AI Search Statistics 2026: Adoption, Behavior & Market Share

You've been asked for a number. Maybe your board wants to know how many people actually use ChatGPT instead of Google. Maybe a client saw a headline about AI killing search and wants to know if their traffic is next. So you went looking for AI search statistics — and found a swamp of recycled figures, dead links and numbers that flatly contradict each other.

This page is the cleanup. Every statistic below links to its primary source — the earnings call, the research report, the live dashboard — not a blog quoting a blog quoting a tweet. Where a number is older or carries a methodology catch, it says so in the same sentence.

Why should you care? Because if you're responsible for a brand's visibility — as a marketing lead or an agency — these numbers decide where your next year of budget goes. Read them wrong and you optimize for a search behavior that no longer exists.

One principle sits underneath everything you're about to read:

The AI answer is becoming the front page of search — and the front page no longer hands out clicks by default.

People haven't stopped searching. They've stopped clicking. Google still fields billions of queries, ChatGPT answers hundreds of millions of people every week, and buyers still land on websites — but the answer now gets read before any link gets touched. Every section below is a different angle on that one shift: how many people are on the answer side, how they behave there, and what it does to the businesses on the other side. (This is also the shift behind the whole GEO vs SEO conversation — but first, the numbers.)

AI search by the numbers: the 2026 snapshot

Short on time, or building a deck? These are the headline figures. Each one links straight to its primary source.

How big is AI search now?

Big enough that AI platforms now report their numbers the way search engines used to. Here's the scoreboard.

Two things in that table deserve a closer look.

Usage is deepening, not just widening. OpenAI's own research on a 0.1% sample of accounts found that six months after signing up, ChatGPT users send 50% more messages per day and have doubled the number of distinct task types they use it for. Non-English-primary users are now over half of active users, led by Spanish, Portuguese and Arabic, and in Q1 2026 users over 35 gained share of messages (consumer plans only — under-35s are still the largest group). That's the profile of a habit forming, not a novelty being sampled.

The traffic is growing faster than any single leader. Inside Similarweb's 9.5 billion monthly visits, ChatGPT's share of gen-AI web traffic fell from roughly 76% to 53% in a year, while Gemini took 27–28% and Claude around 9%. ChatGPT didn't shrink; the category outgrew it. App installs tell the same story — 4.4 billion downloads, up 58%.

Cloudflare, which watches this from the infrastructure layer, adds a stranger data point: more than 50% of internet traffic is now agent-based rather than human. On its live dashboard for the week ending 14 August 2026, AI bots made 7.9% of all HTML page requests, against 39.9% from humans and 48.7% from non-AI bots.

One counterweight before you extrapolate to "everyone lives in ChatGPT now": SparkToro and Datos' clickstream study — 2025 data, desktop only, so treat it as a floor — found just over 20% of Americans used AI tools ten or more times a month, and 95%+ still visited a traditional search engine monthly. Adoption is real. Replacement is not. Which brings us to behavior.

How people actually use AI search

The most important behavioral finding on this whole page is a number that refused to move.

95% of ChatGPT users also use Google — unchanged from September 2025 to May 2026 (Similarweb). Eight months of explosive AI growth and the overlap didn't budge: people bolt AI onto their search habits rather than swapping it in. Semrush's buyer survey (1,030 US consumers, December 2025) lands in the same place — 77% use AI and traditional search together, and only 4% rely mostly on AI.

Who's using it is normalizing too. The 18–34 cohort's share of AI chatbot usage fell from 61% to 50% — not because Gen Z left, but because everyone else arrived. Here's how monthly usage breaks down by tool among US consumers:

Behavior inside AI search differs from classic search in kind, not just degree. Google reports the average AI Mode query is three times the length of a traditional query, and 1 in 6 US searches now use voice or images. People bring AI their full question instead of two keywords — which is exactly why brands get named, compared and recommended inside these answers rather than merely listed.

Meanwhile, AI moved into the regular results page. Pew Research Center's March 2025 study (900 US adults, 68,879 real searches) found 18% of Google searches produced an AI summary, 58% of users saw at least one that month, the median summary ran 67 words, and 88% cited three or more sources. A year on, Similarweb puts AI Overviews on more than 40% of US searches, up from about a third.

And it's already a discovery channel, not just an answer machine: 43% of US consumers have discovered a new brand through AI — though 86% verify AI recommendations before buying. The AI answer opens the buying journey. It rarely closes it alone.

What AI search does to your traffic

Here's where those behavior shifts land in your analytics — and where the honest numbers get uncomfortable.

Zero-click is now the default outcome

More than two-thirds of US Google searches now end without a click. Pew's behavioral data explains the mechanism: when an AI summary appeared, users clicked a result on just 8% of visits, versus 15% without one — and only 1% clicked a link inside the AI summary itself. Sessions were also more likely to simply end there: 26% of pages with a summary closed the browsing session, versus 16% without.

