30 AI search statistics for 2026

AI search is the new front page. Instead of ten blue links, a buyer types a question into ChatGPT, Perplexity, or Google's AI Overviews and reads a single synthesized answer that names a few brands. The statistics below show how fast that shift happened, how much it has drained the click economy, and why being named in the answer is now the unit of visibility. Every figure is sourced; pair them with our deeper State of AI Search report.

The numbers in one minute

AI search has reached mass scale, the click economy has contracted, and the brands an engine names are increasingly the only ones a buyer ever sees. The seven figures below anchor the rest of this page.

• ChatGPT hit 800 million weekly active users in October 2025 and 900 million by February 2026, per TechCrunch.
• Google AI Overviews serve more than 2 billion monthly users across 200+ countries, per Alphabet's Q2 2025 earnings.
• Users click a link just 8% of the time when an AI summary appears, versus 15% without one, per Pew Research.
• About 60% of searches now end click-free and 80% of users lean on AI summaries at least 40% of the time, per Bain & Company.
• Gartner projects a 25% drop in search volume by 2026 as buyers shift to AI assistants, per Gartner.
• AI referral traffic to US retail jumped 1,200% between July 2024 and February 2025, per Adobe Analytics.
• Engines cite different sources: ChatGPT and Google AI Overviews share only 21% of cited domains, per SE Ranking.

How big is AI search in 2026?

AI search reached mass scale in under two years. The four largest answer engines now reach a combined audience in the billions, and adoption is still compounding quarter over quarter rather than leveling off.

1. ChatGPT reached 800 million weekly active users

Sam Altman announced 800 million weekly active users at OpenAI's DevDay in October 2025 — double the 400 million reported in February 2025. That doubling inside eight months reflects how quickly a chat box becomes a default research habit once people trust it once. For a brand, the number sets the stakes: this is the front door where buyers now form first impressions of a category. A product that ChatGPT never names is invisible to a pool that already rivals a major search engine in reach. The same milestone climbed again to 900 million by February 2026, so the audience competing for citation keeps widening. Source: TechCrunch, 2025.

2. ChatGPT passed 900 million weekly users and 50 million subscribers

By February 2026, OpenAI reported ChatGPT had grown to 900 million weekly active users and 50 million paying subscribers, its largest milestone to date. The jump from 800 million in October 2025 shows growth had not plateaued; it kept compounding quarter over quarter. The 50 million subscribers matter as much as the audience size: paying users lean on the tool for work and buying decisions, not casual curiosity. Those are exactly the high-intent moments where a named brand wins or loses a deal. When an engine this large synthesizes one answer, the brands it omits never enter the buyer's consideration set. Source: TechCrunch, 2026.

3. Google AI Overviews reach more than 2 billion monthly users

On Alphabet's Q2 2025 earnings call, Sundar Pichai said AI Overviews had surpassed 2 billion monthly users across more than 200 countries and territories. Unlike ChatGPT, this reach is not opt-in — the summary appears above the links whether or not the user asked for it, so most of those 2 billion meet AI answers by default. That makes Google's own results page the single largest surface where a brand is either named or skipped. The scale dwarfs ChatGPT's 900 million weekly users, yet it cites a largely different set of sources, so winning one engine does not win the other. For a marketer, the implication is that AI visibility has to be tracked on Google's surface specifically, not assumed from organic rankings. Source: TechCrunch, 2025.

4. Google's AI Mode passed 100 million monthly users

The more conversational AI Mode reached over 100 million monthly active users across the US and India within months of launch. AI Mode replaces the link page entirely with a chat-style exchange, so the buyer follows up and refines rather than scanning ten results. That format gives the engine even more room to name a short list of brands and drop everyone else. Reaching 100 million in just two markets, while AI Overviews already serve 2 billion globally, shows how fast Google is moving users from passive summaries to active dialogue. For brands, it means the same citation question now applies across several Google surfaces at once, each with its own behavior. Source: TechCrunch, 2025.

