THE HOOKED DATA LIBRARY · 2026 EDITION
AI Search Statistics (2026): 42 Data Points on Adoption, Clicks and Citations
Google says AI Overviews reach more than 2.5 billion people each month. OpenAI reports more than 1 billion weekly ChatGPT users, while Pew finds that 60% of US adults have read AI search summaries. (Google, Investor Presentation, June 2026; OpenAI, Expanding Access to AI, August 2026; Pew Research Center, Americans and AI 2026)
This reference brings together 42 distinct data points from Google, OpenAI, Pew Research Center, Similarweb, Ahrefs and BrightLocal. It separates adoption, clicks, referrals and citations so readers can use the numbers without confusing their meaning. The 2026 edition includes dated 2025 evidence where useful; each table identifies the study period and population.
How to use this page: counts refer to different products and populations. Do not add them together. No figure here is presented as a Calgary benchmark or a promise of traffic.
Key takeaways
- Google’s AI Overviews exceed 2.5 billion monthly users (Google, Investor Presentation, June 2026).
- ChatGPT exceeds 1 billion weekly users; this includes non-search use (OpenAI, Expanding Access to AI, August 2026).
- 60% of US adults report reading AI search summaries (Pew Research Center, Americans and AI 2026).
- AI contributes 0.4% of retail ecommerce visits in one industry benchmark (Similarweb & Statista, State of Ecommerce, 2026).
- AI referral visits grew 117.4% year over year in Similarweb’s measured window (Similarweb, Generative AI Landscape / Referral Traffic, 2026).
- Adding schema produced no statistically clear uplift for ChatGPT or AI Mode in a matched study (Ahrefs, Schema and AI Citations Study, 2026).
- 45% report using generative AI for local recommendations (BrightLocal, Local Consumer Review Survey, 2026).
- Only 6% of people who encountered AI summaries reported trusting them a lot (Pew Research Center, Views of AI Summaries, 2025).
Definitions: an AI Overview is a summary within Google results; AI Mode is a conversational Google search experience. ChatGPT, Claude and Perplexity are assistants with different retrieval features. A mention names a business, a citation links or attributes a source, a referral is a visit, and a conversion is a separately defined action. This collection does not provide a standalone Claude adoption estimate.
1. AI search adoption and platform reach
Reach tells you whether a channel deserves attention; it does not tell you how much business it will generate. Google reports more than 2.5 billion monthly AI Overview users. Read each audience figure on its own terms: weekly chatbot use, monthly product use, and survey participation cannot be added into a single market size.
| Metric | Value | Source & scope |
|---|---|---|
| 1. AI Overviews monthly users | Over 2.5 billion | Google, Investor Presentation, June 2026June 2026 · worldwide |
| 2. AI Mode monthly users | Over 1 billion | Google, Investor Presentation, June 2026June 2026 · worldwide |
| 3. Gemini app monthly users | Over 900 million | Google, Investor Presentation, June 2026June 2026 · worldwide; all uses |
| 4. ChatGPT weekly users | Over 1 billion | OpenAI, Expanding Access to AI, August 2026August 2026 · worldwide; all uses |
| 5. US adults who have read AI search summaries | 60% | Pew Research Center, Americans and AI 2026February 2026 · US adults |
| 6. AI share of retail ecommerce visits | 0.4% | Similarweb & Statista, State of Ecommerce, 20262026 report · window unspecified publicly |
The final row measures visits, not people. Its narrower retail scope is deliberately labelled. Do not use it to calculate a global AI search market share or a forecast for a Calgary contractor. For a small business, the useful starting point is its own referral and enquiry baseline.
2. What people actually do with ChatGPT
A chatbot audience includes research, writing and many activities unrelated to buying. 49% of messages fell into the “Asking” category in OpenAI’s 2025 analysis. That makes clear explanations useful, but it does not mean every conversation represents a prospective customer.
| Metric | Value | Source & scope |
|---|---|---|
| 7. Messages classified as Asking | 49% | OpenAI, How People Are Using ChatGPT, 2025Consumer usage · 2025 study |
| 8. Messages classified as Doing | 40% | OpenAI, How People Are Using ChatGPT, 2025Consumer usage · 2025 study |
| 9. Messages classified as Expressing | 11% | OpenAI, How People Are Using ChatGPT, 2025Consumer usage · 2025 study |
| 10. Conversations about guidance, information or writing | About 75% | OpenAI, How People Are Using ChatGPT, 2025Consumer usage · 2025 summary |
| 11. Daily messages after six months versus signup | 50% more | OpenAI, How ChatGPT Adoption Has Expanded, 20262026 account-cohort analysis |
| 12. Distinct tasks tried after six months | 2× | OpenAI, How ChatGPT Adoption Has Expanded, 20262026 account-cohort analysis |
The first four rows are a dated description of consumer use, not a 2026 search-query census. The cohort findings describe increasing engagement with the product. Neither establishes which websites will be recommended. Our editorial recommendation is to answer the questions a buyer needs resolved: suitability, process, costs, evidence and limitations.
