Skip to content

SEO statistics 2026: search, AI Overviews and GEO

SEO statistics 2026 brings together dated figures on Google, zero-click search, AI Overviews, GEO, AEO and UK usage.

SEO and GEO statistics 2026: a search box splitting into a classic results list and an AI answer panel, flat vector

SEO statistics 2026 only make sense with the study design beside them. A US panel result cannot answer a UK adoption question, and a vendor keyword tracker is not the behaviour of every searcher.

Key statistics

  • Statcounter recorded Google at 89.94% of worldwide and 91.39% of UK search referrals in September 2026. It recorded Bing at 5.29% worldwide and 6.08% in the UK (worldwide; UK). These are referral shares from measured pageviews, not counts of searches.
  • Pew found an AI summary on 18% of US Google searches in March 2025. The rate was 60% for queries beginning with question words such as who, what, when or why, and 53% for queries of 10 or more words (Pew Research Center).
  • SparkToro and Similarweb reported zero-click rates of 68.01% for US Google searches and 69.5% for UK Google searches in January to April 2026 (US study; UK results).
  • The GEO-bench paper (Aggarwal et al., 2023) reports different lifts for different tactics. Quotation addition produced about +41%, statistics addition about +31%, and source citation about +27.5% on its Position-Adjusted Word Count metric (paper).
  • Ofcom's 2026 survey found that 54% of UK adults used AI tools, while 75% of online adults read AI search summaries at least sometimes and 42% read them often or always (report).
  • BrightLocal's 2026 survey covers the US. It found 71% used Google as a source for reviews and 45% used ChatGPT or other generative AI tools for local recommendations (survey).

What do SEO, GEO and AEO mean?

The labels overlap, but each points to a different job. Search Engine Optimisation improves a page's ability to be crawled, understood and ranked in search results. Generative Engine Optimisation focuses on being selected and cited in generated answers. Answer Engine Optimisation focuses on clear answers that systems can extract for a direct response.

TermMain targetUseful evidenceWhat to check
SEOOrganic results and qualified visitsRankings, impressions, clicks, conversionsCrawlability, intent match, content, links and page experience
GEOCitations in generated answersMention rate, cited URLs, answer shareFactual statements, source quality, entity clarity and coverage
AEODirect answers and answer featuresAnswer visibility, rich-result eligibility, assisted visitsQuestion wording, concise answers, structured data and internal links
Local SEOMap results and local recommendationsProfile actions, calls, directions, local rankingsBusiness details, reviews, categories, service areas and local pages

GEO and AEO are practical labels, not settled standards, and the studies below do not add up to a universal ranking-factor list. Start with technical SEO, then test whether important pages appear in the answers customers actually use. Google's 2026 generative AI guidance says SEO best practices continue to apply to generative AI features in Search.

Search market share: what is the UK position?

Google leads Statcounter's measured referrals in the UK and worldwide. These are monthly shares, not counts of searches; the latest complete month checked is September 2026.

MeasureGeography and dateValueSource
Google search-engine shareWorldwide, September 2026; Statcounter referral/pageview measure, 3bn+ pageviews monthly across 1m+ sites; September sample count not stated89.94%Statcounter
Bing search-engine shareWorldwide, September 2026; same referral/pageview measure; September sample count not stated5.29%Statcounter
Google search-engine shareUK, September 2026; Statcounter referral/pageview measure; country sample count not stated91.39%Statcounter UK
Bing search-engine shareUK, September 2026; same referral/pageview measure; country sample count not stated6.08%Statcounter UK
UK online adults reached by Google SearchUK online adults aged 18+, May 2025; Ipsos iris passive audience panel, sample n not stated82%Ofcom Online Nation 2025
Average daily time onlineUK internet users aged 18+, May 2025; personal smartphones, tablets and computers; audience-panel n not stated4 hours 30 minutesOfcom Online Nation 2025
Online retail sales shareGreat Britain retail businesses, September 2026; ONS Monthly Business Survey, Retail Sales Inquiry; 59.1% response rate and 92.4% turnover coverage, respondent count not stated28.8%ONS, 2026-10-23
Small businesses as a share of UK private-sector businessesUK, start of 2025; DBT estimate, businesses with 0 to 49 employees; 5,690,265 total businesses, not a survey sample99.18%DBT Business Population Estimates 2025