5. The Gemini app surpassed 750 million monthly users

Google's Gemini app crossed 750 million monthly active users in Q4 2025, up from 650 million the prior quarter. The 100 million added in a single quarter mirrors the same compounding curve ChatGPT showed on its way from 800 to 900 million. Gemini matters because it is a standalone destination, separate from the AI Overviews that ride on top of search results — a second Google surface where the same buyer may ask the same question. A brand cited in Overviews is not automatically cited in the Gemini app, since the two draw on different contexts. Tracking both, alongside ChatGPT and Perplexity, is what turns scattered numbers into an actual visibility picture. Source: TechCrunch, 2026.

6. Perplexity processed 780 million queries in a single month

Perplexity handled 780 million queries in May 2025, with CEO Aravind Srinivas citing more than 20% month-over-month growth. At more than 20% growth a month, query volume roughly doubles in well under a year, which is why a smaller engine still deserves attention. Perplexity also leans harder on visible citations than most rivals, so the sources it names get unusual prominence with high-intent researchers. That makes it a place where a single well-structured page can earn outsized visibility before competitors notice. Its query count sits in the same range as the Gemini app's 750 million monthly users, confirming that the answer market is no longer a one-engine story. Source: Perplexity (reported), 2025.

7. AI Overviews now appear on roughly 16% of queries

After spiking from 6.49% in January 2025 to nearly 25% mid-year, Google AI Overviews settled around 16% of all queries by late 2025, per a 10-million-keyword study. The swing from 6.49% to nearly 25% and back to 16% shows Google still tuning where it trusts an AI answer, often informational and how-to queries rather than navigational ones. A brand cannot assume an Overview will or will not appear; coverage shifts by query type and over time. Since the same source measured Overviews answering only 58% of tested queries, roughly 16% sitewide means a large, moving slice of demand now meets a synthesis first. The practical takeaway is to monitor which of your category's questions trigger an Overview, because that set changes month to month. Source: Semrush, 2025.

How has AI search changed buyer behavior?

The headline behavior change is fewer clicks. When an engine answers the question on the page, most people never visit a source — they read the synthesis and move on, which compresses the entire downstream funnel.

8. Users click a link only 8% of the time when an AI summary appears

Pew Research analyzed 68,879 real searches and found users clicked a traditional result just 8% of the time when an AI summary was present, versus 15% when it was not. The mechanism is simple: when the answer sits at the top of the page, the question is already resolved, so there is little reason to click through. That 8% versus 15% gap means an AI summary cuts the chance of any click roughly in half. For a brand that relied on ranking to win traffic, the summary now intercepts the visit before the link is ever seen. The only reliable way back into that moment is to be one of the sources the summary names, which is why citation has replaced ranking as the goal. Source: Pew Research Center, 2025.

9. Only 1% of users click a link inside the AI summary

In the same Pew study, users clicked a source link within the AI-generated summary itself just 1% of the time — citation rarely converts into a visit. This reframes what a citation is worth: the value is the mention seen inside the answer, not the click that almost never follows. Against the 8% who still click a traditional result, that 1% confirms the source link is a footnote, not a traffic channel. So the goal is brand presence in the synthesized text — being named, described, and recommended — rather than a backlink that drives a session. A brand measuring only referral clicks from AI will badly undercount the influence it is actually having on the buyer. Source: Pew Research Center, 2025.

10. People end their session 26% of the time after an AI summary

Pew found users ended their browsing session 26% of the time after seeing an AI summary, compared with 16% on result pages without one. Ending the session is the strongest signal that the answer satisfied the need; the buyer got what they came for and stopped looking. That 26% versus 16% gap means the summary closes the journey outright far more often, not just shifts where the click lands. The brand named in that final answer is the one the buyer carries forward, while the rest are forgotten before any comparison happens. Combined with the 8% click rate, it shows the AI answer is increasingly the whole interaction, not a step toward a website. Source: Pew Research Center, 2025.

