becoming increasingly relevant within high-value and high-trust sectors, not merely retail or informational publishing.

Financial organisations should monitor both referral growth and the accuracy of the information AI systems present about their products.

Source:

Similarweb — AI Referral Traffic by Industry 2026

66. Consumer electronics received 17.1 million average monthly AI referral visits

Global Industry Benchmark. Similarweb reported that consumer-electronics websites received approximately 17.1 million monthly AI referral visits.

The category grew by 215.4% year on year.

Consumer electronics are particularly suited to AI-assisted discovery because purchases often involve:

• technical specifications;
• compatibility;
• features;
• price comparison;
• model comparison;
• and product reviews.

Adobe’s separate retail research also found that AI referral share for consumer electronics was four times the share recorded for apparel and footwear during May 2025.

AI traffic implication: AI adoption appears particularly strong for considered purchases where users benefit from structured comparison before clicking.

Businesses selling complex products should ensure Product Information is:

• complete;
• current;
• machine-readable;
• and easy to compare.

Source:

Similarweb — Generative AI Landscape 2026

67. Beauty was the fastest-growing AI referral sector at 312.5% year on year

Global Industry Benchmark. Similarweb identified beauty as the fastest-growing of the seven sectors analysed, with AI referral traffic increasing by 312.5% year on year.

Average monthly AI referrals increased from approximately:

• 473,000 visits
• to approximately 2.0 million visits.

Monthly traffic reached approximately 4.0 million visits in May 2026, compared with around 1.0 million in May 2025.

Conversational Product Discovery is particularly relevant to beauty because consumers ask detailed questions such as:

• Which product is appropriate for sensitive skin?
• Which ingredients should I avoid?
• Which products work together?
• What is suitable for a particular skin type?

AI traffic implication: Smaller AI referral categories can become commercially meaningful very quickly when conversational discovery aligns strongly with consumer decision behaviour.

Source:

Similarweb — AI Referral Industry Growth 2026

68. Fashion AI referral traffic grew 278.2% year on year

Global Industry Benchmark. Similarweb reported that fashion AI referral traffic grew by 278.2% year on year.

Fashion therefore ranked immediately behind beauty among the fastest-growing categories in its sector analysis.

Similarweb reported monthly Fashion AI referrals rising from approximately:

• 4.4 million visits in May 2025
• to 6.7 million by October 2025.

AI-assisted Fashion Discovery can involve questions about:

• style;
• fit;
• budget;
• occasion;
• colour;
• brand;
• and product combinations.

These are often more naturally expressed conversationally than through conventional short keyword searches.

AI traffic implication: Fashion retailers should increasingly treat AI as a Product Discovery environment rather than simply another source of general website traffic.

Source:

Similarweb — Generative AI Referral Data

69. Travel queries had a 22.6% ChatGPT web-citation rate in May 2026

US ChatGPT Citation Benchmark. Similarweb’s Generative AI Landscape research found that 22.6% of ChatGPT answers relating to travel included web citations in May 2026.

The average across all ChatGPT conversations measured was only 6.8%.

Travel therefore had more than three times the average web-citation rate.

Similarweb also reported that retail-related ChatGPT answers produced web citations at a rate of approximately 13.5%.

This matters because external referral traffic can only occur when the AI system provides a path to an external source.

Industries where AI systems search and cite the live web more frequently may therefore have structurally greater referral potential.

AI traffic implication: Referral opportunity depends not only on user adoption but also on how frequently AI systems use external sources for a particular industry.

This creates the relationship:

AI Usage × Web Retrieval Frequency × Citation Selection × User Click = AI Referral Traffic

Source:

Similarweb — AI Referral Traffic by Industry

70. Technology and software had more than 10 times the AI referral share of retail and banking in Adobe’s sector comparison

US Industry Benchmark. Adobe Digital Insights found that technology and software had the highest AI referral visit share among the major sectors it analysed in May 2025.

Technology and software’s AI referral share was:

• more than 2 times that of media and entertainment;
• 5 times that of travel;
• more than 10 times that of retail and banking.

Technology AI referral traffic had also increased by approximately 13 times between July 2024 and May 2025.

From February to May 2025 alone, it increased by 74%.

Technology users may be particularly likely to use AI systems because the buying journey often involves:

• technical research;
• software comparisons;
• integration questions;
• troubleshooting;
• and complex product evaluation.

AI traffic implication: AI Search penetration varies substantially by industry.

Businesses should therefore avoid applying a single universal AI Traffic target across sectors.

The relevant benchmark is the organisation’s own competitive market.

Source:

Adobe Digital Insights — AI-Driven Traffic Surges Ahead in Q2

Section 8 — AI Attribution, Branded Search & Customer Journey Statistics

One of the most important problems in AI Search measurement is attribution.

A customer can discover a business through ChatGPT, leave without clicking anything, search for the brand on Google several days later and then visit the website.

Conventional analytics will normally record that visit as:

Organic Search

rather than:

AI-Influenced Discovery.

This creates a potentially large gap between the amount of traffic AI systems directly refer and the amount of website activity they actually influence.

The customer journey is increasingly becoming:

AI Discovery
→ Brand Recognition
→ Further Research
→ Branded Search / Direct Visit
→ Website
→ Conversion

rather than a single attributable click.

Data classification: Statistics 71–78 use US and international behavioural data from Similarweb and Semrush. Statistics 79–80 use UK consumer research reported by Adobe. These datasets measure different parts of the customer journey and should not be combined into a single market-share estimate.

71. Users were 2.5 times more likely to visit a brand after ChatGPT recommended it

US Downstream Behaviour Benchmark. Similarweb tracked thousands of real user journeys across:

• Finance;
• Travel;
• and Beauty.

The study followed users who asked ChatGPT an industry-related question, received a specific brand recommendation and did not immediately click through to the brand.

Similarweb then observed what those users did over the following seven days.

Users who received an AI recommendation were 2.5 times more likely to visit the recommended brand’s website than users who were shown a competing brand instead.

The study excluded users who had already visited the brand or included the brand within their original prompt, helping isolate new downstream demand.

AI traffic implication: An AI recommendation can create measurable website traffic even when no direct AI referral occurs.

This means the correct attribution model is not simply:

AI Click → Website

but also:

AI Recommendation → Later Brand Search → Website Visit

Source:

Similarweb — The Downstream Impact of AI Visibility

72. 55.9% of AI-influenced website traffic arrived through search

US Downstream Behaviour Benchmark. Similarweb found that 55.9% of AI-influenced website traffic arrived through a search engine.

In many cases, the AI platform had created the initial awareness or recommendation, but the user later searched for the organisation by name.

The website’s analytics could therefore record:

Google / Organic

even though the earlier ChatGPT interaction helped create the demand.

This is one of the clearest pieces of evidence that direct AI referrals can understate AI’s wider commercial influence.

AI traffic implication: Organisations measuring AI Search should monitor:

• AI referral traffic;
• branded organic search;
• brand-name impressions;
• direct traffic;
• and Brand Demand

together rather than as completely independent channels.

Source:

Similarweb — The Downstream Impact of AI Visibility

73. Search accounted for only around 40% of standard visits in the same Similarweb analysis

US Comparative Benchmark. Similarweb reported that search accounted for approximately 40% of ordinary website visits within the comparison group.

For AI-influenced visitors, that figure was closer to 56%.

The comparison therefore indicates that people exposed to an AI recommendation were disproportionately likely to use search afterwards.

This supports an important customer-journey pattern:

AI Generates Awareness → Google Captures Navigation

The search engine may therefore receive attribution for demand created somewhere else.

AI traffic implication: Growth in branded organic traffic may become a useful supporting indicator of AI influence.

It should not automatically be attributed to AI, but it should be examined alongside changes in:

• AI Recommendation Visibility;
• AI citations;
• Brand Mentions;
• and AI referral traffic.

Source:

Similarweb — AI Visibility Downstream Impact Study

74. AI-influenced visitors showed approximately 2 times deeper engagement

US Downstream Behaviour Benchmark. Similarweb reports that visitors whose journeys had been influenced by an AI recommendation showed approximately 2 times deeper engagement after reaching the website.

The finding supports the idea that AI is frequently performing part of the user’s research before the website visit takes place.

By the time the user arrives, they may already understand:

• what the brand offers;
• why it was recommended;
• how it compares with competitors;
• and whether it fits their requirement.

This creates a different visitor from someone encountering the organisation for the first time through a generic keyword.

AI traffic implication: AI-influenced traffic should be assessed not merely by the acquisition channel recorded in analytics but by downstream visitor behaviour.

The visitor may appear as organic or direct traffic while behaving more like a pre-qualified referral.

Source:

Similarweb — Downstream Impact of AI Visibility

75. 21.6% of ChatGPT’s outbound referral traffic went to Google by February 2026

US Clickstream Benchmark. Semrush found that Google accounted for 21.6% of all outbound referral traffic from ChatGPT by February 2026.

Google was therefore the single largest external destination for ChatGPT referrals within the study.

This is strategically important because it demonstrates that ChatGPT and Google are not necessarily functioning as mutually exclusive alternatives.

Users can move between them during the same research journey.

A possible sequence is:

ChatGPT Research
→ Google Search
→ Brand or Product Result
→ Website

AI traffic implication: AI Search Visibility and conventional SEO should increasingly be treated as interconnected.

Strong AI visibility can create additional Google activity, while strong Google visibility helps capture users who continue their AI-assisted research through conventional search.

Source:

Semrush — ChatGPT Traffic Analysis

76. More than 30% of ChatGPT referral traffic went to only 10 domains

US Clickstream Benchmark. Semrush found that the ten largest destination domains received just over 30% of all ChatGPT referral traffic by February 2026.

