Insights

156 AI Search Statistics (October 2026)

Updated October 2026. Every number is traced to the organisation that produced it, linked to the document it was published in, and dated.

Why this page exists

Most pages about AI search quote each other. This one does not. Every number below is traced to the organisation that produced it, linked to the document it was published in, and dated. Where two credible studies disagree, both appear, because the disagreement is the finding.

The short version: search did not die, it stopped sending clicks. Two thirds of Google searches now end without one. At the same time, a new and much smaller stream of traffic arrives from AI assistants, and it behaves better than the traffic it replaces. Businesses that still measure success in sessions are reading the wrong instrument.

If you run a company that depends on being found, three questions matter. How many buyers now ask an assistant instead of a search box. What those assistants read before they answer. Whether anything you do to your own website changes the answer. The data below addresses all three.

Key Takeaways

68.01% of US Google searches ended without a click in the first four months of 2026 (SparkToro, Jun 2026).
Google users clicked a traditional result on 8% of visits where an AI summary appeared, against 15% of visits where none did (Pew Research Center, Jul 2025).
Users clicked a link inside the AI summary itself on just 1% of visits (Pew Research Center, Jul 2025).
The presence of an AI Overview correlates with a 58% lower average click-through rate for the top-ranking page (Ahrefs, Feb 2026).
Use of generative AI tools for local business recommendations rose from 6% of consumers to 45% in a single year (BrightLocal, Feb 2026).
41% of consumers have made a purchase based explicitly on an AI chatbot recommendation (Semrush, Sep 2026).
57.5% of AI users have decided not to buy something because of what a chatbot told them (Semrush, Sep 2026).
Outbound US referral traffic from ChatGPT to the rest of the web grew 206% in 2025 (Semrush, Apr 2026).
Google AI Mode passed 1 billion monthly active users after its global expansion (Google, Jul 2026).
Training now drives nearly 80% of AI bot activity, up from 72% a year earlier (Cloudflare, Aug 2025).
When an assistant could see neither the brand nor its own domain in the live retrieval path, the brand was mentioned in 2.8% of GPT answers and 3.8% of Gemini answers; when both own-domain exposure and a branded fan-out occurred, mention rates reached 91.4% and 100% (arXiv, Sep 2026).
Ranked "best of" listicles are the most-cited content format in AI answers, at about 21% of all citations (arXiv, Jun 2026).
ChatGPT's share of generative AI website visits fell from about 76% to roughly 53% in twelve months as Gemini and Claude took share (Similarweb, Jul 2026).
Asked a discovery-style question rather than named directly, recognition of real products collapsed to 3.32% in ChatGPT and 8.29% in Perplexity (arXiv, Jan 2026).

How many searches still send you a click

This is the only statistic that reframes everything else. Clicks were already leaking away from websites before generative AI arrived; AI features accelerated a decade-long trend rather than starting it.

The current zero-click rate

In 2024, US zero-click searches on Google stood at 60.45% (SparkToro, Jun 2026).
That is 12.5% growth, or 7.5 percentage points, in clickless queries over two years (SparkToro, Jun 2026).
Ten years ago, roughly 45% of Google searches were zero-click, making the move to today's figure a 33.8% increase, or 23 points, in a decade (SparkToro, Jun 2026).
SparkToro first measured zero-click searches in 2016, when 45% of searches ended without a click (SparkToro, Jun 2026).
By 2019 the figure was 49%, the number that started the public argument (SparkToro, Jun 2026).
The 7.5-point move took 24 months, the fastest acceleration recorded in a decade (SparkToro, Jun 2026).

How much click share websites actually lost

Between June 2025 and May 2026, the traffic share Google sent to tens of thousands of tracked websites fell 8 percentage points, a decline of about 22% (SparkToro, Jun 2026).
The sharpest single drop in the dataset was 9.51 points, a 22.9% decline (SparkToro, Jun 2026).
AI tools themselves send less than 1% of all web traffic (SparkToro, Jun 2026).
More than 20% of Americans now use an AI tool at least ten times a month (SparkToro, Jun 2026).

