AI Search Statistics for 2026: 120 Sourced Numbers, Where They Disagree, and What They Mean for B2B SaaS

One study says Google shows an AI Overview on 18% of searches. Another says 88%. Both are right, because they counted different things: real people’s searches in one case, a hand-picked set of commercial keywords in the other. A lot of AI search statistics travel around without that context, which is how two slides in the same meeting end up contradicting each other.

This page collects 120 AI search statistics, each with who measured it, how, and when, checked against the original publisher as of October 2026. They’re grouped by topic and then by publisher, so you can see which company found what. There’s a section on where the big numbers disagree and why, and a section of first-party data from my own Search Console and AI tracking that you won’t find anywhere else.

I’m a content strategist at Breaking B2B and a B2B SaaS writer, and I track how my own site shows up in Google and in AI answers. I update this page every month, and the log at the bottom shows what changed.

AI search statistics at a glance

Ten numbers that come up most often, with what each one measured:

Statistic	Publisher	What was measured
ChatGPT reaches about 1.2 billion weekly users	OpenAI, Sept 2026	Company-reported weekly users
Google AI Overviews have 2.5 billion+ monthly users	Alphabet, June 2026	People shown an AI Overview
49% of US adults use AI chatbots	Pew Research Center, June 2026	Survey of 5,119 US adults
68.01% of US Google searches end without a click	SparkToro and Similarweb, June 2026	Clickstream panel, Jan–Apr 2026
AI Overviews cut position-1 click-through rate by 58%	Ahrefs, Feb 2026	300,000 keywords, desktop
AI assistants send 0.43% of website traffic	Ahrefs, Oct 2026	115,254 sites, Sept 2026
AI Overviews appear on 88% of commercial keywords	Semrush, Oct 2026	400 commercial keywords
51% of B2B software buyers start research in a chatbot more often than in Google	G2, Apr 2026	Survey of 1,076 B2B buyers
43% of pages cited in AI Overviews rank in Google’s top 10	Conductor, Sept 2026	167.9 million citations
Tracker prompts made up 53% of my long-query impressions	My Search Console, Oct 2026	Top 1,000 queries of 8+ words

How to read an AI search statistic

Four checks will tell you whether two numbers can be compared at all.

What was counted. Sessions, web visits, referral clicks and searches are different units. Graphite puts AI at 56% the size of search, counting sessions in apps and on the web, while Statcounter puts it at 0.4%, counting clicks from chatbots to websites. Both are correct for what they measured.

Whether it’s a survey or a measurement. A survey tells you what people say they do. A clickstream panel or a site’s own analytics tells you what they did. Pew’s finding that 60% of US adults read AI summaries is self-reported, and its finding that people click less when one appears comes from tracked browsing.

How keywords were weighted. Counting every keyword once favors the long tail, where AI Overviews are rarer. Weighting by search volume favors the popular, commercial searches where they’re common. That single choice explains much of the gap between 6.5% and 88%.

Who paid for the study. AI visibility tools publish a lot of the research on citations, and their findings often support the product they sell. That doesn’t make the data wrong, but it’s worth knowing, so each publisher block below notes it.

How many people use AI search

These numbers describe how many people use AI tools, how they use them, and how that compares with search.

OpenAI

• 1.2 billion. At DevDay in late September 2026, OpenAI described ChatGPT as reaching “our collective 1.2B weekly users”. It’s a company-reported figure, and OpenAI publishes weekly users only.
• 14% to 24%. “Seeking information” grew from 14% to 24% of all ChatGPT use between July 2024 and July 2025, in a paper by OpenAI and Harvard economists published through NBER. The authors call that kind of use a very close substitute for web search. The figure comes from automated classification of consumer conversations.

Google and Alphabet

These are company-defined counts. An AI Overview “user” is anyone shown one, and they appear automatically.

• 2.5 billion+. AI Overviews reach more than 2.5 billion users a month, Alphabet said in June 2026.
• 1 billion+. AI Mode passed 1 billion monthly active users after its global expansion in October 2025, Alphabet said in its Q2 2026 earnings remarks. By Q3, AI Mode queries were three times longer than traditional searches on average, and one in five used voice, images or documents.
• 1 billion+. The Gemini app passed 1 billion monthly active users in Alphabet’s Q3 2026 results on October 28, up from 950 million a quarter earlier, and Alphabet called it the fastest-growing product in Google’s history. Gemini is built into Android and Workspace, which raises its reach compared with a standalone app.
• 5 trillion+. Google handles more than 5 trillion searches a year, its first official figure since “2 trillion” in 2016. In Q3 2026, Alphabet said year-over-year query growth was the highest in five years, though it doesn’t publish the rate.