11. 58.5% of US Google searches are zero-click

SparkToro and Datos found 58.5% of US Google searches ended without a click to the open web — a majority, even before the steepest AI Overview growth. Zero-click was already the norm because Google answered many queries with its own panels, snippets, and maps long before AI summaries arrived. AI Overviews did not create the problem; they accelerated a majority that already existed in 2024. The mechanism matters for brands: more than half of all demand was already resolving inside Google's surfaces, so optimizing only for outbound clicks chased a shrinking share. As Overviews expand on top of this baseline, the open web's slice gets thinner still, pushing visibility toward being named on Google's own page. Source: SparkToro + Datos, 2024.

12. Only 360 of every 1,000 searches reach the open web

For every 1,000 US Google searches, just 360 clicks land on a non-Google website, the same study found — the rest stay inside Google's surfaces. This is the 58.5% zero-click figure restated as raw flow, and seeing it as 360 of 1,000 makes the leakage concrete. Nearly two-thirds of search intent now resolves without ever reaching an independent site, whether through Google's own answers or abandoned sessions. For a brand, that means the open web is a minority channel for capturing demand, and shrinking. The reachable audience increasingly lives inside the answer itself, so the work shifts from earning the click to earning the mention that shapes the buyer before any click is possible. Source: SparkToro + Datos, 2024.

13. 80% of users rely on AI summaries; 60% of searches end click-free

Bain & Company found about 80% of search users rely on AI-written summaries for at least 40% of their searches, and roughly 60% of searches now end without the user moving to another site. The 80% figure shows the behavior is mainstream, not an early-adopter habit; most people now treat the summary as a normal first read. Bain's 60% zero-click rate also sits just above the 58.5% SparkToro measured a year earlier, confirming the trend is climbing, not flattening. The mechanism is trust through repetition: once a summary is right often enough, users stop verifying and start acting on it. For brands, that means the summary is where the decision narrows, so being absent from it removes you from most buyers' shortlist before they ever compare options. Source: Bain & Company, 2025.

14. Gartner projects a 25% drop in search volume by 2026

Gartner predicts traditional search-engine volume will fall 25% by 2026 as buyers shift to AI chatbots and virtual agents. The driver is substitution: every question answered in ChatGPT or an Overview is a query that never gets typed into a classic search box. This projection lines up with the adoption curves elsewhere on this page, where ChatGPT alone passed 900 million weekly users and AI Overviews reached 2 billion monthly. A 25% drop does not mean demand vanished; it moved into surfaces that return one answer instead of ten links. For brands, that reroutes a quarter of search intent into a format where being named is the only form of visibility, so SEO budgets aimed only at rankings now address a shrinking pool. Source: Gartner, 2024.

15. 59% of consumers already use generative AI to shop

A consumer survey found 59% of respondents use generative AI tools for shopping tasks such as researching products and comparing options. This moves AI out of trivia and into commercial decisions, where a named brand directly shapes what gets bought. Researching and comparing is the exact stage where a shortlist forms, so the products the AI surfaces become the products under consideration. With 59% of consumers already doing this, the buying funnel now starts inside an answer for the majority, not a minority. That mirrors the trust gap elsewhere on this page: people shop with AI even though only about a third place high trust in it, which means accurate brand information matters even more. A brand the model describes wrongly, or omits, loses the sale before a human ever weighs in. Source: Omnisend / PR Newswire, 2025.

16. 45% of B2B buyers used generative AI for vendor research

Gartner found 45% of B2B buyers used generative AI in their purchase process, primarily to gather information on vendors and products. B2B deals are considered, multi-stakeholder purchases, so an AI-shaped vendor list carries real weight long before a sales call. The mechanism is shortlisting: buyers ask the model which vendors fit, and the names it returns frame the entire evaluation. At 45%, this is no longer a fringe behavior even in slow-moving enterprise buying — close to half of buyers let an engine seed their options. That tracks with the 59% of consumers shopping with AI, showing the shift spans both consumer and B2B demand. A vendor the AI fails to name simply never enters the running, regardless of how strong its product is. Source: Gartner, 2026.