Google alone accounted for 21.6%, while the next nine domains combined accounted for approximately 8.6%.

Semrush reported that the share captured by the top ten domains had ranged between approximately 20% and 32% throughout the study period.

This indicates that AI referral traffic remains relatively concentrated.

The existence of hundreds of thousands of domains receiving ChatGPT referrals does not mean traffic is distributed equally between them.

AI traffic implication: Simply becoming technically eligible for an AI citation is not sufficient.

Businesses also need enough:

• authority;
• relevance;
• Brand Recognition;
• source credibility;
• and topic coverage

to compete for meaningful recommendation and referral share.

Source:

Semrush — 17 Months of ChatGPT Clickstream Data

77. ChatGPT used web search on only 34.5% of queries in February 2026

US Clickstream Benchmark. Semrush found that ChatGPT activated its web-search functionality on approximately 34.5% of queries in February 2026.

That was lower than the approximately 46% recorded in late 2024.

This statistic is crucial for understanding AI attribution because not every ChatGPT interaction involves retrieval from the live web.

Many responses can be generated primarily from the model’s existing knowledge and context.

A business may therefore influence AI answers through:

• its broader digital footprint;
• historic web content;
• third-party coverage;
• entity information;
• and previously acquired authority

without receiving a visible citation every time.

AI traffic implication: AI Search optimisation cannot be reduced to winning live citations alone.

It also involves creating an authoritative digital presence that AI systems can reliably associate with relevant subjects.

Source:

Semrush — ChatGPT Search Insights

78. Average ChatGPT queries per session increased 50% during the final four months of Semrush’s study

US Clickstream Benchmark. Semrush reported that after approximately twelve months of relatively flat engagement, the average number of ChatGPT queries per session increased by 50% during the final four months of its 17-month study.

This reflects the increasingly conversational nature of AI discovery.

A traditional search journey might involve several separate Google searches.

An AI journey can instead involve a sequence such as:

Initial Question
→ Follow-Up
→ Comparison
→ Clarification
→ Recommendation
→ Provider Research

within one continuous conversation.

This means the AI platform can influence a much larger proportion of the decision process before an external visit occurs.

AI traffic implication: Prompt-level measurement should increasingly consider topic journeys rather than isolated queries.

A brand that appears in one answer but disappears from subsequent comparison or recommendation prompts may have substantially weaker commercial visibility than a brand that remains present throughout the conversation.

Source:

Semrush — ChatGPT Traffic Analysis 2026

79. 35% of UK consumers had already used generative AI to assist with online shopping

UK Consumer Data. Adobe reported that 35% of UK consumers surveyed in March 2025 had already used generative AI tools to assist with online shopping.

AI-assisted shopping can involve:

• Product Research;
• Product Comparison;
• gift ideas;
• shopping lists;
• deal discovery;
• and Brand Research.

This means a substantial proportion of UK shoppers were already using AI somewhere within the ecommerce customer journey.

Importantly, that does not mean every AI-assisted shopper generates an identifiable AI referral.

Many may later reach the retailer through:

• Google;
• direct navigation;
• marketplaces;
• or another digital channel.

AI traffic implication: The percentage of UK shoppers influenced by AI may be materially larger than the proportion appearing under AI Referral Traffic in analytics.

Source:

Adobe Digital Insights UK — Retail in Flux

80. 47% of UK consumers planned to use generative AI for shopping during 2025

UK Consumer Data. Adobe’s UK research found that 47% of consumers planned to use generative AI tools to help with shopping during 2025.

That was substantially higher than the 35% who said they had already done so at the time of the survey.

The gap indicated considerable potential for continued adoption.

Adobe subsequently reported that UK AI-driven retail referrals increased by almost 1,200% between August 2024 and June 2025.

The consumer-intention data and measured referral growth therefore point in the same direction:

AI is becoming an increasingly normal part of UK shopping research.

AI traffic implication: Retailers should not rely solely on direct referral data when evaluating the channel.

A more complete UK measurement model should consider:

AI Usage
+ AI Visibility
+ AI Referrals
+ Branded Search
+ Direct Traffic
+ Conversion

Source:

Adobe Digital Insights UK — Generative AI Shopping Research

Section 9 — AI Citations, Brand Mentions & Source Selection Statistics

AI Search Visibility is not a single metric.

A website can be used as a source without the organisation being named.

A brand can be mentioned without receiving a link.

A company can also be recommended even when its own website is not the source used to construct the answer.

These distinctions create three separate forms of AI visibility:

• AI Citation — the website or page is used as a source;
• Brand Mention — the organisation is named within the answer;
• AI Recommendation — the organisation is presented as an option, provider, product or solution.

The commercial value of each can be different.

A citation may build source authority.

A Brand Mention may create awareness.

A recommendation may influence a purchasing decision.

The following statistics examine how frequently these outcomes overlap, which sources AI systems rely upon and how different platforms select information.

Data classification: Statistics 81–85 use 2026 multi-platform research from Semrush. Statistics 86–88 use Yext’s analysis of 6.8 million AI citations across ChatGPT, Gemini and Perplexity. Statistic 89 uses Ahrefs’ March 2026 analysis of four million Google AI Overview URLs. Statistic 90 uses Ahrefs cross-platform source research.

81. 62% of AI citations did not produce an explicit Brand Mention

International AI Visibility Benchmark. Semrush analysed 3,981 domain appearances generated from 115 prompts across 14 countries and four major AI Search environments.

The platforms were:

• ChatGPT;
• Google AI Overviews;
• Gemini;
• and Google AI Mode.

The study found that 62% of AI citations were “ghost citations”.

In other words, the AI system used the website as a source but did not explicitly name the associated brand within the answer.

This is a major distinction for AI Traffic measurement.

A source can contribute evidence to an answer without the user clearly recognising which organisation supplied that evidence.

AI traffic implication: Citation count alone can overstate the amount of visible Brand Exposure being created.

Businesses should separately monitor:

• citations;
• Brand Mentions;
• recommendations;
• and linked citations.

Source: Semrush — Why 62% of AI Citations Don’t Lead to Brand Mentions, June 2026.

82. Gemini named the brand in 83.7% of relevant appearances but cited it only 21.4% of the time

International Platform Benchmark. Semrush found a substantial difference between brand visibility and website citation within Gemini.

When a brand appeared in a Gemini answer, the brand was explicitly named in the response 83.7% of the time.

However, the brand’s domain was cited as a source only 21.4% of the time.

This means Gemini could create considerable brand exposure without providing a direct path to the organisation’s website.

The behaviour demonstrates why AI Search cannot be measured using referral traffic alone.

A user can see and remember a brand even where no clickable citation exists.

AI traffic implication: Gemini visibility reporting should include:

• Brand Mentions;
• recommendation frequency;
• citation frequency;
• and referral traffic

as distinct measures.

Source: Semrush — Ghost Citations Study, 2026.

83. ChatGPT cited brands in 87% of relevant appearances but named them in only 20.7%

International Platform Benchmark. Semrush found almost the opposite pattern within ChatGPT.

When a studied domain appeared in a ChatGPT response, it was cited as a source approximately 87% of the time.

Yet the associated brand was explicitly mentioned in the text only 20.7% of the time.

This demonstrates how dramatically citation behaviour can vary between AI platforms.

The same organisation may therefore experience:

High Citation Visibility + Low Brand Visibility

on one platform, while experiencing:

High Brand Visibility + Low Citation Visibility

on another.

AI traffic implication: Cross-platform AI performance should not be represented by one combined “visibility score” without understanding what that score actually measures.

Businesses should distinguish between:

• being used as evidence;
• being named;
• being linked;
• and being recommended.

Source: Semrush — AI Citation and Brand Mention Study, June 2026.

84. Short conversational queries generated 30 to 50 times more Brand Mentions than long prompts

International Query Benchmark. Semrush found that short, conversational AI queries generated approximately 30 to 50 times more Brand Mentions than longer prompts within its study.

Longer prompts tended to generate more source citations, but substantially fewer explicit brand references.

This distinction is commercially important.

Different prompt types can produce different kinds of visibility.

A detailed research query may encourage the model to assemble evidence from multiple sources.

A shorter commercial query may be more likely to produce a direct list of:

• brands;
• products;
• providers;
• or recommended options.

AI traffic implication: AI Search monitoring should include multiple stages of the customer journey rather than only long informational prompts.

A useful prompt set can include:

• information queries;
• comparison queries;
• recommendation queries;
• provider-selection queries;
• and Brand Research queries.

Source: Semrush — Ghost Citations Study, 2026.

85. Comparative AI content produced 2.4 times more Brand Mentions

International Query-Intent Benchmark. Semrush found that comparative content and comparison-oriented responses generated approximately 2.4 times more Brand Mentions than general informational content.

This makes intuitive sense.

A purely informational question may require factual sources without requiring the AI system to name businesses.

A comparison question explicitly asks the system to distinguish between:

• brands;
• providers;
• products;
• services;
• or alternative solutions.

The probability of explicit Brand Visibility therefore increases.

AI traffic implication: Organisations seeking commercial AI visibility should monitor more than educational prompts.

Queries such as:

• best providers for…
• which company offers…
• A vs B;
• alternatives to…
• recommended suppliers for…

may reveal a substantially different competitive picture.

Source: Semrush — AI Citation and Mention Research, June 2026.

86. 86% of citations in Yext’s AI study came from sources brands could directly manage or influence

Global AI Citation Benchmark. Yext analysed 6.8 million AI citations generated across ChatGPT, Gemini and Perplexity between July and August 2025.

The study covered approximately 1.6 million questions per model across:

• Financial Services;
• Retail;
• Healthcare;
• and Food Service.