Pew Research Center

All four come from Pew’s February 2026 survey of 5,119 US adults, published in June 2026 (margin of error ±1.6 points).

• 49% of US adults use AI chatbots. Pew’s 2024 figure was 33%, but the question was worded differently, so the two aren’t a clean trend.
• 44% of US adults use ChatGPT, up from 34% in 2025, ahead of Gemini (24%), Copilot (17%), Meta AI (14%), Grok (8%) and Claude (6%).
• 60% say they read the AI summaries at the top of search results, while 30% say they don’t.
• 42% use chatbots to search for information. Among chatbot users, that’s 86%, the most common use.

Graphite

Graphite’s numbers count AI sessions in apps and on the web, including coding and writing tasks, so they’re a ratio to search rather than a share of it.

• 56%. Monthly AI sessions are 56% the size of search sessions worldwide and 34% in the US, mostly measured in Q4 2025.
• 28%. Counting only search-like questions, AI equals 28% of search worldwide and 17% in the US, which the author treats as an upper bound.
• 89% to 71%. Google’s share of combined search and AI traffic fell from 89% in 2023 to 71% in Q4 2025, and ChatGPT accounts for 89.4% of AI sessions.

SparkToro and Datos

• 3.2%. AI tools made up 3.2% of US desktop searches in Q4 2025 across 41 search-heavy sites, while Google alone had 73.7%. The data is desktop only, which leaves out the apps where most AI use happens.

Statcounter

Statcounter measures clicks from chatbots to websites, so answers that produce no click don’t show up.

• 80.48%. ChatGPT sent 80.48% of AI chatbot referrals to websites worldwide in September 2026, ahead of Gemini (10.91%), Perplexity (5.85%), Copilot (1.79%) and Claude (0.95%).
• 0.4%. AI chatbots sent 0.4% of referral traffic worldwide in September 2026, against 91% from search engines. In the US, the figure was 0.19%.

Similarweb

• 9.5 billion. Generative AI platforms got 9.5 billion web visits a month in May 2026, up 70% year over year.
• 95%. 95% of ChatGPT users also use Google, unchanged between September 2025 and May 2026. That’s an audience overlap, not a measure of which searches moved.

Wix Studio AI Search Lab

• 27.4 billion. AI search platforms got 27.4 billion visits in Q1 2026, about 9.2% of Google’s traffic. AI grew 42.8% year over year, and Google grew 2.4%.

Semrush

• About 1 billion. ChatGPT’s US traffic levelled off near 1 billion visits a month from November 2025, and it ran a web search on 34.5% of prompts in February 2026, down from 46% in late 2024. Semrush’s clickstream panel covers mobile and desktop.

Bain & Company

• 80%. About 80% of consumers rely on AI summaries for at least 40% of their searches, and about 60% of searches end without a click, in Bain’s December 2024 survey of 1,117 consumers.
• 16%. Only 16% of consumers mostly or always use chatbots instead of search engines, in Bain’s September 2025 survey of 1,500.

Gartner

• 25%. In February 2024, Gartner predicted traditional search volume would drop 25% by 2026. It was a prediction, and the measured data so far doesn’t show it: Google says queries are at an all-time high, and Wix measured Google visits up 2.4% year over year.
• One-third. In a 2025 survey of about 370 US consumers, only about one-third thought chatbots were as good as search engines for learning, and 31% said AI summaries made them spend more time searching.

UK: UKOM and Ipsos iris

• 29.6 million. 29.6 million people in the UK used AI tools in July 2026, up 3.1 million in three months. It’s measured behavior (Ipsos iris), and it leaves out AI inside Google Search.

What it means for B2B SaaS. AI search is adding to Google rather than replacing it, since nearly every ChatGPT user still uses Google. Plan for both channels, and judge AI by whether you’re named in answers, because the click data captures only a small slice of its use.

How AI search affects clicks and traffic

Pew Research Center

These come from Pew’s tracked browsing of 900 US adults in March 2025 (68,879 Google searches). They describe real behavior rather than a keyword sample.