Zero-click rates are not uniform across countries

German Google searchers click at rates more than 20% higher than searchers in the United Kingdom, which has the highest zero-click rate in the comparison (SparkToro, Jun 2026).
Across the countries measured, only about four points separate the highest click rate from the lowest, 43.1% in the UK against 38.2% in Canada (SparkToro, Jun 2026).

Where referrals came from before AI

Close to two thirds, 63.41%, of all US web traffic referrals from the top 170 sites began on Google.com (SparkToro, Mar 2024).
Microsoft-owned sites together accounted for 7.21% of referrals (SparkToro, Mar 2024).
Google often refers more than 70% of a given site's traffic, exceeding all social media and content sites combined (SparkToro, Mar 2024).
Only Microsoft's sites, Facebook, YouTube, Reddit, DuckDuckGo, Yahoo and Twitter sent 1% or more of referral traffic (SparkToro, Mar 2024).
Only 22 sites held a 0.1% or greater share of referrals to the long tail of the web (SparkToro, Mar 2024).
Over the 13 months studied, the long tail lost about 3.2% of share to the top 170 sites (SparkToro, Mar 2024).

What AI Overviews do to click-through rate

The honest position is that every published measurement shows a decline, and the published declines differ widely because each study used a different keyword set and method. Treat the direction as settled and the magnitude as a range.

The measured declines

Users who encountered an AI summary ended their browsing session on 26% of those pages, against 16% of pages with only traditional results (Pew Research Center, Jul 2025).
About 9% of AI Overviews appear somewhere other than position one in the results page (Ahrefs, Feb 2026).
AI Overviews now appear on more than 20% of all searches, and when one is present it cuts click-through rate by nearly 60% (SparkToro, Jun 2026).

Being cited inside the answer changes the maths

When clients appear inside an AI Overview, organic click-through rate rises from 0.74% to 1.02% (Seer Interactive, Feb 2025).
Paid click-through rate for those same cited clients rises from 7.89% to 11% (Seer Interactive, Feb 2025).

Which queries trigger an AI summary at all

18% of all Google searches in Pew's study generated an AI summary (Pew Research Center, Jul 2025).
58% of respondents ran at least one Google search in March 2025 that produced an AI summary (Pew Research Center, Jul 2025).
Just 8% of one-word or two-word searches produced an AI summary (Pew Research Center, Jul 2025).
That share rose to 53% for searches of ten words or more (Pew Research Center, Jul 2025).
60% of queries beginning with a question word such as who, what, when or why produced an AI summary (Pew Research Center, Jul 2025).
36% of searches containing both a noun and a verb generated an AI summary (Pew Research Center, Jul 2025).
The vast majority of summaries, 88%, cited three or more sources (Pew Research Center, Jul 2025).

AI Overviews are moving into commercial queries

Across a six-month analysis of more than 600,000 keywords in 10 industries, AI Overviews grew an average of 71% on results pages with commercial intent (Semrush, Jul 2026).
Results pages containing both Google Ads and an AI Overview grew 394% toward the end of 2025 (Semrush, Jul 2026).
Over the same period, the average share of transactional results pages carrying an AI Overview fell 5% (Semrush, Jul 2026).
Finance saw the largest increase in commercial-intent AI Overviews at 231.25%, followed by computers and electronics at 107.62% and games at 76.66% (Semrush, Jul 2026).
In food and drink the pattern inverted, with AI Overviews on 16% of transactional pages against 14% of commercial ones (Semrush, Jul 2026).
Among the most expensive keywords in jobs and education, more than 60% showed an AI Overview, the highest of any industry measured (Semrush, Jul 2026).

How large AI referral traffic actually is

Two facts sit awkwardly together. AI referral traffic is growing faster than anything else on the web, and it is still a rounding error next to search. Both are true, and planning that ignores either one fails.