3. AI Overviews and clicks to websites
Visibility does not automatically produce a visit. Traditional-result clicks occurred on 8% of visits with an AI summary, versus 15% without one in Pew’s observed sample. A business should therefore track useful enquiries alongside rankings and impressions.
| Metric | Value | Source & scope |
|---|---|---|
| 13. Traditional-result click, summary present | 8% | Pew Research Center, Google AI Summary Click Study, 2025US browsing · March 2025 |
| 14. Traditional-result click, no summary | 15% | Pew Research Center, Google AI Summary Click Study, 2025US browsing · March 2025 |
| 15. Click on a source inside an AI summary | 1% | Pew Research Center, Google AI Summary Click Study, 2025US browsing · March 2025 |
| 16. Session ended after a summary page | 26% | Pew Research Center, Google AI Summary Click Study, 2025US browsing · March 2025 |
| 17. Session ended after a page without a summary | 16% | Pew Research Center, Google AI Summary Click Study, 2025US browsing · March 2025 |
| 18. Searches that produced an AI summary | 18% | Pew Research Center, Google AI Summary Click Study, 2025Study sample · March 2025 |
This is a 2025 snapshot, not a current universal click rate. Query differences prevent a causal conclusion. We recommend comparing similar pages and search intents over time rather than treating every traffic change as an AI effect.
4. AI referral traffic by industry
A growing channel can still account for a small part of a business’s acquisition. Similarweb estimates 770.7 million monthly AI referral visits across its worldwide measurement period. Category totals help describe the channel, but they are not a traffic target for an individual website.
| Metric | Value | Source & scope |
|---|---|---|
| 19. AI referrals, monthly average | 770.7 million | Similarweb, Generative AI Landscape / Referral Traffic, 2026Worldwide web · June 2025–May 2026 |
| 20. AI referrals, year-over-year growth | 117.4% | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
| 21. Marketplace referrals, monthly average | 46.8 million | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
| 22. News referrals, monthly average | 44.5 million | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
| 23. Travel referrals, monthly average | 44.5 million | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
| 24. Beauty referrals, year-over-year growth | 312.5% | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
Volume and growth answer different questions. A small starting base can produce a large percentage increase. Record landing pages, engaged visits and enquiries from each identifiable AI referrer, then compare with other channels. Keep unattributed visits separate instead of assuming they all came from AI.
5. Source citations and the limits of optimization claims
A citation is an opportunity to be seen, not proof of a click or sale. 88% of the AI summaries in Pew’s sample cited at least three sources. The schema study below asks a different question: whether adding markup changed citation frequency for pages already receiving citations.
| Metric | Value | Source & scope |
|---|---|---|
| 25. AI summaries citing at least three sources | 88% | Pew Research Center, Google AI Summary Click Study, 2025Google sample · 2025 |
| 26. Summary sources from Wikipedia, YouTube and Reddit combined | 15% | Pew Research Center, Google AI Summary Click Study, 2025Google sample · 2025 |
| 27. Summary sources on government domains | 6% | Pew Research Center, Google AI Summary Click Study, 2025Google sample · 2025 |
| 28. AI Overview citation change after adding schema | −4.6% | Ahrefs, Schema and AI Citations Study, 2026Relative to matched controls |
| 29. AI Mode citation change after adding schema | +2.4% | Ahrefs, Schema and AI Citations Study, 2026Not statistically distinguishable from zero |
| 30. ChatGPT citation change after adding schema | +2.2% | Ahrefs, Schema and AI Citations Study, 2026Not statistically distinguishable from zero |
Ahrefs found no clear citation uplift. Its small AI Overview decline should not be read as proof that schema harms visibility: other changes could contribute, and already-visible pages limit generalization. Google’s guidance supports ordinary SEO fundamentals rather than special AI markup. Read Google’s guidance. Our recommendation is to make evidence accessible and keep structured data accurate, not to sell a markup-only shortcut.