• 8% vs 15%. Users clicked a regular search result on 8% of visits when an AI summary appeared, against 15% when none did.
• 1%. They clicked a link inside the AI summary on 1% of visits.
• 26% vs 16%. They ended their browsing session on 26% of pages with an AI summary, against 16% of pages without one.

Ahrefs

• 58%. AI Overviews correlate with a 58% lower click-through rate for the position-1 page, comparing December 2023 with December 2025 across 300,000 keywords. Position-1 CTR on those keywords fell from 7.3% to 1.6%. The data is desktop only, and click-through rates fell on keywords without AI Overviews too.
• 0.43%. Across 115,254 sites, AI assistants sent 0.43% of traffic in September 2026, against 26.82% from search engines. ChatGPT alone sent 0.34%.
• 0.1%. In early 2025, AI sent 0.1% of referral traffic across about 35,000 sites, a smaller sample than the benchmark above.
• 23x. On Ahrefs’ own site, AI search was 0.5% of visits but 12.1% of sign-ups over 30 days in mid-2025, a 23x higher conversion rate than organic search. That’s one company with an unusually strong brand.
• 67.8% vs 63.7%. Across about 82,000 sites, AI visitors bounced more often than search visitors and viewed fewer pages (4 against 5.2).

Seer Interactive

Both figures come from Seer’s April 2026 study of 53 brands and 5.47 million queries.

• 1.31% to 2.36%. Organic click-through rate on queries with an AI Overview fell to 1.31% in December 2025 and recovered to 2.36% in February 2026. That’s still below 3.82% on queries without one. Seer calls the rebound directional.
• 2.07% vs 0.94%. On informational queries in 2025, brands cited in the AI Overview earned a 2.07% click-through rate, against 0.94% for brands that weren’t cited.

Google

• “Relatively stable.” Google says total organic click volume to websites has been relatively stable year over year. It published no numbers or method.

First Page Sage

First Page Sage is an agency, and these figures come from its own client and panel data.

• 22.6% vs 3.6%. Position-1 click-through rate is 22.6% with no AI answer and 3.6% when an AI answer appears, across 1,184 domains and 3.67 billion impressions (March–August 2026).
• 5.2%. The first source cited in an AI answer earns a 5.2% click-through rate.
• 4.7% vs 1.9%. In its 218-site panel, ChatGPT visitors converted to leads at 4.7%, against 1.9% for Google organic visitors. The same report puts AI at 6.24% of sessions, far above every other source here.

SparkToro

• 68.01%. 68.01% of US Google searches ended without a click in January–April 2026, using Similarweb panel data.
• 58.5%. In 2024, 58.5% of US Google searches ended with zero clicks, from a different panel (Datos), so SparkToro calls the comparison “a bit of apples and oranges.”

Semrush

• 92–94%. Google AI Mode sessions were zero-click 92–94% of the time in mid-2025, against about 34% for searches without an AI Overview.
• 33.75% to 31.53%. When the same keywords gained an AI Overview, their zero-click rate fell slightly, from 33.75% to 31.53%. That runs against Pew, Ahrefs and Seer, and the clicks may have gone to links inside the AI Overview rather than to regular results.

Similarweb

• 770.7 million. AI platforms sent an average of 770.7 million referral visits a month worldwide between June 2025 and May 2026, up 117.4% year over year.
• 11.4% vs 5.3%. Similarweb estimates ChatGPT-referred ecommerce visits convert at 11.4%, against 5.3% for organic search. That’s a model estimate, not measured conversions.

seoClarity

• Under 1%. AI search is under 1% of organic traffic across hundreds of seoClarity client domains. One B2B SaaS site reached about 4.5% after a 127% rise in three months, which is a single example rather than a benchmark.

Microsoft Clarity

• 1.66% vs 0.15%. Across 1,277 domains, visitors from AI chatbots converted to sign-ups at 1.66%, against 0.15% for search. The sample leans toward publishers and news sites, and AI traffic is a small base.

Kaiser and Schulze, Marketing Science

• Under 0.2%. In a peer-reviewed study of 973 ecommerce sites, traffic from AI chatbots was under 0.2% of sessions and converted worse than organic search, though better than paid social. It’s the only independent, peer-reviewed conversion study here, and it covers ecommerce rather than B2B.

What it means for B2B SaaS. AI Overviews take clicks from informational pages first, and being the cited source wins some of them back. AI referrals are still a small share of traffic, so report AI search on visibility and assisted pipeline rather than referral volume.