Yext found that approximately 86% of citations came from sources that brands could directly manage or meaningfully influence.

These included:

• first-party websites;
• local pages;
• business listings;
• directories;
• reviews;
• and social platforms.

Only a relatively small proportion came from sources over which the organisation had no direct influence.

AI traffic implication: AI source selection is not completely outside organisational control.

Businesses can potentially improve Citation Eligibility through better:

• first-party information;
• listings;
• entity consistency;
• reviews;
• and Digital Reputation.

Source: Yext Research — AI Citations, User Locations & Query Context, 2025.

87. Gemini drew 52.15% of citations from first-party websites

Global Platform Benchmark. Within Yext’s 6.8-million-citation dataset, Gemini showed the strongest preference for brand-owned websites.

Approximately 52.15% of Gemini citations came from first-party web properties.

These included:

• corporate websites;
• local business pages;
• product pages;
• service pages;
• and other organisation-controlled digital assets.

Across the full study, first-party corporate and local websites generated more than 2.9 million citations.

This provides strong evidence that an organisation’s own site remains important within generative discovery.

AI traffic implication: GEO should not be interpreted as replacing website optimisation.

The website remains a primary evidence source that AI systems can use to understand:

• what the organisation is;
• what it offers;
• where it operates;
• and which information is authoritative.

Source: Yext Research — AI Citation Source Analysis.

88. 48.73% of OpenAI citations in Yext’s study came from business listings

Global Platform Benchmark. Yext found a markedly different source pattern for OpenAI.

Approximately 48.73% of OpenAI citations came from controllable third-party listings.

These can include:

• Google-related listings;
• directories;
• industry platforms;
• location databases;
• and other structured third-party profiles.

Yext reported more than 2.9 million citations from listing-type sources across the full study.

This reinforces the importance of entity consistency beyond the organisation’s own domain.

AI traffic implication: AI visibility can depend on whether an organisation is represented consistently across the wider digital ecosystem.

The relevant architecture becomes:

Website
+ Listings
+ Directories
+ Reviews
+ Third-Party References
= AI Evidence Ecosystem

Source: Yext Research — 6.8 Million AI Citation Study.

89. Only 37.9% of AI Overview citations appeared within the first 10 Google result blocks

Google AI Overview Benchmark. Ahrefs updated its AI Overview citation research in March 2026 using:

• 863,000 keyword SERPs;
• and approximately 4 million AI Overview URLs.

It found that only 37.9% of URLs cited in AI Overviews also appeared within the first ten Google search-result blocks for the same query.

The remaining citations were approximately split between:

• 31.2% appearing between positions 11 and 100;
• 31.0% appearing beyond the first 100 result blocks.

When Ahrefs analysed traditional organic blue links only, approximately 37.1% of cited URLs ranked organically within the top ten.

This is significant because an earlier 2025 study had found a much larger overlap between top-ranking pages and AI Overview citations.

Ahrefs cautions that its methodology also improved between studies, so the change should not be attributed entirely to Google behaviour.

AI traffic implication: Ranking on page one remains valuable, but it does not guarantee AI citation visibility.

AI systems can retrieve evidence from a much broader source set through techniques such as query fan-out.

The optimisation target therefore becomes:

Topical Relevance Across the Wider Research Journey

rather than one exact keyword alone.

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

90. Only 14% of the top sources were shared across ChatGPT, Perplexity and Google AI

Cross-Platform AI Benchmark. Ahrefs compared the top 50 most-mentioned sources across:

• ChatGPT;
• Perplexity;
• and Google AI Overviews.

It found that only 14% of those top sources were shared across all three AI environments.

In other words:

86% were not common to all three platforms.

The study was based on large Ahrefs Brand Radar datasets covering millions of AI-generated results and prompts.

The finding demonstrates that there is no single universal AI source-selection system.

Each platform can:

• retrieve different sources;
• apply different ranking or retrieval systems;
• use different data partnerships;
• and generate different recommendations.

AI traffic implication: Strong visibility in one AI platform does not guarantee visibility elsewhere.

Businesses therefore need cross-platform measurement covering:

• ChatGPT;
• Gemini;
• Google AI Overviews;
• AI Mode;
• Perplexity;
• Copilot;
• and other relevant systems.

Source: Ahrefs — Top Mentioned Sources Across AI Assistants, 2025.

Section 10 — AI Search Measurement, Marketing Adoption & Future Traffic Statistics

The final ten statistics examine how organisations are responding to the changes documented throughout this research.

AI Search has moved sufficiently quickly that many marketing teams now recognise the strategic shift but have not yet developed the measurement systems, workflows or reporting models required to manage it effectively.

This creates an important gap between:

AI Search Awareness

and:

AI Search Operational Readiness.

The evidence suggests that most organisations now understand that search extends beyond conventional Google rankings.

The challenge is connecting:

• AI visibility;
• AI citations;
• AI referrals;
• branded demand;
• leads;
• transactions;
• and revenue.

Data classification: Statistics 91–93 use Salesforce’s 2026 State of Marketing research, including UK-specific findings from 250 UK marketers within a global survey of 4,450 marketing decision-makers. Statistics 94–100 use Semrush’s June 2026 survey of 481 marketers, business owners and SEO professionals.

91. 84% of UK marketers say AI is reshaping their SEO strategy

UK Marketing Data. Salesforce’s 2026 State of Marketing research found that 84% of UK marketers say AI is reshaping their SEO strategy.

This indicates that AI Search is no longer being treated primarily as an experimental technology issue.

It is increasingly influencing core Search Strategy.

Traditional SEO historically concentrated on:

• rankings;
• organic traffic;
• links;
• technical performance;
• and conversions.

AI Search adds new visibility outcomes such as:

• AI citations;
• Brand Mentions;
• AI recommendations;
• source selection;
• and AI-generated referrals.

AI traffic implication: Organisations increasingly need one Search Strategy that considers both conventional rankings and generative discovery.

The emerging model is:

SEO + AI Search + GEO + Brand Authority

rather than treating each as an entirely separate activity.

Source:

Salesforce UK — State of Marketing 2026

92. 90% of UK marketers have begun optimising for AI-generated responses

UK Marketing Data. Salesforce reported that 90% of UK marketers had already begun optimising for AI-generated responses across environments such as ChatGPT and Google’s AI experiences.

This is a remarkably high level of stated adoption.

It suggests that AI Search optimisation has moved rapidly from:

emerging experiment

toward:

mainstream marketing activity.

However, beginning optimisation does not necessarily mean organisations have established:

• a formal strategy;
• dedicated measurement;
• repeatable workflows;
• or reliable commercial attribution.

Later statistics in this section show that significant operational gaps remain.

AI traffic implication: Competitive pressure around AI Search Visibility is likely to increase considerably as more organisations actively optimise for generative discovery.

The opportunity is therefore changing from:

“Should we optimise for AI Search?”

to:

“How effectively are we optimising compared with competitors?”

Source:

Salesforce UK — Tenth Edition State of Marketing

93. High-performing marketers are 2.2 times more likely to have optimised for AI Search

International Marketing Benchmark. Salesforce found that high-performing marketing teams were 2.2 times more likely than underperforming teams to have optimised for AI Search.

Salesforce classified high performers according to their reported satisfaction with the overall outcomes of their marketing investments.

This is an association rather than proof that AI optimisation itself caused stronger marketing performance.

Nevertheless, it indicates that organisations reporting better marketing outcomes are more likely to have incorporated AI Search into their strategy.

AI traffic implication: AI Search optimisation is increasingly becoming part of the operating model used by more advanced marketing teams.

That can involve:

• SEO;
• content;
• Digital PR;
• brand;
• entity optimisation;
• analytics;
• and AI visibility measurement.

Source:

Salesforce — State of Marketing 2026

94. Only 22% of marketers have fully integrated SEO and AI Search

International Marketing Benchmark. Semrush surveyed 481 marketers, SEO professionals and business owners in 2026 and found that only 22% had fully integrated SEO and AI Search across strategy, execution and reporting.

The remaining 78% still had some degree of separation between the two disciplines.

Semrush found several common operating models:

• planning SEO and AI together but executing separately;
• parallel workflows with some coordination;
• separate tools;
• or limited overlap between teams.

This reveals a substantial difference between recognising the AI Search transition and actually changing organisational processes.

AI traffic implication: Search teams increasingly need a unified model covering:

Google Search + AI Search + Content + Brand + Authority + Measurement

rather than independent workflows that produce fragmented reporting.

Source:

Semrush — The Operational Gap Between AI Search and SEO

95. 49% of marketers struggle to measure AI’s impact on pipeline or revenue

International Measurement Benchmark. Semrush found that 49% of marketers identified measuring the impact of AI Search on pipeline or revenue as one of their major measurement challenges.

This is one of the most important findings for AI Traffic analysis.

The industry is becoming increasingly capable of measuring:

• AI mentions;
• citations;
• visibility;
• and direct referrals.

Connecting those indicators to commercial outcomes remains much harder.

The attribution problem described in Section 8 contributes directly to this difficulty.

A customer can discover a business through AI but convert later through:

• Google Organic;
• paid search;
• direct traffic;
• email;
• or another channel.

AI traffic implication: The most important future AI Search metric may not be citation volume.

It may be:

Commercial Value Influenced by AI Discovery.

Source:

Semrush — AI Search and SEO Study 2026

96. 45% of marketers struggle to measure visibility in AI-generated answers

International Measurement Benchmark. Semrush found that 45% of respondents struggled to measure Brand Visibility within AI-generated answers.

Traditional Search Visibility is relatively mature.

Organisations can track:

• keyword rankings;
• impressions;
• clicks;
• CTR;
• and organic sessions.

