AI Search and GEO Statistics for 2026: How People Use AI Search and How Engines Cite Sources

By Christoph Olivier | Published 2026-07-26 | Last updated 2026-07-26

This briefing collects verified statistics on how people use AI-powered search and how AI engines select and cite sources. Every figure below is traced to a named study with a working link. Each figure is labeled as measured data, self-reported company data, survey data, or forecast. Where studies disagree, the disagreement is stated. Vendor growth percentages measured off very small 2024 baselines are flagged, because triple-digit year-over-year growth can overstate absolute impact.

Executive summary

• Google AI Overviews appeared on roughly 6.5% of tracked queries in January 2025, peaked near 25% in July 2025, then settled below 16% by November 2025 (Semrush, 10 million keywords).
• When an AI summary appeared, US Google users clicked a traditional result in 8% of visits, versus 15% without a summary (Pew Research Center, March 2025, measured browsing of 900 US adults).
• Links inside AI summaries were clicked in only 1% of visits to pages that had a summary (Pew Research Center, March 2025).
• AI platforms generated about 1.13 billion referral visits in June 2025, up 357% year over year, but that is still under 1% of the roughly 191 billion referrals from Google search in the same panel (Similarweb, measured panel estimate).
• Generative AI traffic to US retail sites rose 1,200% comparing February 2025 with July 2024 (Adobe Analytics, measured across more than 1 trillion visits).
• Citation sources diverge sharply by engine: Wikipedia made up 47.9% of ChatGPT’s top-10 source share while Reddit made up 46.7% of Perplexity’s top-10 source share (Profound, 680 million citations, August 2024 to June 2025).
• Company-stated reach: Google said AI Overviews reach 2 billion monthly users and OpenAI said ChatGPT has 700 million weekly active users (2025, self-reported, not independently audited).
• Gartner’s February 2024 forecast that traditional search volume would fall 25% by 2026 had not clearly materialized by mid-2026, as Google retained a 90%-plus search share (forecast, flagged).

Key findings

1. 8% vs 15% click rate. US Google users clicked a traditional search result in 8% of visits with an AI summary present, against 15% without one, March 2025 (Pew Research Center, measured).
2. 1% in-summary click rate. Users clicked a link inside the AI summary in just 1% of visits to pages that contained a summary, March 2025, US (Pew Research Center, measured).
3. 58% of users met an AI summary. About 58% of tracked US adults ran at least one Google search that produced an AI summary during March 2025 (Pew Research Center, measured).
4. 18% of searches triggered a summary. 18% of all Google searches in the Pew dataset produced an AI summary, March 2025, US (Pew Research Center, measured).
5. 26% vs 16% session abandonment. Users ended their browsing session after 26% of pages with an AI summary, versus 16% of pages with only traditional results, March 2025, US (Pew Research Center, measured).
6. 88% of summaries cited 3 or more sources. 88% of AI summaries cited at least three sources and only 1% cited a single source, April 2025, US (Pew Research Center, measured).
7. 34.5% lower position-1 CTR. The presence of an AI Overview correlated with a 34.5% lower average click-through rate for the top organic result, comparing March 2024 with March 2025 across 300,000 keywords, US (Ahrefs, measured via Search Console data).
8. AI Overview reach swung from 6.5% to 25% to under 16%. AI Overviews triggered on 6.5% of tracked queries in January 2025, near 25% in July 2025, and under 16% in November 2025 (Semrush, 10 million keywords, measured).
9. 1.13 billion AI referral visits. AI platforms sent about 1.13 billion referral visits in June 2025, up 357% from June 2024, to the top 1,000 domains tracked (Similarweb, measured panel estimate).
10. ChatGPT dominates AI referrals. ChatGPT accounted for more than 80% of AI referrals to those top 1,000 domains in June 2025 (Similarweb, measured panel estimate).
11. 1,200% retail AI-traffic growth. Generative AI referral traffic to US retail sites rose 1,200% comparing February 2025 with July 2024 (Adobe Analytics, measured, flagged as off a near-zero base).
12. Wikipedia and Reddit dominate citations, but by different engines. Wikipedia was 47.9% of ChatGPT’s top-10 source share and Reddit was 46.7% of Perplexity’s top-10 source share, August 2024 to June 2025 (Profound, 680 million citations, measured).
13. 2 billion monthly AI Overview users. Google stated AI Overviews reach 2 billion monthly users, 2025 (self-reported, not audited).
14. 700 million ChatGPT weekly users. OpenAI stated ChatGPT reached 700 million weekly active users, 2025 (self-reported, not audited).
15. Gartner forecast: 25% search-volume decline by 2026. Gartner predicted in February 2024 that traditional search engine volume would drop 25% by 2026 (forecast, not yet borne out as of mid-2026).

