AI Search Statistics 2026: Clicks, Traffic and Who Gets Named

How many buyers ask AI before they ask Google?

A large share of your buyers now start in a chatbot. In February 2026, 42% of US adults told Pew they use chatbots to search for information, the most common use Pew measured.

ChatGPT leads. 44% of US adults use it, up from 34% a year earlier. Adults under 50 are twice as likely to use it as older adults.

What this means for you: assume that some of your next ten buyers already asked an assistant about your category. They arrived at your site with a shortlist, or they never arrived at all.

Age matters. If your buyers are under 50, the share who use ChatGPT is closer to six in ten. If they are older, it is closer to three in ten, and Google remains the first stop.

Pew's numbers describe all US adults, not only buyers of your service. Ask your last ten customers how they found you. That answer is worth more than any average here.

• 42%of US adults use chatbots to search for informationPew Research Center, February 2026
• 44%use ChatGPT, up from 34% in 2025Pew Research Center, February 2026
• 57%of adults under 50 use ChatGPT, against 28% of those 50 and olderPew Research Center, February 2026
• 60%say they read the AI summary at the top of search resultsPew Research Center, February 2026

• Gemini app: 950 million monthly active users, July 2026 (Alphabet).
• Google AI Mode: more than 1 billion monthly active users, July 2026 (Alphabet).
• ChatGPT: reported to be approaching 1 billion weekly active users, July 29, 2026 (The Information, via PYMNTS). Pew's 44% counts anyone who ever uses it.
• Generative AI sites: 9.5 billion web visits a month worldwide, up 70% in a year, from 655 million unique visitors, up 57% (Similarweb, June 2025 to May 2026).
• Daily use: 24% of US adults use a chatbot daily, and 12% several times a day (Pew, February 2026).
• AI summaries: 30% of US adults say they do not read them and 10% are not sure (Pew, February 2026).

How big is AI search next to Google?

Still small in visits, and growing fast in questions asked. Ahrefs estimates that ChatGPT handles about 12% of the search volume Google does for things people used to search on Google.

That is an estimate, and Ahrefs shows its method: 65% of ChatGPT's roughly 2.5 billion daily prompts treated as search, divided by Google's 13.7 billion daily searches. The visits tell a different story.

What this means for you: do not expect a flood of chatbot visits. Expect a trickle of better-informed buyers, plus a larger number of buyers who were influenced and never clicked.

Treat the 12% figure as an upper bound on how much of Google's work ChatGPT has taken. Ahrefs itself calls it an estimate, and other estimates differ.

The rest of this page separates three things that are easy to mix up: how many people ask, how many click, and how many become leads.

• 12%ChatGPT's search-like volume against Google's, an estimateAhrefs, February 2026
• 190xmore visits to websites come from Google than from ChatGPTAhrefs, 76,000 sites, February 2026
• 96%lower estimated click-through rate for ChatGPT than for GoogleAhrefs, February 2026
• 1.08%of all website traffic comes from AI referralsConductor, 10 industries, 2026

• Google shows 13.7 billion searches a day in Ahrefs' model. ChatGPT is about 2.5 billion prompts a day, of which Ahrefs counts 65% as search (Ahrefs, February 2026).
• Click-through estimate: about 29.2% for Google and 1.3% for ChatGPT, from Ahrefs' formula of traffic share divided by daily queries (Ahrefs, February 2026).
• Conductor's sample: 13,770 domains, 17 million AI responses and more than 100 million citations (data last updated July 2026).
• AI Mode, Google's chat-style search, handled 0.34% of US Google searches in January to April 2026 (SparkToro).
• AI Mode, Semrush clickstream of almost 69 million US desktop sessions, May 1 to July 5, 2025: 92% to 94% of AI Mode searches were zero-click, and 6% to 8% led to a visit to an outside site.
• AI Mode's share of Google search sessions rose about 4 times, from 0.25% in early May to just over 1% in early July 2025 (Semrush).
• AI Mode queries averaged 7.22 words, against 4.0 for traditional Google search (Semrush, 2025).
• The average AI Mode user had about one daily session of 2 to 3 queries (Semrush, 2025).
• Semrush, from a separate study with different methods: about 34% zero-click on Google search without an AI Overview, about 43% with one, and about 93% in AI Mode.

How often does Google show an AI Overview?

It depends on who counts. Studies range from 16% to 48% of searches, because each one measures a different set of searches.

Pew watched what real people searched, in a browser panel. BrightEdge tracks a chosen keyword set. Semrush counts 10 million keywords. Read each number as a view of its own sample, not as one truth.

What this means for you: find out where your own searches fall. Google Search Console now reports clicks from AI features inside the Web search type, so your own data answers the question for your site.