What AI Overviews do to click-through rates

Ahrefs tracked 300,000 keywords through Google Search Console and found the presence of an AI Overview cut position-1 organic CTR by 58% (from 0.073 to 0.016, December 2023 vs December 2025) — and the damage is widening, up from a 34.5% reduction measured in April 2025. Losses run down the whole page, from −50.8% at position 2 to −19.4% at position 10.

But the damage is not evenly distributed — and this is the nuance most roundups miss:

Look at the branded row. When someone searches your brand by name and an AI Overview appears, clicks went up 18.68%. The squeeze hits generic informational queries — searches where you were borrowing attention. Searches where you are the answer got stronger. Being the brand the AI names is the defensible position.

The crawl-to-click gap

AI engines read far more of your site than they send back. Cloudflare's live Radar dashboard makes the exchange rate explicit — pages crawled per referral visit sent, week ending 14 August 2026:

Note the direction of travel, though: in July 2025, Cloudflare measured Anthropic at 38,065 pages crawled per referral. A year later it's 988:1. The gap is enormous, but referrals are catching up fast. Meanwhile 52% of crawler requests were for AI training as of June 2026, up from 22% in spring 2025 — and heavily-crawled content categories saw human traffic fall by up to 40% in a year.

Referrals exist — they just hide in your analytics

ChatGPT is citing and linking more, not less. Similarweb tracked the citation rate in US prompts rising from 1.6% in June 2025 to 6.8% in May 2026 — with wild variance by vertical: travel and hospitality around 23%, automotive around 20%, professional services under 4%. (About 26% of ChatGPT responses now carry an ad, so paid is arriving too.) When OpenAI changed how links render in May 2026, total ChatGPT referrals jumped 157.7% week over week, homepage referrals rose 354.7%, and pages per visit climbed from 3.8 to 4.7 (desktop panel). A single product decision moved everyone's traffic charts — worth remembering before you credit your strategy or panic at a dip.

The bigger effect never shows up labeled "AI" at all. Similarweb found users who received an AI recommendation for a brand were 2.5x more likely to visit that brand's site within 7 days — and 56%+ of those visits arrived via branded search, against a roughly 40% baseline. The AI answer does its work, then the user types your name into Google — and the assist goes uncredited. If branded search is quietly climbing while non-branded organic sags, this is likely why.

What buyers do after an AI answer

Clicks are scarcer, then. The next question is what the surviving ones are worth.

Semrush's December 2025 consumer survey found 50% of US consumers have made a purchase after using AI in their research — and 22% have purchased directly inside an AI tool. The prompt-to-purchase path is no longer hypothetical; a fifth of consumers have already walked it.

Adobe Analytics, which measures actual transactions across more than a trillion visits to US retail sites, tracked the quality shift: AI-sourced traffic to US retail grew 235% year over year (Jan–May 2026), and during Prime Day 2026 those visitors converted 40% better than average — a complete reversal from a year earlier, when they converted 23% worse. AI visitors used to browse and bounce. Now they arrive pre-sold.

The 2025 holiday season showed the same pattern: AI referrals converted 31% better, drove 254% more revenue per visit, spent 45% more time on site and bounced 33% less. Growth by industry was steep everywhere:

Adobe's companion survey of 1,000+ consumers explains the conversion lift: 65% of AI-assisted shoppers report more purchase confidence, 68% say they're less likely to return the product, and 55% actively click links AI gives them. The AI did the comparison shopping before your site ever loaded. What arrives is a decision, not a browser.

The market: tools, funding and consolidation

Follow the money for a second opinion on everything above. In well under a year, AI visibility went from niche worry to a category venture firms and enterprise software giants pay real prices to enter.

A billion-dollar valuation for a company that measures AI answers, and two acquisitions where the buyer's stated logic was AI-driven discovery — that's the market pricing in the shift before most marketing teams have a line item for it.

The most useful research to come out of the category so far is Semrush's expanded 2026 AI Visibility Index, built on 126 million US AI prompts analyzed between January and April 2026. Four findings stand out:

• ChatGPT averages 15 sources per response; Gemini averages 3. Which engine you optimize for changes the odds of being cited at all.
• Only 36 brands held top-100 visibility across all four major platforms in every month studied. Cross-engine consistency is rare — and therefore valuable. (That consistency is what share of voice in AI answers measures.)
• 45% of marketing leaders say they can't measure their AI visibility. Nearly half the market is flying blind on a channel their buyers already use.
• 81% of organizations running integrated SEO + AI strategies saw traffic or lead gains, versus 36% of those running them separately. Treating AI search as a bolt-on project underperforms treating it as one discipline.

Why the numbers get confusing

Plenty of the numbers above look like they can't all be true. Mostly they can — they're answering different questions. Three collisions are worth defusing, because someone in your next meeting will raise one.

Google says 1 billion people use AI Mode. SparkToro says 0.34% of searches go there. Who's lying?