When an AI summary appears, only 8 in 100 searches still send a click to the open web — and just 1 in 100 follows the source.

Do people trust AI answers?

Adoption has outrun trust. Most people now encounter AI answers regularly, but a large share remains skeptical of their accuracy — and acts on them anyway. That gap is exactly why an accurate, well-cited brand presence matters.

17. 34% of US adults have used ChatGPT — double the 2023 share

Pew found 34% of US adults have used ChatGPT as of early 2025, roughly twice the share two years earlier. Doubling in two years is the adoption curve of a tool crossing from novelty into routine, and ChatGPT's jump from 800 to 900 million weekly users shows the same momentum globally. The mechanism behind a slower US figure is that this counts all adults, including those who rarely search online; the 58% among under-30s reveals where the habit is densest. For a brand, 34% of all US adults is already a large enough base that being absent from ChatGPT's answers means missing a third of the adult population at the source. As that share keeps climbing, the cost of invisibility in AI answers compounds. Source: Pew Research Center, 2025.

18. 58% of adults under 30 have used ChatGPT

Among US adults under 30, 58% have used ChatGPT — up from 43% in 2024 and 33% in 2023 — signaling where buyer behavior is heading. The steady climb from 33% to 43% to 58% is a leading indicator: younger buyers adopt first, then their habits become everyone's baseline. This cohort runs well ahead of the 34% across all adults, so the overall number is being pulled up by the people who will dominate spending over the next decade. The mechanism matters for brands planning beyond the next quarter: today's under-30s default to asking an engine rather than scanning links. Building citation strength now means being established in the answer before this majority-AI generation becomes the core market. Source: Pew Research Center, 2025.

19. About half of users say they don't trust AI search results

Survey data shows roughly half of consumers are skeptical of AI-generated search answers, even as they use them — trust is the lagging indicator of adoption. The tension is that people lean on the summary for convenience while doubting its accuracy, then act on it anyway because checking every claim is too slow. That gap is an opening, not just a caveat: when the AI cites a recognizable, well-regarded brand, it borrows that brand's credibility to close the skeptic's doubt. Against the 59% of consumers already shopping with AI, the skepticism does not stop the behavior; it just raises the value of being a trusted name in the answer. A brand that earns accurate, frequent citation effectively becomes the reassurance a doubtful user is looking for. Source: YouGov, 2025.

20. Only about a third of Americans place high trust in AI answers

Only about one-third of Americans rate their trust in AI search results an 8 or higher on a 10-point scale, leaving most users in the skeptical middle. The striking part is the mismatch: trust sits near the same one-third level as the 34% of adults who have used ChatGPT, even though usage now drives real shopping decisions. Most users are neither true believers nor outright rejecters; they are persuadable and watching for signals of credibility. That middle is exactly where brand recognition tips the balance, because a familiar name in the answer reads as a vote of confidence. For marketers, the lesson is that being cited is necessary but not sufficient — being cited as a credible, well-described option is what converts the skeptical majority. Source: YouGov, 2025.

What is the business impact of AI search?

AI search both gives and takes. It is draining click-through from traditional results while sending a fast-growing, unusually high-intent stream of referral traffic to the brands the engines do name.

21. AI referral traffic to US retail jumped 1,200%

Adobe Analytics found traffic to US retail sites from generative AI sources rose 1,200% from July 2024 to February 2025. A 1,200% jump in eight months is the signature of a channel growing off a small base, so the absolute volume is still modest even as the rate is dramatic. The mechanism is the same adoption curve driving ChatGPT past 900 million users; as more buyers research in AI, the trickle of referral clicks that does escape grows fast. For brands, the takeaway is timing: this channel is small enough to be ignored and growing fast enough that early citation strength compounds. Waiting until AI referrals are large means competing for the mention after rivals have already become the default answer. Source: Adobe Analytics, 2025.