AI Search introduces less standardised measurements.

Businesses may need to track:

• prompt coverage;
• Brand Mention frequency;
• citation frequency;
• Recommendation Share;
• competitor visibility;
• answer sentiment;
• and source selection.

AI traffic implication: Organisations that cannot measure AI visibility cannot reliably determine whether changes in AI referral traffic are caused by:

• increased platform usage;
• improved brand visibility;
• stronger citation performance;
• or interface changes.

Source:

Semrush — AI Visibility Measurement Research

97. Only 9% of marketers say they can measure all the AI Search metrics that matter

International Measurement Benchmark. Semrush found that only 9% of marketers said they could measure all of the metrics that mattered to their AI Search activity.

This means more than nine out of ten respondents still reported some measurement gap.

The problem is understandable because AI Search creates performance signals across several systems.

For example:

AI Platform
→ Citation
→ Brand Mention
→ Search
→ Website
→ Analytics
→ CRM
→ Revenue

No single conventional analytics metric captures that entire sequence.

AI traffic implication: AI Search measurement increasingly requires a combined dashboard integrating:

• AI visibility data;
• Search Console;
• web analytics;
• CRM data;
• revenue data;
• and Brand Demand.

This is necessary to move from visibility reporting toward meaningful business measurement.

Source:

Semrush — Operational Gap Study

98. 40% of marketers still rely on manual ChatGPT checks to monitor AI visibility

International Measurement Benchmark. Semrush found that 40% of marketers relied on manually entering prompts into AI tools to check whether their brand appeared.

Manual testing can be useful for exploratory research.

However, it has significant limitations when used as the primary measurement system.

AI answers can vary according to:

• prompt wording;
• platform;
• location;
• personalisation;
• time;
• and model version.

A handful of manual checks therefore cannot provide a reliable longitudinal picture of AI performance.

Semrush also found:

• 38% used a traditional SEO platform;
• 36% used a specialised AEO or GEO tool;
• 13% had no consistent measurement approach.

AI traffic implication: Mature AI Search measurement should move from occasional prompt checking toward repeatable tracking across defined prompt sets, topics and platforms.

Source:

Semrush — AI Search Measurement Study

99. 38% of marketers plan to invest in AI Search optimisation

International Investment Benchmark. Semrush found that 38% of marketers planned investment in AI Search optimisation, including AEO and GEO.

That compared with 36% planning investment in traditional SEO.

Other planned investment areas included:

• 49% — Content Creation;
• 46% — Brand Visibility Across Channels;
• 25% — dedicated AI visibility tools;
• 14% — analytics and measurement.

The figures demonstrate how quickly AI Search has entered mainstream digital marketing budgets.

They also expose an interesting imbalance.

Although measurement is one of the largest reported problems, only 14% planned dedicated investment in analytics and measurement.

AI traffic implication: Organisations may increasingly spend money attempting to improve AI visibility before developing sufficient systems to determine whether that investment generates:

• traffic;
• leads;
• revenue;
• or Brand Demand.

Measurement infrastructure should therefore develop alongside optimisation activity.

Source:

Semrush — AI Search Investment Research 2026

100. 81% of fully integrated SEO and AI Search teams reported more AI-related traffic or leads

International Operational Benchmark. Semrush found a substantial difference between organisations with integrated and separate AI Search workflows.

Among teams with fully integrated SEO and AI Search execution, 81% reported seeing more traffic or leads connected with AI platforms.

Among teams running SEO and AI Search completely separately, the equivalent figure was only 36%.

This is an association based on survey responses rather than evidence that workflow integration alone caused the improvement.

However, the gap is substantial.

Semrush also found that teams already reporting AI Search results were more likely to have:

• clear ownership;
• integrated workflows;
• dedicated measurement;
• and specialised AI visibility tools.

AI traffic implication: AI Search performance is increasingly becoming an organisational capability rather than an isolated SEO tactic.

The strongest operating model connects:

SEO
+ GEO
+ Content
+ Brand
+ Digital PR
+ Entity Authority
+ Analytics
+ Revenue Measurement

within one Search Visibility strategy.

Source:

Semrush — The Operational Gap Between AI Search and SEO, 2026

Overall Analysis — What the 100 AI Search Traffic Statistics Tell Us

The 100 statistics examined in this research indicate that AI Search is changing website traffic in two different directions at the same time.

On one side, AI-generated answers are reducing the need for users to click conventional search results for some queries.

On the other, ChatGPT, Gemini, Perplexity, Google AI experiences and other generative systems are creating entirely new referral and discovery journeys.

The result is not simply a transfer of traffic from Google to ChatGPT.

It is a much broader restructuring of how digital discovery works.

AI Search Has Become a Mainstream UK Behaviour

The strongest starting point is adoption.

UK consumer evidence shows that generative AI is no longer confined to early adopters.

More than half of UK adults in the Which? research use AI tools to search for products, services or advice, while usage rises substantially among younger adults.

ChatGPT has become the most prominent standalone AI Search environment, but it is only one part of an expanding ecosystem that includes:

• Google Gemini;
• Microsoft Copilot;
• Meta AI;
• Perplexity;
• Claude;
• DeepSeek;
• and generative features embedded directly inside conventional search engines.

This means businesses should not define AI Search exclusively as:

traffic from ChatGPT.

AI Search is better understood as a new discovery layer operating across multiple digital environments.

Google Is Becoming an AI Search Platform as Well as a Search Engine

A major part of the AI Search transition is occurring inside Google itself.

AI Overviews increasingly answer questions before the user reaches an external website.

This changes the relationship between:

Ranking → CTR → Traffic

that traditional SEO forecasting relied upon for many years.

The newer model is closer to:

Ranking
→ AI Answer Exposure
→ Citation / Brand Visibility
→ Possible Click
→ Website

This means a business can maintain strong rankings and high impression volumes while receiving fewer organic visits.

The search result itself has changed.

For organisations, this makes it increasingly important to separate:

• ranking loss;
• CTR loss;
• AI Overview exposure;
• citation loss;
• and broader zero-click behaviour.

Zero-Click Search Is a Structural Change, Not an AI-Only Problem

The research also demonstrates that zero-click search existed at substantial scale before the widespread adoption of generative AI.

Google has progressively answered more questions directly through:

• featured snippets;
• knowledge panels;
• local results;
• shopping modules;
• calculators;
• maps;
• and other SERP features.

AI Overviews and AI Mode extend this behaviour by providing more complete and conversational answers.

The strategic implication is important.

The amount of search activity taking place in a market is no longer equivalent to the amount of website traffic available to capture.

Businesses therefore need to distinguish between:

Total Search Demand

and:

Externally Available Search Traffic.

AI Referral Traffic Is Growing Extremely Quickly

Despite the growth of zero-click behaviour, AI systems are also sending rapidly increasing amounts of traffic to external websites.

Large-scale datasets show substantial year-on-year growth across industries including:

• Retail;
• Travel;
• Finance;
• Banking;
• Technology;
• Media;
• Fashion;
• Beauty;
• Consumer Electronics;
• and Online Marketplaces.

The absolute volume remains smaller than conventional organic search in most sectors.

However, the growth rates show why organisations should begin measuring AI referrals now rather than waiting until the channel becomes large.

The relevant question is no longer:

“Does AI send any traffic?”

The evidence clearly shows that it does.

The more useful questions are:

• How much AI traffic is the organisation receiving?
• Which platforms are generating it?
• Which pages receive it?
• How quickly is it growing?
• How does it compare with competitors?
• and what commercial value does it create?

AI-Referred Visitors Can Be More Engaged

One of the most important findings in the research concerns visitor quality.

Adobe’s retail data repeatedly found stronger engagement among AI-referred visitors, including:

• lower bounce rates;
• longer sessions;
• more pages viewed;
• and stronger overall engagement.

This supports a plausible change in the customer journey.

A conventional search visitor may arrive at a website relatively early in the research process.

An AI user may first spend time asking:

• what they should buy;
• which provider is suitable;
• what features matter;
• how products compare;
• and which organisations deserve consideration.

The AI system can therefore perform part of the qualification process before the external visit occurs.

The resulting referral can arrive later in the buying journey.

AI Traffic Is Becoming Commercially Valuable

The conversion evidence is particularly significant.

Adobe’s retail data showed AI-referred visitors progressing during 2025 from substantially underperforming conventional traffic to ultimately outperforming it.

The pattern moved from:

Lower Conversion → Near Parity → Higher Conversion

within less than one year.

By the holiday season, AI-referred shoppers were converting more frequently than visitors from other traffic sources within Adobe’s US retail dataset.

This suggests that AI referral traffic cannot be dismissed as purely informational or experimental traffic.

For some organisations it may increasingly become a genuine customer-acquisition channel.

AI Traffic Varies Substantially by Industry

The research also shows that AI Search adoption is highly sector-specific.

Some categories already generate very large AI referral volumes.

Others remain smaller but are expanding rapidly.

Industries involving complex comparison and research appear particularly suited to AI-assisted discovery.

These include:

• Travel;
• Technology;
• Finance;
• Consumer Electronics;
• Beauty;
• Fashion;
• and high-consideration Ecommerce.

The commercial significance of AI Search should therefore be assessed against:

• sector;
• customer behaviour;
• query complexity;
• purchase value;
• and competitive environment.

A universal AI Traffic benchmark is unlikely to be meaningful for every organisation.

Direct AI Referral Traffic Understates Total AI Influence

Perhaps the most important measurement finding is that direct AI referral traffic does not capture the full customer journey.

Users can encounter a business through an AI answer without immediately clicking.

They may later:

• search for the company on Google;
• type the domain directly;
• visit through another website;
• or return through another marketing channel.