Probably nobody. Google's 1 billion+ figure is monthly active users — anyone who touched AI Mode at least once in a month. SparkToro's 0.34% is AI Mode's share of total Google search volume (January–April 2026) — and Google runs so many searches that a billion occasional users can still round to a fraction of a percent of queries. Add the instrument differences (SparkToro's panel excludes the Google mobile app) and both numbers stand. The lesson generalizes: user counts measure reach; query share measures habit. Vendors will always quote whichever flatters them.

"AI is killing Google" vs Google's growing ad business

While AI adoption exploded, Google's Search & Other ad revenue grew 17% year over year in Q2 2026, with Alphabet revenue up 24% — and Google claims its AI features drive "billions of clicks to websites every week." For raw scale, the most recent like-for-like comparison (August 2025 snapshot, so a year old now) had Google at 84.5 billion monthly visits against ChatGPT's 5.72 billion — ChatGPT growing 117% year over year, and still under a tenth the size. Google isn't dying. Google is becoming an answer engine — which is precisely why your clicks are falling while its revenue rises. The threat to you and the threat to Google were never the same threat.

Grading the famous Gartner prediction

In February 2024, Gartner predicted traditional search engine volume would drop 25% by 2026. It's 2026. Verdict: wrong metric, right instinct. Search volume held — 95%+ of Americans still visited a traditional search engine monthly in 2025, and Google's revenue says demand is fine. What actually fell was clicking: the zero-click share of US Google searches rose 19 points, from 49% in 2019 to 68.01% in early 2026. Searches stayed; their value to publishers didn't. If you graded your own strategy against "search is dropping 25%," you prepared for the wrong disaster.

The general rule for reading any AI statistic: check the instrument. Company-stated figures are marketing shaped like measurement. Panels (Similarweb, Datos) skew desktop and miss apps. Surveys measure what people say, not what they do. The numbers on this page disagree less than they appear to once you know which instrument produced each one.

What these numbers mean for you

Enough measuring. Here's what the data actually tells you to do, and the one number backing each move.

• Find out what AI engines say about your brand — this quarter, not next year. 43% of consumers have discovered a brand through AI; if you're not the recommendation, a competitor is.
• Stop grading yourself on non-branded clicks alone. With 68% of searches ending clickless, flat traffic can hide growing influence — and falling traffic can hide stable demand.
• Watch branded search as your AI-attribution proxy. 56%+ of AI-influenced visits arrive as branded search, so that's where the AI dividend shows up in your analytics.
• Treat AI referrals as your highest-intent segment. They converted 40% better on Prime Day 2026; a small stream that converts like that deserves its own landing experience.
• Run SEO and AI visibility as one program, not two. 81% of integrated programs saw gains versus 36% of separated ones — the mechanics overlap more than the acronyms suggest.
• Structure content so engines can lift it cleanly. ChatGPT pulls from 15 sources per response — clear claims, real data and honest sourcing are what get you into those 15. (That practice has a name: generative engine optimization — GEO, the AI-answer counterpart to SEO.)

Start with the first one. Every other move on the list depends on knowing where you currently stand in the answers — and most of your competitors, remember, can't measure it yet.

Frequently asked questions

How many people use AI search?

More than 900 million people use ChatGPT every week (OpenAI, February 2026), Google AI Mode has passed 1 billion monthly users (Google, May 2026), and the Gemini app reports 950 million monthly users (Alphabet, July 2026). Cloudflare estimates around 2.5 billion people, roughly 30% of humanity, now use generative AI regularly. Overlap between platforms is heavy, so these numbers cannot simply be added together.

Is ChatGPT replacing Google?

No. The data says people add ChatGPT alongside Google rather than switching. 95% of ChatGPT users also use Google, a figure that held flat from September 2025 to May 2026 (Similarweb), and Google's search ad revenue grew 17% year over year in Q2 2026 (Alphabet). As of July 2025, Google still drew 84.5 billion monthly web visits to ChatGPT's 5.72 billion.

What percentage of searches are AI now?

It depends what you count. Only 0.34% of US Google searches went to AI Mode between January and April 2026 (SparkToro). But AI now appears inside ordinary search: AI Overviews show on more than 40% of US Google searches, up from about a third a year earlier (Similarweb); a separate Pew Research Center panel measured AI summaries on 18% of Google searches in March 2025. Measured as destinations, generative AI platforms drew 9.5 billion monthly web visits in the year to May 2026 (Similarweb).

Do AI Overviews reduce website traffic?

For non-branded queries, yes. Ahrefs measured a 58% drop in click-through rate for the number-one organic result when an AI Overview is present, and Pew found users clicked a result on only 8% of searches with an AI summary versus 15% without. The exception is branded queries, which gained 18.68% CTR when AI Overviews appeared (Amsive, 2025 data).

Was Gartner right that search would drop 25% by 2026?

Not as stated. Gartner predicted in February 2024 that traditional search engine volume would fall 25% by 2026. In 2025, SparkToro and Datos found more than 95% of Americans still visited a traditional search engine monthly, and Google's search ad revenue rose 17% year over year in Q2 2026. What did fall is clicking: 68.01% of US Google searches now end without a click. Volume held; clicks did not.