6. Trust, reviews and local recommendations
Being recommended is only part of winning a customer’s confidence. 45% reported using generative AI for local recommendations in BrightLocal’s 2026 survey. Trust and review findings offer context for what a useful business presence needs to communicate.
| Metric | Value | Source & scope |
|---|---|---|
| 31. At least some trust in AI summary information | 53% | Pew Research Center, Views of AI Summaries, 2025US summary users · August 2025 |
| 32. A lot of trust in AI summary information | 6% | Pew Research Center, Views of AI Summaries, 2025US summary users · August 2025 |
| 33. Little or no trust in AI summary information | 46% | Pew Research Center, Views of AI Summaries, 2025US summary users · August 2025 |
| 34. Use of generative AI for local recommendations | 45% | BrightLocal, Local Consumer Review Survey, 2026US consumers · 2026 survey |
| 35. Consumers reading local-business reviews | 97% | BrightLocal, Local Consumer Review Survey, 2026US consumers · 2026 survey |
| 36. Likely to visit a website after positive reviews | 54% | BrightLocal, Local Consumer Review Survey, 2026US consumers · 2026; multiple responses allowed |
These are separate studies, not one funnel. They do not establish a review count that causes an AI citation. Our practical advice is to keep your services, service area, contact details and client evidence consistent. The local SEO work behind that presence remains useful whether someone arrives from a search result, a recommendation or a direct visit.
7. Differences across ages and audiences
An overall adoption figure can hide differences in who uses a tool. 72% of US adults aged 18–29 reported reading AI search summaries, compared with a smaller share among older adults. Avoid assuming that a national average describes your own customer base.
| Metric | Value | Source & scope |
|---|---|---|
| 37. Read AI search summaries: ages 18–29 | 72% | Pew Research Center, Americans and AI 2026US adults · February 2026 |
| 38. Read AI search summaries: ages 30–49 | 68% | Pew Research Center, Americans and AI 2026US adults · February 2026 |
| 39. Read AI search summaries: ages 50–64 | 59% | Pew Research Center, Americans and AI 2026US adults · February 2026 |
| 40. Read AI search summaries: ages 65+ | 38% | Pew Research Center, Americans and AI 2026US adults · February 2026 |
| 41. Feminine-name share among classifiable ChatGPT users | 52% | OpenAI, How People Are Using ChatGPT, 2025July 2025; name-based proxy, not gender census |
| 42. Adoption growth: lowest- versus highest-income countries | Over 4× | OpenAI, How People Are Using ChatGPT, 2025By May 2025; growth comparison, not penetration |
Country-level growth and name classification are imperfect audience indicators. For campaign decisions, ask actual customers how they discovered you and record the answer alongside analytics. This is especially important when applying US survey results to a Canadian business.
Summary: AI search by the numbers
A selection of the figures above for quick reference. These repeat existing statistics and are not counted again in the headline total.
| Metric | Value | Source & scope |
|---|---|---|
| AI Overviews monthly users | Over 2.5 billion | Google, Investor Presentation, June 2026June 2026 · worldwide |
| AI Mode monthly users | Over 1 billion | Google, Investor Presentation, June 2026June 2026 · worldwide |
| Gemini app monthly users | Over 900 million | Google, Investor Presentation, June 2026June 2026 · worldwide; all uses |
| ChatGPT weekly users | Over 1 billion | OpenAI, Expanding Access to AI, August 2026August 2026 · worldwide; all uses |
| US adults who have read AI search summaries | 60% | Pew Research Center, Americans and AI 2026February 2026 · US adults |
| AI share of retail ecommerce visits | 0.4% | Similarweb & Statista, State of Ecommerce, 20262026 report · window unspecified publicly |
| Messages classified as Asking | 49% | OpenAI, How People Are Using ChatGPT, 2025Consumer usage · 2025 study |
| Messages classified as Doing | 40% | OpenAI, How People Are Using ChatGPT, 2025Consumer usage · 2025 study |
| Daily messages after six months versus signup | 50% more | OpenAI, How ChatGPT Adoption Has Expanded, 20262026 account-cohort analysis |
| Distinct tasks tried after six months | 2× | OpenAI, How ChatGPT Adoption Has Expanded, 20262026 account-cohort analysis |
| AI referrals, monthly average | 770.7 million | Similarweb, Generative AI Landscape / Referral Traffic, 2026Worldwide web · June 2025–May 2026 |
| AI referrals, year-over-year growth | 117.4% | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
| Marketplace referrals, monthly average | 46.8 million | Similarweb, Generative AI Landscape / Referral Traffic, 2026Same measurement window |
| AI Overview citation change after adding schema | −4.6% | Ahrefs, Schema and AI Citations Study, 2026Relative to matched controls |
| ChatGPT citation change after adding schema | +2.2% | Ahrefs, Schema and AI Citations Study, 2026Not statistically distinguishable from zero |
| Use of generative AI for local recommendations | 45% | BrightLocal, Local Consumer Review Survey, 2026US consumers · 2026 survey |
| Read AI search summaries: ages 18–29 | 72% | Pew Research Center, Americans and AI 2026US adults · February 2026 |
What these findings mean for a local business
Our interpretation is to treat AI discovery as part of a broader customer journey. Start by identifying which questions potential customers ask before they call. A useful page should make its subject, audience, location and evidence easy for a person to understand.