Conventional analytics will normally attribute that later visit to the final channel.

The AI interaction that created the initial awareness can disappear from the attribution record.

This produces a hidden journey:

AI Recommendation
→ Brand Recognition
→ Later Search
→ Website Visit
→ Conversion

The organisation may therefore receive substantially more AI-influenced traffic than its referral report suggests.

Citations, Mentions and Recommendations Are Different Outcomes

The citation research demonstrates another important distinction.

Being used as a source does not necessarily mean the organisation is named.

Being named does not necessarily mean the organisation receives a link.

Being linked does not necessarily mean the organisation is recommended.

AI Search measurement should therefore distinguish between:

• Source Citation
• Brand Mention
• Recommendation Inclusion
• Linked Citation
• Direct Referral
• and Commercial Outcome

These form different stages of the AI discovery funnel.

AI Source Selection Extends Beyond Traditional Rankings

Traditional organic rankings remain important, but AI systems can draw evidence from a much broader range of sources.

Research across AI Overviews, ChatGPT, Gemini and Perplexity demonstrates that:

• high-ranking pages are often cited;
• lower-ranking pages can also be cited;
• third-party listings can influence answers;
• first-party websites remain important;
• and source preferences differ between platforms.

The emerging optimisation challenge is therefore broader than:

Rank One Page for One Keyword.

It increasingly involves:

Building an Authoritative Digital Evidence Ecosystem Around an Entity and Topic.

AI Search and SEO Are Converging

The final group of statistics shows that marketers increasingly recognise this convergence.

SEO, AI Search and GEO are becoming interconnected rather than independent disciplines.

The strongest Search Visibility strategy increasingly connects:

• Technical SEO;
• Content Authority;
• Entity Authority;
• Brand Signals;
• Digital PR;
• structured information;
• AI citation monitoring;
• Search Analytics;
• and commercial attribution.

The operating model therefore moves from:

Traditional SEO

toward:

Search Ecosystem Optimisation.

100 AI Search Traffic Statistics at a Glance

The following summary groups the 100 statistics into the ten major areas examined throughout the research.

CGO AI Search Traffic Measurement Framework™

Traditional web analytics were designed primarily to answer a relatively simple question:

Which channel sent the visitor to the website?

AI Search makes that question increasingly incomplete.

A user can now discover an organisation inside ChatGPT, Gemini, Google AI Overviews, AI Mode, Perplexity or another generative system without immediately visiting the organisation’s website.

The eventual visit may arrive through:

• an AI referral;
• a branded Google search;
• direct navigation;
• a marketplace;
• another publisher;
• or a completely different marketing channel.

The CGO AI Search Traffic Measurement Framework therefore separates AI performance into seven connected measurement layers.

The objective is to connect visibility inside AI systems with measurable website activity and eventual commercial outcomes.

1. AI Search Exposure

The first layer measures whether an organisation has an opportunity to appear within relevant AI-generated answers.

This begins with understanding the queries and prompts customers use when researching:

• problems;
• products;
• services;
• providers;
• brands;
• locations;
• comparisons;
• and recommendations.

Useful measurements include:

• number of tracked AI prompts;
• percentage of prompts producing relevant answers;
• platform coverage;
• topic coverage;
• commercial-intent coverage;
• informational-query coverage;
• and country or market coverage.

This layer establishes the size of the AI visibility opportunity.

A business cannot receive an AI citation, recommendation or referral if it is not present within the research environment customers use.

2. AI Visibility

The second layer measures whether the organisation actually appears.

Visibility can take several forms:

• Brand Mention;
• organisation mention;
• Product Mention;
• service mention;
• Source Citation;
• linked citation;
• or recommendation inclusion.

The most useful measurement is not simply:

“Did we appear?”

but:

“How frequently did we appear compared with the organisations competing for the same customer?”

Potential KPIs include:

• AI Visibility Rate;
• Brand Mention Share;
• Citation Share;
• Recommendation Share;
• competitive visibility;
• platform-specific visibility;
• and visibility by query intent.

3. Citation & Recommendation Quality

Not every AI appearance has the same value.

An organisation mentioned incidentally within an answer has a different commercial position from one explicitly recommended as a leading option.

Similarly, a citation linking directly to the organisation can create a clearer referral opportunity than an unlinked Brand Mention.

This layer therefore evaluates the quality and context of AI visibility.

Measurement can include:

• whether the organisation is recommended;
• whether the organisation is merely referenced;
• whether the citation is linked;
• citation prominence;
• recommendation position;
• sentiment or description;
• accuracy of information;
• and competitive context.

For commercial queries, the organisation should also track whether it appears:

• first;
• within a shortlist;
• as an alternative;
• or not at all.

This helps distinguish simple AI visibility from commercially meaningful AI visibility.

4. Direct AI Referral Traffic

The fourth layer measures users who leave an AI platform and arrive directly at the organisation’s website.

This is the portion of AI Search activity most easily captured using conventional web analytics.

Traffic should be segmented by source wherever possible, including:

• ChatGPT;
• Gemini;
• Perplexity;
• Copilot;
• Claude;
• Google AI experiences where identifiable;
• and emerging AI platforms.

Core traffic KPIs include:

• AI referral sessions;
• AI users;
• new users;
• returning users;
• AI traffic share;
• year-on-year AI traffic growth;
• month-on-month growth;
• landing pages;
• and platform mix.

This layer answers:

“How much identifiable website traffic is AI Search directly generating?”

5. AI Referral Engagement

Traffic volume alone does not establish value.

The fifth layer measures what AI-referred visitors do after reaching the website.

Important engagement metrics include:

• engagement rate;
• bounce rate;
• average engagement time;
• pages viewed;
• scroll depth;
• product views;
• service-page views;
• downloads;
• form starts;
• and repeat visits.

These should be compared against other major acquisition channels such as:

• Organic Search;
• Paid Search;
• Direct;
• Social;
• Email;
• and conventional referral traffic.

The key question becomes:

“Are AI-referred visitors more or less engaged than visitors acquired through other channels?”

6. AI-Influenced Demand

This is the measurement layer most conventional analytics systems are likely to underestimate.

A user may see an organisation recommended in an AI answer but not click immediately.

Later activity may include:

• a branded Google search;
• a direct website visit;
• a search for a specific product;
• a search for the organisation plus reviews;
• or a later visit from another device.

Potential supporting indicators include:

• growth in branded Search Console impressions;
• growth in branded organic clicks;
• homepage direct traffic;
• brand-name search demand;
• branded paid-search activity;
• new-user direct visits;
• and customer self-reported discovery source.

None of these metrics proves AI causation on its own.

However, when analysed alongside improving AI visibility, they can provide supporting evidence that AI Search is contributing to Brand Demand.

The measurement objective is therefore:

Detect AI-Influenced Demand That Direct Referral Reports Cannot See.

7. Leads, Revenue & Commercial Value

The final layer connects AI Search activity with business performance.

For lead-generation organisations, relevant measures can include:

• AI-generated enquiries;
• qualified leads;
• sales opportunities;
• pipeline value;
• closed revenue;
• and customer acquisition cost.

For ecommerce organisations, relevant measures can include:

• Add-to-Cart rate;
• checkout initiation;
• transactions;
• conversion rate;
• average order value;
• revenue per AI session;
• repeat purchase;
• and Customer Lifetime Value.

For publishers or research organisations, commercial value may also include:

• subscriptions;
• research downloads;
• media enquiries;
• citations;
• newsletter sign-ups;
• and authority development.

This final layer answers the most important question:

“What measurable business value is associated with AI Search?”

How to Measure AI Search Traffic

Organisations should begin by separating directly measurable AI traffic from AI-influenced traffic.

These are related but different categories.

Direct AI Referral Traffic

Direct AI referral traffic consists of website visits where the referrer can be identified as an AI platform.

A simple measurement is:

AI Referral Share (%) = AI Referral Sessions ÷ Total Website Sessions × 100

For example, if a website receives:

• 100,000 total monthly sessions;
• and 2,000 identifiable AI referral sessions;

its direct AI referral share is:

2,000 ÷ 100,000 × 100 = 2%

AI Referral Growth

Growth should be monitored independently because a small channel can still be developing extremely quickly.

AI Traffic Growth (%) = (Current AI Sessions − Previous AI Sessions) ÷ Previous AI Sessions × 100

This can be calculated:

• month on month;
• quarter on quarter;
• or year on year.

AI Referral Conversion Rate

Traffic quality should be measured by the proportion of AI visitors completing a commercially meaningful action.

AI Conversion Rate (%) = AI-Referred Conversions ÷ AI Referral Sessions × 100

Conversions may include:

• purchases;
• qualified enquiries;
• appointments;
• subscriptions;
• downloads;
• or another defined business outcome.

AI Revenue per Visit

For transactional organisations, one of the most useful measures is the average commercial value of each AI-referred visit.

AI Revenue per Visit = Revenue Attributed to AI Referrals ÷ AI Referral Sessions

This makes it possible to compare AI traffic economically against channels with much larger traffic volumes.

AI Visibility Rate

Website analytics alone cannot measure the answers in which an organisation appears.

A separate AI visibility dataset is therefore required.

AI Visibility Rate (%) = Relevant AI Answers Featuring the Organisation ÷ Total Tracked AI Answers × 100

AI Citation Share

Where citations are measurable, organisations can calculate their share of citations within a defined query set.

AI Citation Share (%) = Organisation Citations ÷ Total Relevant Citations × 100

This should normally be analysed by:

• platform;
• topic;
• market;
• query type;
• and competitor set.

AI Recommendation Share

For commercial queries, a Recommendation Share can be more meaningful than citation count.