Volume and growth

Average monthly web visits across generative AI platforms worldwide grew 70% year over year to 9.5 billion between June 2025 and May 2026 (Similarweb, Jul 2026).
Unique visitors across those platforms grew more slowly, up 57% to 655 million (Similarweb, Jul 2026).
ChatGPT's US traffic plateaued around November 2025, hovering near 1 billion monthly US visits through early 2026 (Semrush, Apr 2026).
Comparing January 2025 with January 2026, outbound US ChatGPT referral traffic grew 206% year over year (Semrush, Apr 2026).
Marketplaces lead every industry for AI referral traffic at 46.8 million average monthly visits between June 2025 and May 2026 (Similarweb, Sep 2026).
The marketplace category tripled from 13.9 million average monthly visits the year before (Similarweb, Sep 2026).
Beauty was the fastest-growing industry for AI referral traffic, up 312.5% year over year, with fashion close behind at 278.2% (Similarweb, Sep 2026).
Not one of the seven categories measured grew less than 115% (Similarweb, Sep 2026).
Finance reached 19.1 million average monthly visits, up 236.5%, and consumer electronics 17.1 million, up 215.4% (Similarweb, Sep 2026).
Marketplaces grew 237.3% from the largest base of all, which undercuts the argument that high growth is only a base effect (Similarweb, Sep 2026).

Where that traffic goes once it leaves the assistant

More than 30% of US ChatGPT referral traffic goes to just 10 domains (Semrush, Apr 2026).
Google alone accounts for 21.6% of US ChatGPT referral traffic (Semrush, Apr 2026).
The share of US ChatGPT referral traffic going to Google climbed from roughly 14% at the start of the study period to more than 21% by early 2026 (Semrush, Apr 2026).
The next nine domains after Google add another 8.6% of US referrals (Semrush, Apr 2026).
The top 10 domains have held between 20% and 32% of US ChatGPT referrals since October 2024 (Semrush, Apr 2026).
Homepage referrals from ChatGPT rose from roughly 26% to 29% of all its referral traffic before one product update to about 62% to 63% by late May 2026, and stayed there (Similarweb, Jul 2026).

How often the assistant even looks at the live web

As of February 2026, ChatGPT enabled web search on 34.5% of US queries, down from 46% in late 2024 (Semrush, Apr 2026).
The share of search-enabled US sessions ranged from as low as 15% to as high as 66.3% across the study period (Semrush, Apr 2026).
For most of the period, between 65% and 85% of prompts could not be matched to any traditional search keyword in Semrush's database (Semrush, Apr 2026).
After twelve months of flat engagement, average queries per session jumped 50% in the final four months of the study (Semrush, Apr 2026).
Citations appeared in about 1.6% of US ChatGPT prompts in June 2025, rising to roughly 6.8% by May 2026 (Similarweb, Jul 2026).
Travel and hospitality prompts carry a citation about 23% of the time, the highest rate tracked, with automotive near 20% (Similarweb, Jul 2026).
Professional services prompts carry a citation less than 4% of the time (Similarweb, Jul 2026).

How buyers behave after an AI recommendation

This is the section most businesses skip, and it is the one that decides revenue. An assistant does not only recommend. It also talks people out of things.

AI recommendations that end in a purchase

Among consumers who use AI at least occasionally, 57.51% have bought a product an AI chatbot recommended (Semrush, Sep 2026).
Of consumers who use AI multiple times a week, 80.45% have bought an AI-recommended product at least once, and 54.95% have done so more than once (Semrush, Sep 2026).
59.27% of AI users have discovered a brand or product through a chatbot recommendation that they did not previously know about (Semrush, Sep 2026).
More than a quarter, 25.79%, have discovered new brands and products this way frequently (Semrush, Sep 2026).
Across all consumers, 43.14% have discovered a new brand through an AI tool (Semrush, Sep 2026).
Among consumers actively using AI for product research, only 9.38% have never discovered a new product or brand from a chatbot (Semrush, Sep 2026).

AI recommendations that kill a purchase

Among consumers who name AI chatbots as a source they use when buying online, 80.93% have decided against a purchase on the basis of AI guidance (Semrush, Sep 2026).
74.15% of all consumers would be at least somewhat less likely to buy a product if a chatbot flagged mixed or negative reviews (Semrush, Sep 2026).
Among current AI users that figure rises to 85.31% (Semrush, Sep 2026).
70.62% of current AI users believe chatbots can recommend the best brands and products, rising above 80% among weekly users (Semrush, Sep 2026).
Almost half of all consumers, 47.54%, now at least occasionally consult an AI chatbot about a company or seller before buying (Semrush, Sep 2026).
Among those AI users, 32.68% consult chatbots about a seller frequently (Semrush, Sep 2026).