For a UK company, Google still belongs in the measurement plan, though market share excludes discovery on ChatGPT, TikTok, Amazon and YouTube. Google's structured data documentation explains how markup helps Google understand page content and may make a page eligible for a rich result. Google does not guarantee that structured data will appear or provide a general click-through uplift.

What percentage of searches end without a click?

SparkToro and Similarweb define a zero-click Google search as one that ends without a click. Their 2026 browser clickstream panel covers January to April and excludes Google's mobile search app. A zero-click search can end with the session or another query, so it does not mean the searcher found nothing useful.

MeasureGeography and dateResultSource
Google searches ending without a clickUS, January to April 2026; Similarweb desktop and mobile browser clickstream; panel n not reported, Google search app excluded68.01%SparkToro and Similarweb, 2026-06-08
Google searches ending without a clickUK, January to April 2026; Similarweb desktop and mobile browser clickstream; panel n not reported, Google search app excluded69.5%SparkToro and Similarweb, 2026-06-16
Clicks to non-Google properties per 1,000 searchesUS, January to April 2026; same Similarweb browser panel; panel n not reported231SparkToro and Similarweb, 2026-06-16
Clicks to non-Google properties per 1,000 searchesUK, January to April 2026; same Similarweb browser panel; panel n not reported232SparkToro and Similarweb, 2026-06-16
Google result-page visits followed by a traditional result click without an AI summaryUS panel, March 2025; 900 KnowledgePanel Digital adults, 68,879 searches; next URL matched to a traditional result15%Pew Research Center
Google result-page visits with an AI summary followed by a traditional result clickUS panel, March 2025; same 900 adults and browsing sample; next URL matched to a traditional result8%Pew Research Center
Visits to summary pages followed by a click on a cited sourceUS panel, March 2025; same 900 adults; next URL matched to one of the first three cited links1%Pew Research Center
Pages with an AI summary where the session endedUS panel, March 2025; same browsing sample; session end means browser exit for at least five seconds26%Pew Research Center

How are AI Overviews changing click-through rates?

Pew's study is the clearest source here for observed US behaviour. An AI summary appeared on 18% of 68,879 Google searches in a panel of 900 adults in March 2025. It appeared on 60% of queries beginning with question words such as who, what, when or why, and 53% of queries with 10 or more words. Traditional result clicks followed 15% of visits without a summary and 8% of visits with one. A click on a source link in the summary occurred on 1% of summary-page visits.

Traditional result, no AI summary on thepage15%Traditional result, AI summary on the page8%Source link inside the AI summary1%
Share of Google result-page visits followed by a click, March 2025, 900 US adults, 68,879 searches. Source: Pew Research Center.
ItemValue
Traditional result, no AI summary on the page15%
Traditional result, AI summary on the page8%
Source link inside the AI summary1%

A separate Ahrefs analysis of 300,000 keywords reported a 58% lower average click-through rate for the top-ranking page when an AI Overview appeared. That is a keyword analysis, not a panel observation, so read it as an association rather than a reduction every site will see (Ahrefs study).