How often Google shows AI Overviews

Published estimates run from 6.5% to 88%. The table puts them in order, with how each study counted:

Estimate	Publisher	How it was counted
6.49% (Jan 2025) → 24.61% (Jul 2025) → 15.69% (Nov 2025)	Semrush	10 million+ keywords, each counted once
18% of searches (Mar 2025)	Pew	68,879 searches by real US users
20.5% of result pages (Sept 2025)	Ahrefs	146 million desktop result pages
40%+ of US searches (2026)	Similarweb	Weighted by search volume
About 30% → 48% (Feb 2025 → Feb 2026)	BrightEdge	Tracked client keywords
73.3% branded, 78.3% non-branded (Sept 2026)	Ahrefs	232.8 million US desktop crawls
81.6% of US queries (2026)	First Page Sage	Volume-weighted, includes AI Mode
88% of commercial keywords (Aug 2026)	Semrush	400 commercial keywords

The spread comes from a few patterns.

• Question searches trigger them far more often. Pew found AI summaries on 60% of question-word searches, and Ahrefs found them on 57.9% of question queries against 9.5% of single-word queries.
• Commercial searches now trigger them almost every time. Semrush’s commercial panel went from 45.75% in September 2025 to 88% in August 2026, or 97% weighted by search volume. Software searches for “alternatives” and “best X for Y” hit 100%.
• The numbers move fast. Ahrefs measured a 34.5% click drop for position 1 in April 2025, and 58% in its February 2026 re-run.

When you quote AI Overview prevalence, give the denominator with it. A figure without one can be made to say almost anything.

What it means for B2B SaaS. your money keywords, such as alternatives, comparisons and “best X” searches, are the ones where Google now shows an AI Overview on nearly every result. Your highest-intent pages already compete with an AI answer above them.

What AI answers cite

A large share of the research on citations comes from companies that sell AI visibility tracking or SEO software, and each block notes it once.

Profound

Profound sells AI visibility tracking.

• 7.8%. Wikipedia was ChatGPT’s most-cited source at 7.8% of all citations, across 680 million citations from August 2024 to June 2025. The widely repeated 47.9% is Wikipedia’s share of ChatGPT’s top 10 sources only.
• #11 to #5. LinkedIn rose from about #11 to #5 among ChatGPT’s most-cited domains between November 2025 and February 2026, and became the most-cited domain for professional queries across six AI platforms.
• 1.5–2.5x. After an AI assistant mentions a brand, people visit that brand’s site at 1.5 to 2.5 times their expected rate over the next seven days, and only about 2.5% of those visits carried a trackable AI referral. Profound says this effect is the bet its product is built on.

Peec AI

Peec AI sells AI visibility tracking.

• 16.5 points. Being ranked #1 in a frequently cited third-party list raised a B2B SaaS brand’s chance of being mentioned in AI answers by 16.5 percentage points, across nearly 200,000 AI responses from eight engines. Peec says the result is observational.
• 8+ providers. ChatGPT draws search results from its own index plus at least eight outside providers, so Bing rankings aren’t a stand-in for ChatGPT visibility.
• 15.47% to 8.19%. After the GPT-5.6 rollout in August 2026, listicles fell from 15.47% to 8.19% of pages ChatGPT retrieved, and product pages took the lead at 16.39%.

Semrush

Semrush sells an AI visibility toolkit.

• 60% to 10%. ChatGPT cited Reddit in close to 60% of responses in early August 2025, then about 10% by mid-September, across 230,000+ prompts tracked weekly.
• 62%. In a 115-prompt study with Kevin Indig, 62% of citations linked a brand’s site without naming the brand in the answer. ChatGPT cited domains in 87% of appearances but named the brand in 20.7%.
• 82% vs 35%. Perplexity’s cited URLs overlapped Google’s top 10 by 82%, against about 35% for AI Mode, measured per query across 5,000 keywords.

Ahrefs

Ahrefs sells Brand Radar, the tool behind most of these figures.