Local business discovery through AI

For a local service business this is the whole story. The source mix for local recommendations changed faster in one year than it had in the previous ten.

Adoption

Use of AI tools for local business recommendations climbed from 6% of consumers in 2025 to 45% today, making AI the third most popular source of business recommendations (BrightLocal, Mar 2026).
Adults aged 30 to 44 lead adoption, with 64% having asked an AI tool for a business recommendation in the past year (BrightLocal, Mar 2026).
Consumers over 60 are the most cautious, with 24% turning to AI for local business guidance (BrightLocal, Mar 2026).
ChatGPT is the frontrunner, used by 31% of consumers for business recommendations in the last 12 months (BrightLocal, Mar 2026).
Google's AI Mode follows at 23%, ahead of Gemini (BrightLocal, Mar 2026).
In early 2025, 40% of consumers reported actively using generative AI when searching online (BrightLocal, Mar 2026).
Only 3% of consumers treat an AI platform as their default for local searches (BrightLocal, Mar 2026).

Trust, and its limits

Among active AI users, nearly two thirds, 63%, trust AI recommendations, while 10% express distrust (BrightLocal, Mar 2026).
Among people who do not use AI, 53% say they do not trust its business recommendations (BrightLocal, Mar 2026).
AI platforms are already trusted equally with reviews by 42% of consumers (BrightLocal, Mar 2026).
Most consumers, 88%, fact-check the reviews an AI tool cites (BrightLocal, Mar 2026).

The review layer the assistant reads

97% of consumers read reviews for local businesses (BrightLocal, Feb 2026).
41% of consumers always read reviews when browsing for businesses, up from 29% the previous year (BrightLocal, Feb 2026).
Google's share as a review source dipped from 83% in 2025 to 71% (BrightLocal, Feb 2026).
Just 35% of small and medium businesses have a Google Business Profile at all (BrightLocal, Feb 2026).
Apple Maps nearly doubled in usage, from 14% in 2025 to 27% (BrightLocal, Feb 2026).
Local news sites fell sharply as a discovery source, from 48% to 29% (BrightLocal, Feb 2026).
47% of consumers will not use a business with fewer than 20 reviews (BrightLocal, Feb 2026).
74% of consumers only care about reviews written in the last three months (BrightLocal, Feb 2026).
31% of consumers will only use a business rated 4.5 stars or higher (BrightLocal, Feb 2026).

What AI engines actually cite

The uncomfortable finding for brand websites is how much of the citation pool is other people's property. The useful finding is which format wins when a corporate site does get cited.

Source mix

When engines cite sources, about 78% of citations go to corporate websites, and among non-corporate sources YouTube leads, ahead of Reddit, editorial media and Wikipedia (arXiv, Jun 2026).
Government sites made up 6% of sources linked in AI Overviews, against 2% for standard search results (Pew Research Center, Jul 2025).
News websites accounted for 5% of AI summary links and 5% of standard search results (Pew Research Center, Jul 2025).
Wikipedia, YouTube and Reddit together accounted for 15% of the sources listed in the AI summaries Pew examined, and a similar 17% of standard search results (Pew Research Center, Jul 2025).

Whether your own pages can move the answer

With an own-domain citation present but no branded fan-out, target brand mention rates rose to 49.0% in GPT and 58.4% in Gemini (arXiv, Sep 2026).
A previous non-mention combined with no current own-domain exposure produced next-run mention rates of 1.6% and 1.9% (arXiv, Sep 2026).
A previous mention combined with current exposure produced next-run mention rates of 80.5% and 83.7% (arXiv, Sep 2026).
On a 30% holdout, the full model predicting brand mention reached AUC 0.963 on GPT and 0.942 on Gemini, against 0.880 and 0.840 for live signals alone (arXiv, Sep 2026).
When users asked about products by name, recognition was near perfect at 99.4% for ChatGPT and 94.3% for Perplexity (arXiv, Jan 2026).