If your buyers search with long questions, you sit at the high end of the range. If they search a service plus a city, you sit nearer the low end, and the map pack matters more.

• Semrush, by topic, November 2025: science 25.96%, computers and electronics 17.92%, people and society 17.29%. Real estate, shopping and arts and entertainment each under 3%.
• Semrush: food and drink grew fastest, up 7.25 percentage points since March 2025.
• Semrush, by intent: AI Overviews on informational queries fell from 91.3% of those shown in January to 57.1% in October 2025, while commercial queries rose from 8.15% to 18.57% and transactional from 1.98% to 13.94% (October 2024 to October 2025).
• Google's own rule: a page must be indexed and eligible for a snippet to appear. There are no additional technical requirements (Google Search Central, 2026).

Which searches get an AI answer?

Long, specific questions and comparisons. Pew found that 8% of one and two word searches produced an AI summary, against 53% of searches of ten words or more.

Seer Interactive found the same by type. Comparison searches (X versus Y) showed an AI Overview 95.4% of the time. Plain buying searches almost never did.

What this means for you: list the ten searches that bring your best leads. Sort them into comparison, question and ready-to-hire searches. The first two groups are where an AI answer now sits above your result.

The ready-to-hire group is the safest today. It is also the group where being named in the AI answer for the earlier searches decides whether your name gets typed at all.

• Pew: 36% of searches with both a noun and a verb produced an AI summary (March 2025).
• Seer: average share of searches with an AI Overview was 36% of informational, 8% of commercial and 5% of transactional (full-year 2025).
• Pew: 60% of searches that began with a question word produced an AI summary.

What happens to clicks when an AI answer appears?

Fewer people click. In Pew's panel, searchers clicked a regular result on 8% of visits when an AI summary showed, and on 15% when it did not.

Only 1% of visits clicked a link inside the summary. And 26% of visits with a summary ended the browsing session, against 16% without.

What this means for you: plan for the lower end of the range on question and comparison searches, and for no loss on ready-to-hire searches. Then check your own Search Console trend since 2024 for pages that carry a question.

The recovery in the Seer series is real but modest. Click-through on AI Overview searches reached 2.36% in February 2026, and searches without an AI Overview sat at 3.82%. The gap is still about 38%.

Read the cited and uncited gap with care. Seer cannot say whether being cited causes the lift or whether stronger brands are both cited more and clicked more.

• 8%of visits click a result when an AI summary showsPew Research Center, 2025
• 15%click when no AI summary showsPew Research Center, 2025
• 1%of visits click a link inside the AI summaryPew Research Center, 2025
• 26%of visits with a summary end the session, against 16% withoutPew Research Center, 2025

• Seer, paid CTR on searches with an AI Overview: 14.64% in January 2025, 16.21% in February 2026. Without one: 25.98%, then 21.85%.
• Seer, informational searches: cited brands earned 2.07% organic CTR, uncited brands 0.94%, searches with no AI Overview 3.35%.
• Seer, per million impressions: about 33,500 clicks with no AI Overview, about 20,743 for a cited brand, about 9,445 for an uncited brand.
• Seer caveat: one account made up 47% of the uncited segment. Without it that segment's CTR is about 1.61%, not 0.94%.
• Ahrefs, other positions: the drop runs from 19.4% at position 10 to 50.8% at position 2.
• Google says clicks from AI features count in Search Console under the Web search type (Google Search Central, 2026).

How many Google searches end with no click at all?

More than two thirds. SparkToro's January to April 2026 study found 68.01% of US Google searches ended without a click to any result.

The trend is up. SparkToro puts the figure at 60.45% in 2024, 49% in 2019 and about 45% a decade ago. Its own note says the panels differ, so read the trend as a direction.

What this means for you: a rising no-click rate does not mean your phone stops ringing. Some of those searches are answered by a map, a call button or a number in the result. Count calls and form fills, not only sessions.

It does mean that clicks are a weaker measure of demand than they were. Track leads by source and brand searches, and use sessions as a supporting number.

• 68.01%of US Google searches ended with no click, January to April 2026SparkToro, June 2026
• 60.45%ended with no click in 2024SparkToro with Datos, 2024 baseline
• 9.51percentage points fewer searches led to any click, 2024 to 2026SparkToro, 2026
• 0.34%of searches moved into AI ModeSparkToro, 2026

• SparkToro, source: Similarweb desktop and mobile web panel, US, January to April 2026. The Google app is excluded, where zero-click may be higher.
• SparkToro: searches that led to another search rose 7.2 points from 2024 to 2026.
• Untraced arithmetic: SparkToro's post calls the rise from about 45% to 68% a 33.8% increase. 23 points on a base of 45 is about 51%. Quote the points, not the percentage.
• A 2024 baseline of 58.5% appears in earlier coverage of the same series. SparkToro's 2026 post uses 60.45%. Check which one a page quotes.
• Not all of the rise is AI. SparkToro says AI Overviews are likely contributing but did not isolate their share.