22. AI-referred visitors bounce 23% less and browse 12% more

Adobe found visitors arriving from generative AI had a 23% lower bounce rate, viewed 12% more pages, and spent 8% more time on site than other visitors. The reason is pre-qualification: by the time the AI sends someone to a site, it has already answered the basic questions and recommended the brand, so the visitor arrives further along. Lower bounce, more pages, and longer sessions all point to a warmer, more committed click than a cold search result. This is why the 1,200% growth matters more than its small base — the few visitors who do come through convert better. It also explains the 4.4x value figure elsewhere on this page: AI referrals are fewer but qualitatively stronger, so the citation that produced them is worth fighting for. Source: Adobe Analytics, 2025.

23. An AI-search visitor is 4.4x as valuable as an organic one

Semrush calculated the average AI-search visitor is 4.4 times as valuable as a traditional organic visit, measured by conversion rate. The mechanism is intent concentration: the AI has already filtered and recommended, so the person who clicks through is closer to buying than a generic searcher. This figure puts a price on the citation — each AI-referred visit converts at several times the rate of an ordinary organic one, which reframes the lost click economy. It also reconciles two numbers on this page: clicks from AI are rare (just 8% of searches send one, 1% from inside the summary), yet the few that land are 4.4 times as valuable. So the channel trades volume for quality, which means a brand should optimize for being the recommended answer rather than chasing raw traffic. Source: Semrush, 2025.

24. AI Overviews cut the #1 result's click-through 58%

Ahrefs analyzed 300,000 keywords and found the presence of an AI Overview correlated with a 58% lower average click-through rate for the top-ranked page. The position that used to capture the most traffic is now the one the Overview undercuts most, because the summary sits above it and answers first. This is the same dynamic Pew measured at the user level — an 8% click rate with a summary versus 15% without — seen here from the website's side as a 58% drop for the rank-one page. The hardest-won SEO position no longer guarantees the visit it once did. For brands, it means a top ranking and an Overview citation are now separate prizes: you can hold number one and still lose most of the clicks unless you are also named in the answer above it. Source: Ahrefs, 2025.

25. The zero-click shift is cutting organic traffic 15–25%

Bain estimates the move to AI summaries and zero-click answers is reducing organic web traffic by 15% to 25% across the sites it studied. This is the aggregate, site-wide version of the page-level damage Ahrefs found at the top result; spread across a whole site, the loss lands in the 15% to 25% range. The mechanism compounds: more queries trigger summaries, more summaries end the session, and fewer clicks reach the open web overall. Tellingly, this 15% to 25% drop sits in the same band as Gartner's projected 25% fall in search volume, two different methods pointing to a similar contraction. For brands, the figure sets the size of the gap to be filled — the traffic ranking no longer delivers has to be recovered through citation inside the answers that replaced it. Source: Bain & Company, 2025.

How many sources does an AI answer cite?

The citation surface is both narrow and engine-specific. Each answer names a handful of sources, and which sources differ so much between engines that winning one tells you little about the others. We unpack the patterns in How AI Engines Choose What to Cite.

26. ChatGPT cites 10.42 links per answer; Bing Copilot just 3.13

Across 2,000 keywords, SE Ranking found ChatGPT cited an average of 10.42 links per answer, Google AI Overviews 9.26, Perplexity 5.01, and Bing Copilot 3.13. The number of citation slots sets the odds of being named, and they vary more than threefold across engines. On Bing Copilot, with about 3 slots, the competition is fierce and only the strongest sources make it; on ChatGPT, with around 10, there is more room but also more rivals to outrank. The mechanism is each engine's tolerance for listing sources, which shapes how concentrated visibility is. For a brand, this means the same content can win citation on one engine and miss on another purely because of how many slots exist. It is the first reason visibility must be tracked engine by engine rather than as one score. Source: SE Ranking, 2025.