How AI crawlers treat your website

Crawling is the supply side of AI search, and the imbalance is extreme. Most of what AI companies take from a website is used to train a model, not to send a visitor.

Crawl volume and growth

AI and search crawler traffic grew 18% from May 2024 to May 2025, or 48% including newly added Cloudflare customers (Cloudflare, Jul 2025).
Crawling by search engines and AI services rose 24% year over year in June 2025 before slowing to 4% year-over-year growth in July (Cloudflare, Aug 2025).
Crawling traffic peaked in April 2025 at 32% above May 2024 (Cloudflare, Jul 2025).
Around 30% of global web traffic comes from bots, exceeding human internet traffic in some locations (Cloudflare, Jul 2025).
AI user-action crawling, the kind triggered by a live user prompt, increased more than fifteenfold in 2025 (Cloudflare, Dec 2025).
Googlebot alone accounted for 4.5% of HTML request traffic, while all other AI bots together accounted for 4.2% (Cloudflare, Dec 2025).
Global internet traffic grew 19% in 2025 (Cloudflare, Dec 2025).

Which bots are doing the crawling

Googlebot remains the anchor at 39% of all AI and search crawler traffic (Cloudflare, Aug 2025).
GPTBot's share of AI crawling grew from 4.7% in July 2024 to 11.7% in July 2025 (Cloudflare, Aug 2025).
ClaudeBot rose from 6% to nearly 10% over the same period, while Meta's crawler jumped from 0.9% to 7.5% (Cloudflare, Aug 2025).
Amazonbot fell from 10.2% to 5.9% and ByteDance's Bytespider from 14.1% to 2.4% (Cloudflare, Aug 2025).
Measured against a different baseline, GPTBot's share rose from 2.2% to 7.7%, a 305% rise in requests (Cloudflare, Jul 2025).
Requests from ChatGPT-User surged 2,825%, reaching a 1.3% share (Cloudflare, Jul 2025).
PerplexityBot recorded the highest growth rate of any crawler at a 157,490% increase in raw requests, from a 0.2% share (Cloudflare, Jul 2025).
Measured against a different crawler panel, Googlebot's share rose from 30% to 50%, which is not the same denominator as the 39% figure above and should not be read as a contradiction (Cloudflare, Jul 2025).
GPTBot gained 16 percentage points of share, Meta's crawler more than 15, and ClaudeBot 8 (Cloudflare, Aug 2025).

Why the crawling happens

Training traffic accounts for nearly 80% of crawling from AI bots and follows no clear cyclical pattern (Cloudflare, Aug 2025).
User-action and undeclared crawling purposes together accounted for less than 5% of AI bot traffic (Cloudflare, Aug 2025).
For one vertical measured, three quarters of crawling traffic in the first week of August came from just four bots, and 80% of it was for model training (Cloudflare, Aug 2025).

Crawl-to-refer ratios

In the default view, Anthropic had the highest crawl-to-refer ratio at nearly 50,000 to 1, followed by OpenAI at 887 to 1 and Perplexity at 118 to 1 (Cloudflare, Aug 2025).
For news and publications sites the ratios tightened, with Anthropic at 2,500 to 1, OpenAI at 152 to 1 and Perplexity at 32.7 to 1 (Cloudflare, Aug 2025).
For the week of 19 to 26 June 2025, ratios ranged from Anthropic's 70,900 to 1 down to Mistral's 0.1 to 1, meaning Mistral sent ten times as many referrals as crawl requests (Cloudflare, Jul 2025).
Over the same comparison, ratio changes ranged from increases above 6% for DuckDuckGo and Yandex to a 19.4% decrease for Google (Cloudflare, Jul 2025).
Anthropic increased referrals but still crawled 38,000 pages per referred visitor in July 2025, down from 286,000 to 1 in January (Cloudflare, Aug 2025).
Google referrals to news sites in March 2025 were about 9% lower than in January, and April was 15% lower (Cloudflare, Aug 2025).