Keep three separate records: where your business is mentioned, which visits reach your website, and which enquiries become suitable work. A screenshot of an AI recommendation is evidence of that response at that moment. It is not a ranking guarantee or an attribution report.
For a service business, we recommend a monthly review of relevant landing pages, identifiable referral sources, qualified calls and forms, and customer-reported discovery. Use consistent definitions so changes mean something. If a report says “conversion,” confirm whether it means a click, a form, a qualified lead or a completed sale.
A clear service website can explain who you help and make the next step straightforward. Our SEO versus Google Ads guide helps put the longer-term discovery work alongside paid acquisition. The studies on this page do not establish a universal budget split.
Our recommendation is to improve a page when it has a useful answer missing, weak evidence, confusing navigation or an unclear offer. Publishing a large volume of near-identical answers simply to chase AI exposure is not the goal. Google’s official guidance continues to emphasize useful content and technical clarity.
Methodology and sources
Last updated: September 28, 2026. This is a secondary synthesis of original research and provider disclosures, not a Hooked survey. The 42 numbered table rows are unique retained data points; summary rows, chart labels and takeaways are duplicates and excluded from that count.
We screened 64 candidate statistics and retained 42. The collection uses 11 original research/provider publications plus one official guidance document. No statistic is sourced to a blog quoting another organization’s research. Vendor research is included when the vendor is reporting its own dataset; that does not make it independent of commercial interests.
- Dates: 24 retained rows come from publications released in 2026, some using 2025–2026 observations. The remaining 18 are explicitly dated 2025 findings. A publication date is not necessarily its measurement date.
- Geography: US surveys, worldwide web estimates and product disclosures stay separate. No Canadian or Calgary estimate was manufactured from them.
- Comparability: weekly and monthly users are not comparable market shares. Retail visit share and global referral growth have different denominators. We do not combine them.
- Market size: no dollar valuation or forecast is included; we found no need to equate the AI software market with AI search. This avoids combining incompatible market definitions.
- Causality: observed differences and matched estimates do not prove that a single feature caused a business outcome.
- Recency: no retained source is more than three years old. The 2025 behavioural studies are historical snapshots, not claims about today’s interface. New click-behaviour and demographic releases should be checked at the next review.
- Update policy: quarterly editorial review, with corrections when identified. The date above records this review; future updates require a fresh source check.
Source register
- Google, Investor Presentation, June 2026
Company-reported worldwide monthly users; products overlap. - OpenAI, Expanding Access to AI, August 2026
Company-reported weekly ChatGPT users; not search-only usage. - Pew Research Center, Americans and AI 2026
US adult survey, February 17–23, 2026; self-reported reading. - OpenAI, How People Are Using ChatGPT, 2025
Privacy-preserving consumer-plan research; 1.5 million conversations. Category definitions and denominators differ. - OpenAI, How ChatGPT Adoption Has Expanded, 2026
0.1% account sample; accounts created October 15, 2025–May 1, 2026; activity through May 31, 2026. - Pew Research Center, Google AI Summary Click Study, 2025
900 US adults; March 2025 browsing, 68,879 searches; results retrieved April 2025. Observational, not causal. - Similarweb, Generative AI Landscape / Referral Traffic, 2026
Modelled worldwide desktop/mobile-web referrals, June 2025–May 2026; not app-inclusive audience counts. - Similarweb & Statista, State of Ecommerce, 2026
Retail ecommerce headline benchmark; the public summary does not fully specify the observation window. Context only. - Ahrefs, Schema and AI Citations Study, 2026
1,885 treated pages and 4,000 controls; August 2025–March 2026 additions. Already-cited pages; matched observational analysis. - Pew Research Center, Views of AI Summaries, 2025
5,153 US adults surveyed August 18–24, 2025. Trust figures concern people who encountered summaries. - BrightLocal, Local Consumer Review Survey, 2026
Representative panel of 1,002 US adults; self-reported local-business behaviour. - Google Search Central, Generative AI Optimization Guide
Official guidance, checked September 28, 2026; not a statistical study.
How to cite this resource
Hooked Marketing. “AI Search Statistics (2026): 42 Data Points on Adoption, Clicks and Citations.” Updated September 28, 2026. When quoting a number, also credit and link its original study, including the date and population.
Browse the digital marketing statistics hub for related research. To discuss the practical implications for your business or report a source correction, contact Hooked Marketing.