How often AI Overviews appear

AI Overviews are Google’s AI-generated summaries placed above organic results. Their appearance rate was not stable in 2025. Semrush tracked more than 10 million keywords from January to November 2025 and found the appearance rate rose from 6.5% of queries in January to just under 25% in July, then fell back below 16% by November. Because the rate moved this much inside a single year, any single “X% of searches” claim is time-sensitive and should carry its month.

The intent mix of AI Overview queries also shifted. Semrush reported that informational queries made up 91.3% of AI Overview triggers in January 2025 but 57.1% by October 2025, meaning commercial, transactional, and navigational intents rose to 42.9%. Ads on AI Overview result pages grew 394% from March to October 2025, from 5.17% of such pages to 25.56%. Geography and device coverage were not specified in the Semrush report, so treat these as directional for US-plus-global desktop tracking.

Source: Semrush AI Overviews Study, 2025.

Click-through impact of AI Overviews

Two independent measurement approaches agree on direction, disagree on magnitude, and measure different things. Pew Research Center installed browsing trackers on 900 consenting US adults and recorded 68,879 Google searches during March 2025, of which 12,593 produced an AI summary. Users clicked a traditional result in 8% of visits with a summary, versus 15% without one, a drop of nearly half. They clicked a link inside the summary in only 1% of visits. Session abandonment rose from 16% to 26% when a summary was present.

Ahrefs took a different route, aggregating Google Search Console data for 300,000 keywords (150,000 with an AI Overview, 150,000 without) and comparing March 2024 with March 2025. It found the presence of an AI Overview correlated with a 34.5% lower position-1 click-through rate. The top-result CTR for AI Overview keywords fell from 0.073 to 0.026 over that year.

These are not the same statistic. Pew measures the share of all searches that produced any click on any result. Ahrefs measures the CTR of the single top organic position. Both point down. A later Ahrefs update, reported in December 2025 secondary coverage, put the position-1 reduction near 58%, but that specific figure is sourced here only through secondary reporting and is not treated as verified.

Sources: Pew Research Center, July 2025; Ahrefs, 2025.

AI search adoption and referral traffic

AI referral traffic is growing quickly from a small base. Similarweb, using its proprietary traffic panel, reported that AI platforms (ChatGPT, Gemini, DeepSeek, Grok, Perplexity, Claude, and Liner) generated over 1.13 billion referral visits in June 2025, up 357% from June 2024, to the top 1,000 domains it tracked. For scale, the same analysis counted about 191 billion referrals from Google search. AI referrals therefore equal roughly 0.6% of Google’s referral volume in that panel. ChatGPT alone accounted for more than 80% of AI referrals. Referrals to news and media sites grew fastest at 770% year over year.

Adobe Analytics, drawing on more than 1 trillion visits to US retail sites, reported that generative AI referral traffic to US retail rose 1,200% comparing February 2025 with July 2024, after a 1,300% jump during the 2024 holiday season and 1,950% on Cyber Monday 2024. Growth this large reflects a near-zero 2024 starting point, so the percentages describe trajectory, not absolute share.

On self-reported reach, Google stated that AI Overviews reach 2 billion monthly users and that Google AI Mode reached 100 million users in the US and India in 2025. OpenAI stated ChatGPT reached 700 million weekly active users in 2025. These are company statements, not independently audited figures, and are labeled as such.

Sources: Similarweb, 2025; Adobe Analytics, 2025; Google, 2025; OpenAI, 2025.

How AI shoppers behave

Adobe reported that visitors arriving from generative AI sources behaved differently from other US retail visitors: 8% higher engagement time, 12% more pages per visit, and a 23% lower bounce rate, though they were 9% less likely to convert. That conversion gap had narrowed from 43% in July 2024. In a paired survey of 5,000 US respondents, 39% said they had already used generative AI for online shopping and 53% planned to during the year. Among AI shoppers, 92% said it improved their experience. Survey figures are self-reported and should be read as stated intent, not observed behavior.