AI Recommendation Share (%) = Relevant Recommendation Answers Featuring the Organisation ÷ Total Recommendation Answers Tracked × 100

This measures whether the organisation is being placed into the AI-generated consideration set.

Research Methodology

This research was developed to provide a structured evidence base for understanding how AI-powered search is affecting website traffic, user behaviour, commercial referral patterns and digital measurement in 2026.

The study combines UK-specific evidence with clearly identified international benchmarks where robust UK datasets are not yet available.

The objective is not to treat every international statistic as directly representative of UK users.

Instead, the research separates evidence according to:

• geography;
• dataset type;
• platform;
• measurement methodology;
• time period;
• and commercial context.

This approach is particularly important in AI Search because adoption, interface design and traffic behaviour can change rapidly between markets and platforms.

Research Scope

The research examines 100 statistics across ten areas:

1. UK AI Search Adoption & Traffic Scale
2. Google AI Overviews & Organic Traffic
3. Zero-Click Search & Open-Web Traffic
4. AI Referral Traffic Growth
5. AI Referral Engagement
6. AI Referral Conversion & Commercial Value
7. AI Traffic by Industry & Sector
8. AI Attribution & Customer Journeys
9. AI Citations, Brand Mentions & Source Selection
10. AI Search Measurement, Adoption & Future Traffic

The study therefore examines both:

traffic generated directly by AI systems

and:

traffic or commercial activity that may be influenced indirectly by AI discovery.

Primary Evidence Categories

Statistics were classified according to the geographic and methodological scope of the underlying evidence.

UK Data

UK-specific evidence was prioritised wherever suitable datasets were available.

Sources used include:

• Ofcom;
• Which?;
• Adobe Digital Insights UK;
• Salesforce UK;
• Ipsos iris data reported by Ofcom;
• and UK Similarweb data reported within Ofcom research.

UK statistics were used particularly for:

• AI Search adoption;
• ChatGPT audience scale;
• AI Overview exposure;
• consumer shopping behaviour;
• and marketing adoption.

European Benchmarks

European evidence was used when it provided useful regional context but should not be interpreted as specifically representing the United Kingdom.

This includes European Union clickstream evidence relating to:

• zero-click search;
• open-web traffic;
• and search-result behaviour.

US Behavioural Benchmarks

A substantial amount of high-quality AI Search research currently originates from the United States.

These datasets were included where they provide useful behavioural evidence relating to:

• Google AI Overviews;
• click-through behaviour;
• AI referral traffic;
• ecommerce engagement;
• conversion;
• and revenue.

They are explicitly identified as US benchmarks and should not be interpreted as direct UK measurements.

International and Global Benchmarks

Global datasets were used to examine:

• AI referral volumes;
• industry differences;
• citation behaviour;
• source selection;
• AI marketing adoption;
• and cross-platform visibility.

These datasets are useful for identifying structural patterns in AI Search but may contain significant country-level variation.

Source Selection Principles

The research prioritised sources that provide identifiable information about:

• sample size;
• measurement period;
• geographic scope;
• methodology;
• platform;
• or underlying dataset.

Preference was given to:

• regulators;
• established research organisations;
• large analytics platforms;
• major digital measurement companies;
• and studies based on substantial behavioural datasets.

Key source organisations include:

• Ofcom
• Which?
• Pew Research Center
• Adobe Digital Insights
• Similarweb
• Semrush
• Ahrefs
• Yext
• SparkToro
• Datos
• Salesforce
• and Ipsos iris

How Statistics Were Selected

A statistic was included where it contributed measurable evidence to one or more of the following research questions:

• How many people are using AI Search?
• How frequently are AI-generated answers appearing?
• What happens to conventional organic clicks when AI answers appear?
• How much external website traffic are AI systems generating?
• How quickly is AI referral traffic growing?
• How do AI-referred visitors behave after arrival?
• How well does AI referral traffic convert?
• How does AI traffic differ between industries?
• How much AI influence is hidden by conventional attribution?
• How are AI systems selecting sources and citations?
• and how are organisations adapting their Search Strategies?

Qualitative statements were not counted among the 100 statistics unless they were supported by an underlying numerical measurement.

This is an important distinction from broad industry commentary.

The objective of the Statistics Library is to separate:

measured evidence

from:

strategic interpretation.

Statistical Interpretation

The statistics in this research should not automatically be interpreted as causal relationships.

For example, if AI-referred visitors convert more frequently than other visitors, this does not necessarily prove that the AI platform itself caused the higher conversion rate.

Possible contributing factors include:

• different user intent;
• greater research completed before the visit;
• different product categories;
• seasonality;
• platform demographics;
• and referral selection effects.

Where research demonstrates correlation rather than causation, the interpretation throughout this paper is intentionally framed as an association.

Time Period

The evidence included in this research primarily covers:

2024 to 2026

with the greatest emphasis placed on the most recent available 2025 and 2026 datasets.

Earlier evidence is included only where it provides useful context for measuring the speed of change.

This matters because AI Search is evolving unusually quickly.

A benchmark measured only twelve months earlier may no longer represent current platform behaviour.

Platform Coverage

The research includes evidence relating to platforms and interfaces including:

• ChatGPT;
• Google Search;
• Google AI Overviews;
• Google AI Mode;
• Gemini;
• Microsoft Copilot;
• Perplexity;
• Claude;
• Meta AI;
• DeepSeek;
• and other generative AI environments where included in source datasets.

Not every source measures every platform.

Results should therefore be interpreted within the platform scope of the original research.

Traffic Definition

This study deliberately uses a broader definition of AI Search Traffic than direct referrals alone.

Three different traffic categories should be distinguished:

Direct AI Referral Traffic

Website sessions where an identifiable AI platform is recorded as the referring source.

AI-Influenced Traffic

Website traffic where earlier AI exposure may have influenced the user, but the eventual visit arrives through another channel such as Google Organic or Direct.

AI Visibility Without Traffic

Situations where an organisation is cited, mentioned or recommended inside an AI-generated answer but the user does not visit the website.

These outcomes should not be combined into a single metric because they represent different forms of Search Visibility and commercial influence.

Research Limitations

AI Search is developing more quickly than many established forms of digital measurement.

The statistics in this research should therefore be interpreted with several limitations in mind.

1. UK-Specific Data Remains Limited

Although UK-specific evidence is used wherever possible, many of the largest behavioural datasets currently available originate from the United States or combine users from multiple countries.

International evidence can identify important patterns, but it should not automatically be assumed to represent UK behaviour precisely.

2. AI Platforms Change Rapidly

ChatGPT, Google AI Overviews, AI Mode, Gemini, Perplexity and other generative systems are continually changing.

Updates can affect:

• answer construction;
• link prominence;
• source selection;
• citation behaviour;
• web retrieval;
• and referral traffic.

A statistic accurately measured in one period may therefore change significantly after a platform update.

3. Different Studies Use Different Definitions

Terms such as:

• zero-click;
• AI referral;
• AI visibility;
• citation;
• Brand Mention;
• and conversion

are not always defined identically across research organisations.

Direct comparisons between datasets should therefore be made carefully.

4. Referral Data Understates AI Influence

Direct referral tracking captures only users who click from an identifiable AI environment to a website.

It does not fully capture:

• AI-assisted branded search;
• later direct visits;
• cross-device behaviour;
• offline conversion;
• or delayed customer journeys.

AI’s commercial influence may therefore be larger than referral reports indicate.

5. AI Answers Can Vary Between Users and Sessions

AI responses can differ according to:

• prompt wording;
• conversation history;
• geography;
• platform;
• model version;
• personalisation;
• and retrieval conditions.

An organisation appearing in one prompt test is therefore not evidence that it will appear consistently for every user.

6. Citation Does Not Equal Recommendation

A source can be cited because it provides useful evidence without the AI system recommending the associated organisation.

Likewise, an organisation may be recommended using information gathered from third-party sources rather than from its own website.

Citation, mention and recommendation should therefore remain separate KPIs.

7. Industry Differences Are Significant

AI Search adoption and referral behaviour differ considerably between industries.

A benchmark from ecommerce should not automatically be applied to:

• Healthcare;
• Legal;
• Financial Services;
• B2B SaaS;
• Manufacturing;
• or Professional Services.

Sector-specific measurement is preferable wherever sufficient data exists.

8. Very High Growth Rates Can Reflect a Small Starting Base

AI referral traffic frequently shows percentage increases of several hundred percent.

These figures can appear dramatic because the channel was very small at the beginning of the comparison period.

Growth rate should therefore always be considered alongside:

• absolute traffic volume;
• share of total website traffic;
• conversion rate;
• and revenue contribution.

9. Survey Data Reflects Reported Behaviour

Research based on surveys measures what participants report doing or intending to do.

Reported behaviour can differ from observed clickstream behaviour.

Where possible, this paper therefore combines:

• survey evidence;
• audience measurement;
• clickstream data;
• and website analytics research.

10. Attribution Remains an Unresolved Measurement Challenge

Current analytics systems were not originally designed to measure conversational AI discovery.

A user may move across:

ChatGPT
→ Google
→ Website
→ CRM
→ Offline Sale

without the complete journey being connected.

This means current AI attribution should often be considered an estimate rather than a complete record of every influenced customer journey.

Recommended Related CGO Media Research

AI Search Traffic should not be studied in isolation.

Traffic represents only one part of a wider discovery system involving:

• AI Search adoption;
• Search Visibility;
• source selection;
• citations;
• Brand Mentions;
• recommendations;
• customer behaviour;
• conversion;
• and commercial return.

The following CGO Media resources provide additional research and measurement frameworks for understanding these connected areas.