MeasureSample and dateResultSource
Google searches with an AI summaryUS panel, March 2025; 900 KnowledgePanel Digital adults, 68,879 searches18%Pew
Queries beginning with question words with an AI summaryUS panel, March 2025; same 900 adults and 68,879-search dataset60%Pew
Queries of 10 or more words with a summaryUS panel, March 2025; same 900 adults and 68,879-search dataset53%Pew
AI summaries citing three or more sourcesUS panel, March 2025; summaries found among 12,593 summary-producing searches88%Pew
AI summaries citing one sourceUS panel, March 2025; same set of summary-producing searches1%Pew
Position-one CTR association when an AI Overview appearsAhrefs keyword analysis, December 2025; 300,000 keywords, 150,000 with AIOs and 150,000 informational controls; desktop GSC CTR58% lowerAhrefs

Track impressions and citations alongside clicks: a citation may not produce a visit, and none of these sources measures its business value. Google's guidance says sites do not need special AI markup or a separate AI file to appear in AI features.

What do GEO and AEO studies say about citations?

The GEO-bench experiment used 10,000 queries split into training, validation and test groups. It measured visibility with Position-Adjusted Word Count. Its results describe tactics under the paper's conditions, not gains a live page will repeat.

Experiment measureReported resultInterpretationSource
GEO-bench size10,000 queriesExperimental benchmark sizeAggarwal et al., KDD 2024
Maximum reported visibility liftUp to 40%Maximum under the paper's conditionsAggarwal et al.
Quotation Addition+41%Position-Adjusted Word Count over baselineAggarwal et al.
Statistics Addition+31%Position-Adjusted Word Count over baselineAggarwal et al.
Cite Sources+27.5%Position-Adjusted Word Count over baselineAggarwal et al.
AI summaries citing three or more sources88%Observed US Google summariesPew Research Center
Government share of AI-summary links6% versus 2%AI summaries compared with standard resultsPew Research Center

Do not add the paper's tactic results together. A page that adds quotations, statistics and citations does not automatically receive a 41% plus 31% plus 27.5% gain. The experiment uses GPT-3.5 with Google results and a separate Perplexity test; it does not establish a fixed order of importance across Google, ChatGPT, Perplexity or Gemini.

Our own editorial advice, not a measured effect: state who collected a figure, when and what it means, put the answer near the question, and explain specialist terms before using them. The related guide on what agentic SEO means covers the execution layer that follows this editorial work.

What does Ofcom Online Nation 2025 show about the UK?

Ofcom's Online Nation 2025 recorded 1.8 billion ChatGPT visits in the first eight months of 2025, compared with 368 million in the same period of 2024. Its 2026 Adults' Media Use and Attitudes report found that 54% of UK adults used AI tools such as ChatGPT, Copilot or Gemini. It also found that 75% of online adults read AI search summaries at least sometimes and 42% read them often or always. The survey was of 7,533 UK adults aged 16+, with fieldwork from 2025-09-29 to 2025-11-28. ChatGPT visits are website visits, not unique people or searches (Ofcom 2025; Ofcom 2026 report).

MeasureUK resultDate or periodSource
ChatGPT website visits1.8 billionJanuary to August 2025; UK desktop and mobile browser visits, SimilarwebOfcom Online Nation 2025
ChatGPT website visits368 millionJanuary to August 2024; UK desktop and mobile browser visits, SimilarwebOfcom Online Nation 2025
UK adults using AI tools such as ChatGPT, Copilot or Gemini54%published 2026-04-02; UK adults aged 16+, total survey n=7,533; fieldwork 2025-09-29 to 2025-11-28Ofcom Adults' Media Use and Attitudes 2026
Online adults reading AI search summaries at least sometimes75%published 2026-04-02; UK adults 16+, total survey n=7,533, fieldwork 2025-09-29 to 2025-11-28; shown an example summaryOfcom Adults' Media Use and Attitudes 2026
Online adults reading AI search summaries often or always42%Same Ofcom survey, published 2026-04-02; fieldwork 2025-09-29 to 2025-11-28; self-reported frequency after seeing an example summaryOfcom Adults' Media Use and Attitudes 2026

What do local search and review statistics show?

BrightLocal's 2026 Local Consumer Review Survey is a US survey, so do not read it as a UK adoption forecast.