Source: Adobe Analytics, 2025.

How AI engines cite sources

Which sources an AI engine cites depends heavily on the engine. Profound analyzed 680 million citations across ChatGPT, Google AI Overviews, and Perplexity from August 2024 to June 2025. Wikipedia made up 47.9% of ChatGPT’s top-10 source share and 7.8% of ChatGPT’s overall citation volume. Reddit made up 46.7% of Perplexity’s top-10 source share and 6.6% of its overall volume. For Google AI Overviews, Reddit was 21.0% and YouTube 18.8% of the top-10 source share. Across all engines, .com domains carried more than 80% of citations and .org sites were second at 11.29%.

The practical read: a brand cited well by one engine can be absent from another, because ChatGPT leans encyclopedic while Perplexity leans toward community and forum content. Note that top-10 source share and overall citation volume measure different things and should not be conflated.

Source: Profound AI Platform Citation Patterns, 2025.

How our own data fits in

Our AI Search Readiness Study (July 2026) crawled 419 live US small-business websites across six verticals and ten cities. It found 22.4% of live sites effectively invisible to AI search (no structured data, no llms.txt, no FAQ content), a mean AI-Readiness Index of 59.7 out of 100, HTTPS present on 99.0% of sites against FAQPage schema on only 4.3%, and no stated position on OpenAI’s GPTBot on 91.2% of sites. That supply-side picture explains part of the citation concentration above: when few small sites expose answer-ready structured data, engines fall back on a small set of large, well-structured sources such as Wikipedia and Reddit.

Source: Christoph Olivier Consulting, AI Search Readiness Study, 2026.

Projected change in search behavior

Forecasts here are labeled as forecasts. In February 2024 Gartner predicted that traditional search engine volume would fall 25% by 2026 as AI chatbots and virtual agents absorbed queries. As of mid-2026 that scale of decline had not clearly appeared: Google retained a 90%-plus search market share and reported AI Overviews reaching 2 billion monthly users, which suggests AI changed the shape of the results page more than it cut total search volume. Separately, Semrush projected that AI search traffic may surpass traditional search traffic by 2028. Treat both as vendor and analyst projections, not measured outcomes.

Sources: Gartner, 2024; Semrush, 2025.

Original synthesis

1. The one-in-six click squeeze

Combining three measured studies: AI Overviews appeared on under 16% of tracked queries by November 2025 (Semrush), and when a summary appears the odds of any click roughly halve, from 15% to 8% (Pew), while the top organic position loses about 34.5% of its CTR (Ahrefs). Direction is consistent across a keyword-tracking panel, a browser panel, and Search Console aggregates. Logic: exposure times suppression gives a compounding effect on a growing minority of searches. Inputs: Semrush 10M-keyword study; Pew 900-adult browser panel; Ahrefs 300K-keyword GSC study. Limitations: three different populations, scopes, and geographies; the figures cannot be multiplied into a single national traffic-loss number.

2. AI referrals are loud in growth, quiet in volume

Similarweb’s own panel shows about 1.13 billion AI referral visits against roughly 191 billion Google referrals in June 2025. Dividing the two, AI referrals equal about 0.6% of Google’s referral volume even after 357% year-over-year growth. Logic: percentage growth and absolute share tell opposite stories, and both are true. Inputs: Similarweb June 2025 panel. Limitations: panel estimate, top 1,000 domains only, not a census; the ratio understates AI’s influence on high-consideration queries where it concentrates.

3. Citation concentration flips by engine

From Profound’s 680 million citations, the single most-cited source in an engine’s top-10 is Wikipedia for ChatGPT (47.9%) but Reddit for Perplexity (46.7%), with Google AI Overviews spreading across Reddit (21.0%) and YouTube (18.8%). Derived insight: AI citation is concentrated, but the concentration point is engine-specific, so a single “get cited by AI” strategy is unlikely to work across engines. Inputs: Profound Aug 2024 to June 2025 dataset. Limitations: top-10 share is not total volume; the study period predates mid-2026 model updates.

4. Forecast versus the record

Gartner’s 25%-decline-by-2026 forecast can be checked against Google’s own 2026 statements: a 90%-plus search share and AI Overviews at 2 billion monthly users. Derived insight: substitution reshaped the results page rather than collapsing query volume, at least through mid-2026. Inputs: Gartner Feb 2024 forecast; Google 2025 self-reported reach. Limitations: Google figures are self-reported; forecasts still have runway through end of 2026.