The platform split is no longer only ChatGPT

Any plan that treats AI search as one destination is already out of date. The traffic is fragmenting, and the share shift inside twelve months was larger than most forecasts allowed.

Share of AI web traffic

Gemini rose from under 9% to around 27% to 28% of generative AI web traffic in twelve months (Similarweb, Jul 2026).
Claude moved from barely 2% to close to 9%, the largest proportional gain of any platform tracked (Similarweb, Jul 2026).
Claude's unique visitors grew 180% in the second half of the year compared with the first half (Similarweb, Jul 2026).
One platform's worldwide app monthly active users grew roughly 1,318% between June 2025 and May 2026 (Similarweb, Jul 2026).
Only 0.34% of Google searches reached AI Mode between January and April 2026 (SparkToro, Jun 2026).

What Google reports about its own AI surfaces

Google AI Mode surpassed 1 billion monthly active users after expanding globally the previous October (Google, Jul 2026).
The Gemini app reached 950 million monthly active users, with daily active users tripling in a year (Google, Jul 2026).
Google reported 17% revenue growth in Search and Other for the quarter (Google, Jul 2026).
Alphabet revenue grew 24% year over year in the same quarter (Google, Jul 2026).
Google's model APIs process approximately 22 billion tokens per minute, up from 16 billion a quarter earlier (Google, Jul 2026).
Search revenue grew 19% year over year in the prior quarter (Google, Jun 2026).
Google reduced Gemini serving costs by 78% during 2025 (Google, Jun 2026).

What still earns a click

The useful question is not how to restore 2019 click-through rates. It is which pages still get chosen when the assistant has already answered the question.

Page and brand characteristics that correlate with winning

Nearly 84% of sites that gained traffic let users finish what they came to do, against half of the sites that lost traffic (SparkToro, Jun 2026).
Sites with only one of the measured strategic features won just 15% of the time (SparkToro, Jun 2026).
Sites with four or five of those features won around 68% to 70% of the time (SparkToro, Jun 2026).
Branded terms make up roughly 44% of Google searches, and AI answers preferentially cite entities they already recognise (SparkToro, Jun 2026).
Asked discovery-style questions such as which tools are best, recognition of real products fell to 3.32% in ChatGPT and 8.29% in Perplexity, which indicates that optimisation multiplies existing visibility rather than creating it (arXiv, Jan 2026).

If your category is one where assistants rarely cite anything, the opportunity is different from a category where they cite constantly. A professional services firm facing a citation rate under 4% is fighting for a smaller pool than a travel brand facing 23%, and the correct strategy differs accordingly.

Glossary

AEO: answer engine optimisation, the practice of optimising to be the answer rather than a ranked result.
AI Mode: Google's conversational search surface, separate from AI Overviews.
AI Overview: the AI-generated summary Google places above traditional search results.
Crawl-to-refer ratio: how many pages a bot requests for every visitor it sends back.
CTR: click-through rate, clicks divided by impressions.
Fan-out: the set of follow-up queries an assistant issues before answering.
GBP: Google Business Profile, the listing that powers local results and map packs.
GEO: generative engine optimisation, optimising to be cited by AI engines.
MAU: monthly active users.
SMB: small and medium business.
Zero-click search: a search that ends without the user clicking through to any website.

Methodology and Sources

Every statistic on this page was taken from the organisation that produced the data, not from a roundup that republished it. Each source page was fetched and its text searched for the exact figure quoted, and each publication date was read from the source page's own published-date metadata rather than estimated. Statistics whose source page could not be fetched, or which carried no verifiable publication date, were dropped rather than cited, which is why several widely quoted AI search figures do not appear here. Where studies disagree on magnitude, both figures are shown with their methods attributed, because prevalence and click-through numbers vary with keyword set, geography, device mix and detection method. This page is reviewed and refreshed monthly.

Sources

What to do with these numbers

The data says three things plainly. Clicks are scarcer, the assistants cite sources they already recognise, and your own pages measurably change whether your brand gets named. That is a build problem before it is a marketing problem.

If you want the mechanics, readwhy a business doesn't show up in ChatGPTand how AI search works for a Scottsdale business. For how we work and what it costs, see ourmethod and pricing.