MeasureUS resultSource
US consumers using Google as a source for business reviews71%BrightLocal
Consumers using ChatGPT or similar tools for local recommendations45%BrightLocal
Consumers reading reviews when choosing a business97%BrightLocal
Consumers requiring a rating of at least 4.5 stars31%BrightLocal
Consumers unwilling to use a business with fewer than 20 reviews47%BrightLocal
Consumers prioritising reviews from the last three months74%BrightLocal

For a UK service business, keep the profile accurate, make services clear and ask for genuine reviews after real work. Do not turn a US survey into a claim about Bristol, Birmingham or any other UK market. Work on local SEO should connect profile data with the matching service page and a measurement plan.

What business data helps explain search demand?

The UK business and retail figures below put search demand in context. They come from official or named first-party publications and do not show that SEO caused a change in sales.

MeasureValueDateSource
Online retail sales as a share of Great Britain retail sales28.8%September 2026; ONS Monthly Business Survey, Retail Sales Inquiry; response rate 59.1%, turnover coverage 92.4%, respondent count not statedONS, 2026-10-23
Small businesses as a share of UK private-sector businesses99.18%Start of 2025; DBT estimate, 0 to 49 employees, total population 5,690,265 businessesDBT Business Population Estimates 2025
UK private-sector employment in small and medium-sized businesses16.9 millionStart of 2025; DBT estimate, SMEs with 0 to 249 employeesDBT Business Population Estimates 2025
US retail-site traffic from AI sources+1,219%October 2024 to July 2026; Adobe Analytics referral clicks to US retail sites, over 1 trillion visits analysed; not a survey sampleAdobe Digital Insights, 2026-08-19

The Adobe figure is US retail data, not a UK forecast. ONS and DBT describe the size and shape of the market, not the return from one channel.

How should a business use these statistics?

Start with the searches that already matter. Export queries and pages from Search Console, mark the country, intent and business outcome, then record whether the result includes an AI summary, local pack, featured snippet or another answer feature. Review a fixed sample each month because feature visibility changes by query and location.

  1. Choose a small set of commercial and informational queries. Keep the wording and location fixed.
  2. Check the ranking page and any answer feature, including its cited sources.
  3. Record impressions, clicks, enquiries, calls and assisted conversions. Do not treat a citation as a sale.
  4. Improve one evidence gap at a time: unclear definition, missing source, weak internal link, poor local detail or technical error.
  5. Repeat the check after a defined period and keep a dated change log.

For people-first content, compare the page against Google's guidance and the practical checks in our helpful content update guide. For pages produced at scale, use the programmatic SEO for ecommerce guide to test whether each URL adds distinct value. The workflow automation tools guide is relevant when reporting and monitoring need repeatable steps.

SEO, GEO and AEO evidence checklist

  • Name the publisher and link to the underlying report or dataset.
  • State the geography, sample and fieldwork period.
  • Define the metric. CTR, visibility, visits and citations are different measures.
  • Separate an observed result from a forecast or a vendor recommendation.
  • Keep US, UK, EU and global figures in separate rows.
  • Record the page, prompt or query set checked by your team.
  • Use a change log when a figure is replaced or a source updates its page.
  • Re-test the commercial outcome. A higher impression count is not automatically a higher lead count.

Limitations and what this guide does not cover

There is no universal AI ranking-factor list here. Google, ChatGPT, Perplexity and the rest use different retrieval and generation processes, and the studies use different units (a user panel, a keyword set, a prompt benchmark, a clickstream dataset, a survey) that cannot be combined into one number.

The UK figures here come mainly from Ofcom, ONS, DBT, Statcounter and UK-located reporting. The Pew, SparkToro, Datos, Ahrefs, BrightLocal and Adobe results are not UK benchmarks unless the row says so.

Some useful studies are older than 2026 because fieldwork and publication take time. The change log records replacements when a newer source measures the same metric with a clear method.