AI Search Visibility Research Observations UK 2026

AI Search Visibility Research examines how organisations can measure whether they are recognised, cited, mentioned and recommended across generative search platforms.

It provides a useful companion to this traffic research because visibility occurs before referral traffic.

Together, the two resources examine the relationship:

AI Visibility → Recommendation → Referral → Commercial Outcome

Explore AI Search Visibility Research UK 2026 →

AI Citation Authority Research Observations UK 2026

AI Citation Authority Research examines how organisations become sources that AI systems select, reference and cite.

It is directly relevant to AI Search Traffic because external referral opportunities frequently begin with source selection.

The research explores:

• Citation Authority;
• source credibility;
• external validation;
• Entity Authority;
• content evidence;
• and AI recommendation visibility.

Explore AI Citation Authority Research UK 2026 →

SEO ROI Statistics UK 2026

SEO ROI Statistics UK 2026 examines the wider commercial economics of Search Visibility.

The resource contains 100 statistics covering:

• UK Search Behaviour;
• organic rankings;
• click-through rates;
• lead generation;
• conversion;
• Local SEO;
• Ecommerce SEO;
• AI Overviews;
• Technical SEO;
• and financial return.

It provides a useful benchmark for comparing established SEO economics with the emerging economics of AI Search.

Explore SEO ROI Statistics UK 2026 →

CGO Media Statistics Library

The CGO Media Statistics Library brings together quantitative resources examining changes across traditional Search, AI Search, GEO, digital visibility and commercial performance.

Research areas include:

• AI Search;
• ChatGPT usage;
• Google AI Overviews;
• AI citations;
• Search Behaviour;
• SEO ROI;
• Local SEO;
• Technical SEO;
• Ecommerce;
• and wider Digital Marketing.

Explore the CGO Media Statistics Library →

Latest CGO Media Research

Because AI Search changes rapidly, traffic benchmarks, citation behaviour and platform functionality can change within relatively short periods.

The Latest Research page provides access to newly published and substantially updated CGO Media research across:

• AI Search;
• GEO;
• source selection;
• citation systems;
• recommendation systems;
• Entity Authority;
• and Future Search.

View the Latest CGO Media Research →

Frequently Asked Questions About AI Search Traffic

What is AI Search Traffic?

AI Search Traffic is website traffic generated through AI-powered discovery environments such as ChatGPT, Gemini, Perplexity, Copilot and other generative search systems.

The term can also be used more broadly to examine website activity influenced by AI recommendations even when the final visit arrives through another channel.

For measurement purposes, it is useful to distinguish between:

• Direct AI Referral Traffic — identifiable visits directly from an AI platform;
• AI-Influenced Traffic — visits occurring later through search, direct navigation or another channel after earlier AI exposure.

How much traffic does ChatGPT send to websites?

There is no single percentage applicable to every website.

ChatGPT referral volumes vary according to:

• industry;
• Brand Authority;
• content;
• query type;
• citation frequency;
• and how often ChatGPT uses the live web for relevant queries.

Large clickstream studies show that ChatGPT is sending increasing amounts of traffic to external websites, but conventional Google Search remains much larger overall.

Is AI Search replacing Google?

Current evidence does not support treating AI Search and Google as completely separate alternatives.

Users frequently move between them.

A journey might involve:

ChatGPT → Google → Website

or:

Google AI Overview → Brand Search → Website.

AI Search and traditional Search are increasingly becoming interconnected parts of the same customer journey.

Are Google AI Overviews reducing organic traffic?

Several behavioural studies have found substantially lower conventional organic click-through rates when AI Overviews appear.

However, the size of the effect varies according to:

• query;
• search intent;
• industry;
• country;
• device;
• and study methodology.

Businesses should therefore measure AI Overview exposure across their own keyword portfolio rather than applying one universal traffic-loss percentage.

What is zero-click search?

A zero-click search occurs when a search does not generate an external website click.

This can happen because the user:

• receives the answer directly;
• changes the search;
• ends the session;
• uses a Google-owned feature;
• or continues research without visiting an external website.

Zero-click search existed long before generative AI, but AI-generated answers can expand the amount of information available without leaving the search environment.

Does being cited by an AI platform generate traffic?

Not necessarily.

A citation creates a potential referral opportunity, but many users consume the answer without clicking the source.

AI citations can still create value through:

• Brand Exposure;
• source authority;
• trust;
• later branded searches;
• and Recommendation Visibility.

Citation performance and direct referral traffic should therefore be measured separately.

What is the difference between an AI citation, Brand Mention and recommendation?

An AI citation means a source has been used or linked within the answer.

A Brand Mention means the organisation is explicitly named.

An AI recommendation means the organisation is presented as a relevant option, provider, product or solution.

One answer can contain all three, but they do not necessarily occur together.

Can AI Search increase branded Google searches?

Yes, AI-assisted discovery can create a customer journey in which the user first learns about an organisation from an AI system and later searches for that organisation on Google.

In that situation, conventional analytics may attribute the final visit to Organic Search even though AI contributed to the earlier discovery stage.

This is why branded search demand should be monitored alongside direct AI referrals.

Is AI referral traffic more valuable than organic traffic?

There is no universal answer.

Some large retail datasets have found AI-referred users showing:

• stronger engagement;
• lower bounce rates;
• longer sessions;
• and, in later 2025 periods, higher conversion rates.

However, performance varies by sector and website.

The correct comparison should use the organisation’s own:

• engagement;
• conversion;
• revenue;
• and Customer Lifetime Value data.

How should a business track AI referral traffic?

At minimum, organisations should create a dedicated AI Traffic segment within their analytics environment.

The segment should monitor known AI referral sources and report:

• sessions;
• users;
• landing pages;
• engagement;
• conversions;
• revenue;
• and growth over time.

This should be combined with separate AI visibility and citation tracking.

Why does AI attribution remain difficult?

AI platforms can influence a user without generating a direct referral.

The eventual conversion may occur through:

• Google Organic;
• Paid Search;
• Direct;
• Email;
• Social;
• or another device.

Traditional last-click analytics may therefore record the final acquisition channel while missing the earlier AI interaction.

Which industries are seeing the strongest AI referral growth?

Current international datasets show substantial AI referral growth across sectors including:

• Retail;
• Travel;
• Finance;
• Technology;
• Beauty;
• Fashion;
• Consumer Electronics;
• Media;
• and Online Marketplaces.

Industries involving complex research and comparison appear particularly well suited to conversational discovery.

Will AI Search Traffic become more important?

Current evidence shows continuing growth in:

• AI Search adoption;
• AI referral traffic;
• AI-assisted shopping;
• marketing investment;
• and AI Search optimisation.

The exact balance between zero-click answers and external referrals will continue changing as platforms modify their interfaces and citation behaviour.

For this reason, organisations should establish measurement systems now and monitor the direction of change over time.

Does traditional SEO still matter in an AI Search environment?

Yes.

AI systems continue to rely on digital information created across:

• websites;
• search indexes;
• business listings;
• publishers;
• structured information;
• reviews;
• and wider web sources.

Technical accessibility, Content Authority, Entity Authority, Digital PR and Brand Signals therefore remain important.

The change is that SEO increasingly operates as one component of a broader Search and AI Visibility strategy.

Conclusion — AI Search Traffic in 2026

AI Search is changing the relationship between digital visibility, website traffic and commercial discovery.

The 100 statistics examined in this research show that this change cannot be described simply as:

“People Are Moving From Google to ChatGPT.”

The reality is considerably more complex.

Consumers are now moving between:

• Google Search;
• Google AI Overviews;
• AI Mode;
• ChatGPT;
• Gemini;
• Copilot;
• Perplexity;
• marketplaces;
• publisher websites;
• brand websites;
• and direct navigation.

The modern Search Journey is therefore becoming multi-platform, conversational and increasingly difficult to attribute through conventional last-click analytics.

AI Search Is Already Part of UK Consumer Behaviour

UK-specific evidence shows that AI-assisted search has moved into mainstream consumer behaviour.

Consumers are already using generative AI systems to:

• research products;
• compare services;
• find information;
• evaluate providers;
• plan purchases;
• and seek recommendations.

Younger audiences show particularly high adoption, suggesting that AI-assisted discovery is likely to become even more significant as these behaviours mature.

Traffic Is Becoming Harder to Interpret

Traditional SEO analysis often relied on a relatively direct relationship:

Ranking → Click → Visit

AI Search introduces several intermediate stages.

The emerging journey can instead be:

Query
→ AI Answer
→ Citation or Recommendation
→ Brand Recognition
→ Possible Referral
→ Later Search
→ Website Visit

This means a decline in direct organic traffic does not necessarily mean an organisation has become less visible.

Likewise, an increase in branded search or direct visits may partly reflect discovery that occurred earlier within an AI system.

Zero-Click Search Will Increase the Importance of Visibility Beyond the Website Visit

The growth of zero-click search means organisations increasingly need to create value before a visitor reaches their website.

That value can include:

• Brand Recognition;
• Citation Authority;
• Recommendation Visibility;
• Product Discovery;
• trust;
• and inclusion within the customer’s consideration set.

Website traffic remains important.

But it can no longer represent the entire value created by Search Visibility.

AI Referral Traffic Is Becoming a Real Acquisition Channel

At the same time, generative AI systems are sending rapidly increasing volumes of traffic to external websites.

The evidence examined in this study shows strong referral growth across sectors including:

• Ecommerce & Retail;
• Travel & Hospitality;
• Financial Services;
• Technology;
• Media;
• Beauty;
• Fashion;
• Consumer Electronics;
• and Online Marketplaces.

The absolute importance of AI referrals varies considerably by industry.

However, the speed of growth means organisations should already be establishing baseline data.