How much traffic do AI assistants send, and does it convert?

Small today, growing fast, and it converts well in retail. Adobe measured AI-referred traffic to US retail sites up 138% in May 2026 against a year earlier, and up 1,324% since it began tracking in October 2024.

Those visitors converted 54% better than other visitors. A year earlier they converted at about half the rate. Time on site was 53% higher and pages per visit 23% higher.

What this means for you: the channel is small enough that one tracking mistake hides it. A few dozen visits a month can contain several leads, and they can sit in the direct or referral bucket.

Be careful with the best numbers. Adobe's figures are for US retail, and your business may not be. Your own conversion data decides.

• 138%growth in AI-referred visits to US retail sites, May 2026 against May 2025Adobe, June 2026
• 54%better conversion than non-AI visits, US retail, May 2026Adobe, June 2026
• 53%longer time on site for AI-referred visitorsAdobe, June 2026
• 6.8%of US ChatGPT prompts carried a citation in May 2026, up from 1.6% in June 2025Similarweb, July 2026

• Adobe: more than 1 trillion visits to US retail sites in its sample (reported June 17, 2026).
• Adobe: AI-referred visitors were 15% more engaged than other visitors (May 2026).
• Ahrefs: AI search visitors viewed about 50% more pages than organic visitors, with a lower bounce rate but shorter visits (June 2025).
• Ahrefs' own models: AI search users click links about 75% less than organic search users.
• Conductor: AI referrals were 1.08% of traffic in 10 industries, and ChatGPT sent 87.4% of them (data from May to September 2025).
• Why it is hard to see: AI visits often arrive without a clear source label. The measurement guide shows how to separate them.
• Semrush: the average AI search visitor is worth 4.4 times the average organic search visitor, by conversion rate. It is a projection from 500+ topics, not a measured sales result (July 2025).
• Semrush's extrapolation: AI search could pass traditional search as a source of visitors by early 2028. It is a forecast, not a measurement.
• Semrush: ChatGPT weekly active users grew 8 times from October 2023 to April 2025, to over 800 million.
• Semrush: pages ChatGPT cites rank in position 21 or lower for related queries almost 90% of the time, and 50% of links in ChatGPT 4o answers point to business and service sites.
• Cloudflare: AI-referred consumers convert at 3 to more than 5 times the rate of traditionally referred consumers. Cloudflare quotes this without naming its source, so treat it as unverified.

Which sources do AI engines cite?

Mostly other people's pages, not yours. Muck Rack analysed more than 25 million links from ChatGPT, Claude and Gemini across 17 industries. Earned media was 84% of the citations, and journalism alone 27%.

Ranking well still helps, but less than most owners think. How much depends on how you count, as the chart shows.

What this means for you: the list of sites AI cites for your category is your marketing plan. Ask the five buyer questions, copy the cited sites, and work out which of them could name you.

Professional services sit at the low end of the Similarweb series. That means fewer answers carry a link at all, so being named in the text of the answer matters more than being linked.

• 84%of AI citations are earned mediaMuck Rack, May 2026
• 27%of cited sources are journalismMuck Rack, May 2026
• 37.9%of AI Overview citations rank in the Google top 10; 31.0% rank outside the top 100Ahrefs, March 2026

• Ahrefs, March 2026: 31.2% of cited URLs rank 11 to 100. Counting blue links only, 37.1% are in the top 10 and 36.7% outside the top 100.
• Ahrefs, July 2025: 76.10% of cited pages ranked in the top 10 (top three cited URLs per answer, 1.9 million citations).
• Why the two Ahrefs numbers differ: the 2025 study looked at the top three cited links in each answer. The 2026 study counts every cited URL across all result blocks, on 863,000 searches.
• BrightEdge: the overlap between cited sources and the organic top 10 held near 17% throughout (16.1% to 16.9%, February to July 2025).
• Muck Rack: the share of earned media has ranged from 82% to 89% across its editions.
• Similarweb: the industry citation figures cover US ChatGPT prompts and exist to show spread, not a promise for your sector.
• Freshness, Ahrefs Brand Radar, 16.975 million cited URLs (July 2025): AI assistants cited pages published 1,064 days earlier on average, organic results 1,432 days, 25.7% older.
• Last updated: 909 days for AI assistants against 1,047 for organic results, a 13.1% gap (Ahrefs, 2025).
• Days since published by platform: ChatGPT citations 958, ChatGPT references 1,023, Copilot 1,056, Gemini 1,118, Perplexity 1,166 (Ahrefs, 2025).
• Muck Rack October 2026: when the same prompt ran twice in one day, the cited domains repeated 41% of the time in ChatGPT, 45% in Gemini and 94% in Perplexity.
• Muck Rack October 2026: only 212 domains were cited by all four providers in its September analysis of 1,020 prompts.