27. Google AI Overviews answered only 58% of tested queries

Google AI Overviews returned an answer for just 58.15% of the 2,000 tested queries — the lowest coverage of the engines studied, leaving the rest to classic results. The mechanism is selective trust: Google withholds an Overview where it judges a synthesized answer risky or unhelpful, so coverage is uneven by query type. This pairs with the roughly 16% of all queries that trigger an Overview sitewide; on the queries Google does test, more than four in ten still fall back to ordinary links. For brands, that means the playbook is split — on covered queries the goal is citation in the summary, while on the rest classic ranking still earns the click. Knowing which of your category's questions get an Overview is what decides where to aim, and that mix shifts over time. Source: SE Ranking, 2025.

28. ChatGPT and Google AI Overviews share only 21% of cited domains

The same study found ChatGPT and Google AI Overviews overlapped on just 21.26% of cited domains; Perplexity and ChatGPT shared 25.19% — different engines surface largely different sources. With roughly four in five cited domains differing between ChatGPT and Google, each engine effectively runs its own ranking system on its own index. The mechanism is that engines pull from different underlying data and apply different selection rules, so a source that satisfies one may be invisible to another. This is the strongest single argument against a one-number view of AI visibility: leading on ChatGPT tells you almost nothing about where you stand on Google AI Overviews. Combined with the varying citation counts above, it means a brand must measure and earn its place engine by engine, then close the specific gaps each one reveals. Source: SE Ranking, 2025.

29. News zero-click rates rose from 56% to 69% in a year

In the news segment, the share of searches ending without a click climbed from 56% to 69% within twelve months of the US AI Overviews launch. News is a leading case because its facts compress neatly into a summary, so the engine can satisfy the query without sending anyone onward. That a single year pushed zero-click from 56% to 69% shows how fast a category can tip once summaries cover it well. The starting point already sat near the 58.5% baseline SparkToro measured for all Google searches, so news ran ahead of the field and accelerated. The warning for other categories is directional: as summaries get better at a topic, its click-free share rises the same way, and brands relying on those clicks should expect the floor to keep moving. Source: Similarweb (reported), 2025.

30. The right content can lift AI visibility up to 40%

The Princeton and Georgia Tech "GEO" study found that optimizing a page for generative engines — with statistics, quotations, and cited sources — lifted its visibility in AI answers by up to 40%. The mechanism is that models favor content they can verify and quote, so pages dense with sourced facts and clear attributions are easier to lift into an answer. This is the optimistic counterweight to every loss figure on this page: visibility in AI answers is not fixed, it responds to how content is written. Against the 58% click drop at the top result and the 15% to 25% organic decline, a controllable 40% lift in citation gives brands a lever to pull rather than a trend to absorb. The "so what" is direct — the same statistics-and-citations structure this article uses is what the study found earns the mention, so the path forward is to build pages engines can confidently quote. Source: Aggarwal et al., KDD 2024. We break the levers down in the GEO Playbook.

What this means for brands

AI search did not just add a channel — it changed the unit of victory. When the engine answers the question, the brands it names are often the only ones a buyer considers, and the brands it omits never get a chance.

• Hundreds of millions of buyers now start in an AI answer, so being absent there is being absent at the start of the journey.
• Clicks to the open web are falling 15–25%, so traffic you used to win from rankings has to be replaced by citation inside the answer.
• Each engine cites different sources, so visibility has to be measured engine by engine, not as a single number.
• The traffic AI does send converts unusually well — a reason to compete for the citation, not just lament the lost click.

The practical next step is to see how you actually appear across these engines today. Mentionova runs your category's buying questions across six engines and shows where you are named, cited, or missing.