Traffic Quality May Matter More Than Traffic Volume

One of the most commercially important findings is that AI-referred visitors can demonstrate strong post-click behaviour.

Large retail datasets have reported AI visitors showing:

• lower bounce rates;
• longer sessions;
• greater page depth;
• and increasingly strong conversion performance.

This supports the possibility that AI systems are performing part of the user’s research and qualification process before the external visit occurs.

The most commercially useful AI Traffic KPI may therefore become:

Value per AI-Referred Visitor

rather than traffic volume alone.

Direct AI Referrals Will Understate the True Influence of AI Search

The evidence also indicates that many AI-influenced customer journeys do not generate an immediately identifiable AI referral.

A customer may:

Discover a Brand in ChatGPT
↓
Remember the Brand
↓
Search Google Later
↓
Visit the Website
↓
Convert

Analytics may classify that customer as Organic Search.

The earlier AI recommendation disappears from conventional attribution.

This is why organisations increasingly need to measure:

• AI Referral Traffic;
• Brand Search Demand;
• direct traffic;
• AI Visibility;
• Recommendation Visibility;
• and commercial outcomes

together.

AI Search Visibility Is Not the Same as AI Traffic

Another major finding is the difference between:

• being cited;
• being mentioned;
• being recommended;
• being linked;
• and receiving traffic.

These are separate stages.

An organisation may be highly visible without receiving many clicks.

Another may receive traffic from only a small number of highly commercial recommendations.

The most useful reporting therefore evaluates the complete chain:

Visibility
→ Citation
→ Mention
→ Recommendation
→ Referral
→ Engagement
→ Conversion

SEO and AI Search Are Becoming One Search Ecosystem

The evidence does not suggest that organisations should abandon conventional SEO.

It suggests that the scope of SEO is becoming broader.

AI Search systems continue to rely on digital evidence produced across:

• websites;
• search indexes;
• publishers;
• directories;
• business listings;
• reviews;
• structured information;
• research;
• and wider third-party sources.

Traditional SEO therefore remains an important foundation.

But the target is expanding from:

Organic Ranking

toward:

Search Ecosystem Visibility.

The Measurement Model for 2026

The strongest organisations will increasingly connect seven areas:

1. AI Search Exposure
2. AI Visibility
3. Citation & Recommendation Quality
4. Direct AI Referral Traffic
5. AI Referral Engagement
6. AI-Influenced Demand
7. Commercial Value

This moves measurement beyond the question:

“How many visits did ChatGPT send?”

toward:

“How much customer discovery, demand, traffic and revenue is AI Search influencing?”

References

The following sources provide the statistical and methodological evidence used throughout AI Search Traffic Statistics UK 2026.

Where research is based on international rather than UK-specific data, that distinction has been identified throughout the paper.

1. Ofcom. (2025). From Apps to AI Search: How the UK Goes Online in 2025 / Online Nation 2025. UK communications regulator. View source.
2. Ofcom. (2025). The Era of Answer Engines: Generative AI’s Impact on Search Experiences and Online Safety. Analysis incorporating Ipsos iris audience measurement. View source.
3. Which? (2025). Consumer Use and Attitudes Towards AI Search Tools. Nationally representative survey of 4,189 UK adults conducted in September 2025. View source.
4. Pew Research Center. (2025). Google Users Are Less Likely to Click on Links When an AI Summary Appears in the Results. Analysis of 68,879 Google searches generated by 900 US adults. View source.
5. Ahrefs. (2026). Update: AI Overviews Reduce Clicks by 58%. Analysis of 300,000 keywords using aggregated Google Search Console data. View source.
6. SparkToro & Datos. (2024). 2024 Zero-Click Search Study: For Every 1,000 EU Google Searches, Only 374 Clicks Go to the Open Web. In the US, It’s 360. View source.
7. Semrush. (2025). Google AI Mode’s Early Adoption and SEO Impact. Clickstream analysis examining Google AI Mode behaviour and external referral activity. View source.
8. Adobe Digital Insights. (2025–2026). AI-Driven Traffic Surges Across Industries. Analysis based on Adobe Analytics data covering more than one trillion visits to US retail websites. View source.
9. Adobe Digital Insights. (2025). Generative AI-Powered Shopping Rises with Traffic to U.S. Retail Sites. Research examining AI referral engagement and ecommerce conversion. View source.
10. Adobe Digital Insights UK. (2025). Retail in Flux: GenAI, Prime Day and the Future of Shopping. UK consumer and AI referral research. View source.
11. Semrush. (2026). ChatGPT Traffic Analysis: Insights from 17 Months of Clickstream Data. Analysis of more than one billion lines of US clickstream data. View source.
12. Similarweb. (2026). The Downstream Impact of AI Visibility. Research examining how AI recommendations influence subsequent brand visits, search activity and website engagement. View source.
13. Similarweb. (2026). AI Referral Traffic by Industry. Generative AI Landscape analysis examining global AI referrals across major commercial sectors. View source.
14. Semrush. (2026). Why 62% of AI Citations Don’t Lead to Brand Mentions. Cross-platform study of AI citations, Brand Mentions and ghost citations. View source.
15. Yext Research. (2025). AI Citations, User Locations & Query Context. Analysis of 6.8 million citations across Gemini, OpenAI and Perplexity. View source.
16. Ahrefs. (2026). Update: 38% of AI Overview Citations Pull From the Top 10. Analysis of 863,000 keyword SERPs and approximately four million AI Overview URLs. View source.
17. Ahrefs. (2025). 86% of Top Mentioned Sources Are Not Shared Across ChatGPT, Perplexity and AI Overviews. Cross-platform source analysis using Ahrefs Brand Radar. View source.
18. Salesforce. (2026). State of Marketing 2026. Global research involving 4,450 marketing decision-makers, including 250 UK marketers. View source.

CGO Media Research Ecosystem

AI Search Traffic Statistics UK 2026 forms part of the wider CGO Media Search and AI Research Programme.

The programme examines how traditional Search, generative AI and digital authority are converging into a broader discovery environment.

Research is organised across several connected libraries and knowledge resources.

Research Library

The CGO Media Research Library contains longer-form studies examining the structural development of Search, AI Discovery, source selection, citation systems, recommendation systems and Digital Authority.

Explore the CGO Media Research Library →

Framework Library

The Framework Library translates research findings into structured models covering areas including:

• AI Search Readiness;
• AI Citations;
• Entity Authority;
• Content Authority;
• Brand Signals;
• Search Visibility;
• Technical SEO;
• and GEO.

Explore the CGO Media Framework Library →

Statistics Library

The Statistics Library brings together quantitative resources designed for:

• business leaders;
• marketers;
• journalists;
• researchers;
• SEO professionals;
• and organisations evaluating Search and AI trends.

Explore the CGO Media Statistics Library →

Research Observations

CGO Media Research Observations examine emerging patterns where evidence is developing quickly and long-term conclusions remain premature.

This allows new changes in AI Search to be documented while keeping a clear distinction between:

• measured statistics;
• research observations;
• and established frameworks.

Latest Research

The Latest Research resource provides access to newly published studies, updated datasets and emerging CGO Media research.

Explore Latest Research →

Research Methodology

CGO Media’s research methodology explains the evidence, classification and source-selection principles used across the wider research programme.

Read the CGO Media Research Methodology →

About the Research

Roger Wilkinson

Roger Wilkinson is an independent researcher, SEO practitioner and Founder of CGO Media.

His work focuses on the evolution of Search from conventional search-engine discovery toward an environment increasingly shaped by generative AI, entity understanding, source selection, citations and recommendation systems.

Current research areas include:

• AI Search;
• Generative Engine Optimisation;
• AI answer construction;
• source selection;
• AI Citation Authority;
• Entity Authority;
• Brand Authority;
• Knowledge Architecture;
• Search Visibility;
• and the future economics of Search.

ORCID:

0009-0004-3325-0740

CGO Media Research Team

The CGO Media Research Team develops research, statistical resources and practical frameworks examining the changing relationship between traditional Search, AI Search, GEO and organisational Digital Authority.

The team separates quantitative statistics from research observations and strategic interpretation wherever possible, with particular attention to:

• source provenance;
• geographic relevance;
• methodological limitations;
• and the distinction between correlation and causation.

Meet the CGO Media Research Team →

Research Usage & Citation

This research may be cited by journalists, researchers, academics, businesses and other organisations when discussing AI Search Traffic, generative AI referrals, Search Behaviour or the changing relationship between AI Search and website traffic.

When citing an individual statistic, CGO Media recommends also reviewing and referencing the original underlying source identified within the relevant section.

Recommended Citation

Digital Object Identifier: 10.5281/zenodo.22958500

Journalists & Media

Journalists and media organisations may reference statistics and analysis from this research with attribution to CGO Media and a link to the research page.

Where a statistic originates from an external study, the original source should also be consulted before publication.

For press enquiries, expert commentary, methodology questions or research data enquiries, visit:

CGO Media Press & Media Resources →

Researchers & Academic Use

Researchers may cite the CGO Media analysis, framework and classification structure while retaining the source attribution attached to individual third-party statistics.

The research is intended to support further investigation into:

• AI Search Traffic;
• AI Search Behaviour;
• zero-click Search;
• AI referral economics;
• AI citation systems;
• Search Attribution;
• and the commercial impact of generative discovery.

Businesses & Marketing Teams

Businesses may use the framework within this paper to develop internal AI Search reporting and measurement systems.

The seven recommended measurement layers are:

1. AI Search Exposure
2. AI Visibility
3. Citation & Recommendation Quality
4. Direct AI Referral Traffic
5. AI Referral Engagement
6. AI-Influenced Demand
7. Leads, Revenue & Commercial Value