How Aristral approaches this

We start with the customer's search journey and the business action behind it. We check the technical foundation, page intent, internal links, local signals and the sources that appear around important questions, then build a dated baseline for rankings, impressions, clicks, enquiries and citations. A vendor benchmark can inform a test; the client's own query set decides it.

The same process covers sites that need both conventional SEO and AI search visibility: see our SEO services page. Where a business depends on maps, we add local work. CH Development came to us as a brand-new company and was ranking in its local map pack and booked to capacity within two months.

If you want to check where your pages appear and what to fix first, contact us with the site and the searches that matter.

Sources and change log

Figures were re-checked against primary sources on 23 and 24 October 2026. The next review follows the quarterly update process.

Frequently asked questions

How much have AI Overviews reduced Google click-through rates?
Pew's US panel recorded traditional-result clicks after 8% of Google result-page visits with an AI summary, compared with 15% without one, in March 2025. A click on a source link inside the summary occurred on 1% of summary-page visits. In a separate 300,000-keyword analysis, Ahrefs found a 58% lower average click-through rate for the top-ranking page when an AI Overview appeared. These study designs differ, so neither is a universal UK reduction.
What percentage of searches now end without a click?
SparkToro and Similarweb reported that 68.01% of US Google searches ended without a click in January to April 2026. Their country study found 69.5% for the UK over the same period (results). The browser clickstream panels do not disclose their sample sizes and exclude Google's mobile search app. Zero-click includes a session ending or another query.
How big is AI search in the UK?
Ofcom's 2026 survey found that 54% of UK adults used AI tools such as ChatGPT, Copilot or Gemini. It also found that 75% of online adults read AI search summaries at least sometimes and 42% read them often or always. The survey included 7,533 UK adults aged 16+ and was fielded 2025-09-29 to 2025-11-28. Ofcom's 2025 Online Nation report recorded 1.8 billion ChatGPT website visits in January to August 2025, compared with 368 million in the same period of 2024. Visits are not searches or unique people.
What is Generative Engine Optimisation and does it work?
Generative Engine Optimisation is a practical term for work intended to improve content visibility in generative engine responses. The GEO-bench paper reported tactic-level gains of about +41% for quotation addition, about +31% for statistics addition and about +27.5% for source citation on its Position-Adjusted Word Count metric. It tested 10,000 benchmark queries, with reported results on a 1,000-query test split averaged over five seeds. It was an experiment, not a guaranteed live-search result.
Does structured data increase click-through rate?
Google's structured-data documentation says markup helps Google understand page content and can make pages eligible for rich results. Google does not guarantee that structured data will appear or produce a CTR increase. Measure clicks on your own pages.
Are UK businesses investing more or less in SEO?
The sources in this guide do not provide a single comparable UK spending trend. ONS and DBT describe retail and business-population context, while Ofcom reports online behaviour. Treat any claim about a UK SEO budget increase as a separate survey finding. A business can assess its own position through search demand, qualified enquiries, conversion data and the cost of maintaining the work.

Methodology

I re-checked the figures against primary sources on 23 and 24 October 2026. The review covered geography, sample, fieldwork period, metric definition and whether a source was first-party, academic or an industry survey. UK and US figures remain separate. Readers should re-check live market-share readings, AI feature availability and vendor studies before making a decision. Older studies are labelled with their fieldwork date. The update process is quarterly: the owner checks new Pew, Ofcom, Google, Statcounter and vendor releases, records changed figures, then asks for a technical review. Technical review: Huzaifa Jan Asim, co-founder and CTO. Last reviewed 23 and 24 October 2026. Corrections: admin@aristral.com.

About the author

Taha Bilal

Co-founder, Aristral

Taha Bilal is a co-founder of Aristral, a UK AI automation and SEO agency based in Clifton, Bristol. He has been running SEO and digital-growth campaigns for SMB and SaaS clients since 2018, and now leads Aristral's combined SEO + GEO programmes for service businesses across the UK and US. Corrections and source requests: admin@aristral.com.

LinkedIn