What predicts whether AI names your business?

Being talked about in places AI reads. Ahrefs studied 75,000 brands. YouTube mentions of a brand correlated at about 0.737 with AI visibility, the strongest signal it found.

Branded web mentions came next, at 0.66 to 0.71. Page count on the brand's own site barely mattered, at about 0.194. Correlation is not cause, and larger brands collect more of everything.

What this means for you: the signals are all reputation signals. Mentions in video, mentions across the web and searches for your name move together with AI visibility. Page count and markup do not.

That fits how these systems work. An assistant that answers a question looks for pages that agree about who the options are. One source saying it is you is a claim. Several independent sources saying it is a fact.

• OpenAI: sites that opt out of OAI-SearchBot are not shown in ChatGPT search answers. GPTBot, the training crawler, is a separate setting.
• OpenAI: changes to robots.txt can take about 24 hours to apply.
• Primal summarising Bing's guidelines: Bing search and Copilot share one base for crawling, indexing and ranking. Duplicate pages make Bing less sure which one to cite.
• Ahrefs: backlinks showed a weak correlation with AI visibility in the 75,000-brand study.
• The fuller method is in how to get cited by AI, and crawler setup is in AI crawler access.

What does Megawebvision see in its own search data?

Our search impressions grew from 825 to 10,572 in a month, 12.8 times, led by statistics pages. This is Google Search Console for megawebvision.com, October 8 to November 3, 2026, against the prior 28 days.

Google showed our pages for 481 distinct queries. 160 of them, one in three, were 8 words or longer, or used search operators such as site: and after:. That pattern matches AI assistants querying the web for a user. It is an observation, not proof.

What this means for you: the same pattern may already be in your Search Console. Filter queries by length and by operators, and see how many look like an assistant at work.

We do not claim these queries prove citations. They show that automated tools read pages like ours, and that sourced statistics pages are what they find first.

• 12.8xgrowth in search impressions in one month, 825 to 10,572Google Search Console, megawebvision.com, Oct 8 to Nov 3, 2026 against the prior 28 days
• 1 in 3of 481 distinct queries were 8+ words or used operators (160)Google Search Console, same window
• 58%of all impressions came from our five statistics pagesGoogle Search Console, same window

• Reading: long, precise queries and search operators are how a tool searches for a person. People type short.
• Reading: pages that carry sourced numbers get shown for those queries. This page follows the same rule: every number links to its source.

How do B2B buyers use AI to pick a vendor?

Half of software buyers now start in a chatbot. G2 surveyed 1,076 B2B software buyers in March 2026. 51% begin research with an AI chatbot more often than with Google, up from 29% in April 2025.

The chatbot also changes the outcome. 69% chose a different vendor than they planned after chatbot guidance, and 33% bought from a vendor they had never heard of. If you sell to businesses, this is your shortlist.

What this means for you: the buyer arrives pre-sold or pre-rejected. A chatbot named your competitor, and the first call starts from that answer.

• 51%of B2B software buyers start research in an AI chatbot more often than in Google, up from 29%G2, 1,076 buyers, March 2026
• 69%chose a different vendor than planned after chatbot guidanceG2, March 2026
• 33%bought from a vendor they had never heard ofG2, March 2026
• 64%often or very often see inaccurate chatbot recommendationsG2, March 2026

• G2 ranks AI chatbots as the top influence on which vendors make a buyer's shortlist (March 2026).
• G2's buyers distrust the answers and use them anyway: 64% often see inaccurate recommendations.
• Gartner's 69% shows the human step remains. Buyers check the chatbot's answer with a person.

What should an owner do about AI search this quarter?

Four moves: be findable, be named, answer directly, measure leads. Start with the cheap checks. One blocked crawler can remove you from an entire assistant.

Then run the Ask-Five test below. It takes ten minutes and shows where you stand today. The self-check after this section turns your answers into a score.

The order matters. Crawler access and measurement cost an afternoon each. Mentions and video take a quarter and carry the highest weight in the self-check.

1. Write the five questions a buyer types before hiring a business like yours. Include one comparison and one price question.
2. Ask each in ChatGPT, Gemini and Google. Use a fresh session for each.
3. Note whether your business is named, and which sites are cited.
4. Repeat each question twice more on another day. Answers change from run to run (SparkToro and Gumshoe, 2026).
5. Count your mentions out of the total. Save the cited sites as your target list.