Methodology

Source selection favored primary and independent measurement: a nonpartisan research center (Pew), large-scale keyword and Search Console panels (Semrush, Ahrefs), a citation-analysis platform (Profound), a traffic panel (Similarweb), a first-party analytics dataset (Adobe), and official company statements (Google, OpenAI). Inclusion required a named study, a stated timeframe, and a working source link. Figures that could not be traced to a fetched named source were excluded, including a widely repeated 11% ChatGPT-Perplexity domain-overlap number that did not appear in the primary Profound page reviewed here, and an Ahrefs 58% figure available only through secondary coverage. Conflicts were handled by presenting each measurement with its scope rather than averaging incompatible methods. Every figure is labeled measured, self-reported, survey, or forecast. Growth percentages measured against near-zero 2024 baselines are flagged. Last updated 2026-07-26.

Source quality ranking

Tier 1 (primary, official bodies, first-party research): Pew Research Center (independent measured study); Google and OpenAI (official self-reported reach, labeled as such); Gartner (named analyst forecast); Christoph Olivier Consulting AI Search Readiness Study (first-party measured crawl); Adobe Analytics (first-party measured dataset of 1 trillion-plus visits).

Tier 2 (credible market research and platform data): Semrush (10M-keyword study); Ahrefs (300K-keyword Search Console study); Similarweb (traffic panel estimate); Profound (680M-citation analysis).

Tier 3 (reputable journalism and commentary used only for context): Search Engine Land and Digital Commerce 360 reporting that quotes the primary studies above.

Journalist-friendly section

Most quotable statistics

• US Google users clicked a result in 8% of searches with an AI summary, versus 15% without (Pew, March 2025).
• Links inside AI summaries were clicked in only 1% of visits (Pew, March 2025).
• AI Overviews swung from 6.5% of queries in January 2025 to nearly 25% in July before falling below 16% in November (Semrush).
• AI platforms sent 1.13 billion referral visits in June 2025, up 357% year over year, but only about 0.6% of Google’s referral volume (Similarweb).
• Wikipedia was 47.9% of ChatGPT’s top-10 sources; Reddit was 46.7% of Perplexity’s (Profound).

Data limitations

AI Overview appearance rates fluctuated within 2025, so single-point figures need their month. CTR studies use different scopes (all clicks in Pew, position-1 in Ahrefs) and cannot be combined. Similarweb and Adobe growth rates start from near-zero 2024 bases. Similarweb figures are panel estimates, not a census. Company reach figures from Google and OpenAI are self-reported. Profound top-10 shares are not total citation volume. Gartner and Semrush projections are forecasts.

Downloadable dataset: recommended fields

metric_name, exact_value, unit, year_or_range, geography, engine_or_platform, source_name, source_url, evidence_type (measured / self-reported / survey / forecast), sample_size, methodology_note, baseline_caveat.

Press summary (about 150 words)

AI search reshaped the results page in 2025 and 2026 without yet collapsing search volume. Pew Research Center measured US browsing and found that when Google shows an AI summary, users click a traditional result in only 8% of searches, down from 15% without one, and click links inside summaries just 1% of the time. Semrush tracked AI Overviews swinging from 6.5% of queries in January 2025 to nearly 25% in July before settling below 16%. AI referral traffic grew fast, 357% year over year to 1.13 billion visits in June 2025 per Similarweb, yet still equals under 1% of Google’s referral volume. Citation patterns differ by engine: Wikipedia dominates ChatGPT while Reddit dominates Perplexity, per Profound’s analysis of 680 million citations. Gartner’s forecast of a 25% search-volume drop by 2026 had not clearly materialized, as Google kept a 90%-plus market share.

Suggested headlines

1. Google AI Summaries Halve Click-Throughs, Pew Data Shows
2. AI Overviews Hit One in Four Searches at 2025 Peak, Then Pulled Back
3. AI Referral Traffic Grew 357% but Is Still Under 1% of Google
4. Wikipedia Rules ChatGPT, Reddit Rules Perplexity: Inside 680 Million AI Citations
5. Gartner Said Search Would Drop 25% by 2026. The Data Says Otherwise.

Frequently asked questions

See the FAQ section rendered from structured data below.