• 0.737. Across 75,000 brands, YouTube mentions had the strongest correlation with AI visibility (about 0.737), followed by branded web mentions (0.66–0.71). The number of pages on a site showed almost none.
• 38%. 38% of pages cited in AI Overviews also ranked in the top 10 for the same query, and 31% ranked outside the top 100. Ahrefs’ July 2025 figure was about 76%, but it changed how it detects citations, so the two aren’t a clean trend.
• About half. ChatGPT cites about half of the URLs it retrieves, across 1.4 million prompts, and URLs from its separate Reddit channel were cited only 1.93% of the time.
• 16.8%. In September 2026, Reddit led ChatGPT citations in the US with a 16.8% share, ahead of Wikipedia at 7.0%.
• 22.9%. YouTube was the most-cited domain in AI Overviews in September 2026, at 22.9%, followed by Reddit (18.5%) and Facebook (10.1%). Both shares are of the top 50 sources, not of all citations.

SE Ranking

SE Ranking sells AI visibility tracking.

• 14%. AI Mode answers averaged 12.6 links, but only 14% of cited URLs matched the organic top 10, and repeated runs of the same query shared 9.2% of URLs (10,000 keywords, June 2025).
• 3x. Domains with profiles on review platforms such as G2, Capterra and Trustpilot had three times the odds of being cited by ChatGPT. That’s a correlation, and it likely overlaps with brand size.

BrightEdge

BrightEdge sells enterprise SEO and AI search software.

• 54.5%. 54.5% of AI Overview citations came from pages that rank organically, up from 32.3% at launch, and 71.0% in B2B tech.
• 16.7%. Only 16.7% of AI Overview citations came from top-10 results, with most of the growth from pages ranking 21 to 100.

Conductor

Conductor sells an enterprise answer engine optimization platform.

• 43%. Across 167.9 million AI Overview citations from June to August 2026, 43% of cited pages also ranked in the top 10. A page ranking #1 was cited 24.9% of the time, falling to 10.2% at #10. It’s the largest and most recent sample here.

AirOps

AirOps sells AI search content tools.

• 15%. ChatGPT cited only 15% of the 548,534 pages it retrieved, and 95% of its follow-up searches had no measurable search volume.
• 43.2%. 43.2% of pages ranking #1 in Google were cited by ChatGPT, and 55.8% of cited pages ranked in Google’s top 20 for at least one related query.
• 85%. In commercial discovery searches across B2B software categories, 85% of brand mentions came from third-party sites, and 13.2% from the brand’s own. Nearly 90% of those third-party mentions came from lists, comparisons or reviews.
• 30%. Only about 30% of brands stayed visible from one AI response to the next run of the same prompt.

Scrunch

Scrunch sells AI visibility software.

• 4.5 weeks. The average cited source lost half its citation activity in about 4.5 weeks, across 3.5 million citation events. ChatGPT cycled fastest (3.4 weeks) and Perplexity slowest (5.8 weeks). The study was run with Stacker, a content distribution company.
• 4% to 7%. When AI cited a vendor’s own “best tools” list ranking itself first, that vendor’s recommendation rate rose only from about 4% to about 7%.
• Reddit. Reddit looks dominant in aggregate studies mainly because it appears in every industry, and specialist sites beat it inside individual industries.

Otterly.ai

Otterly.ai sells AI visibility tracking.

• 66.6%. On buyer questions across seven AI engines in September 2026, listicles made up 66.6% of cited pages, and brands published 35.6% of those lists.

Search Atlas

Search Atlas sells SEO and AI visibility software.

• 21%. ChatGPT’s citations shared 21% of domains and 7% of URLs with Google’s results for matched queries. Perplexity shared 43% of domains.

Writesonic

Writesonic sells AI visibility tooling.

• 17%. For the same prompt, any two AI engines shared only about 17% of their cited domains, and 3.8% of domains were cited by all four engines studied.

seoClarity

seoClarity sells enterprise SEO software.

• 94% vs 56%. 94% of AI Overviews cited at least one top-20 result, but only 56% of all citations came from the top 20, across 5.1 million citations.
• 19%. Only 19% of AI Mode citations came from the top 20 organic results for the same query, in a 1,000-query sample.

Similarweb

Similarweb sells an AI citation analysis tool.

• 13.15% and 11.97%. Wikipedia and Reddit took about a quarter of US ChatGPT citations in January and February 2026.

Wix Studio AI Search Lab

• 40.86%. Listicles were the most-cited page type, at 21.9% overall and 40.86% on commercial prompts, across 75,000 AI answers. In professional services, 80.9% of cited lists were third-party rather than self-promotional.
• 8.2x. Wix sites with 50+ organic search visits a month were 8.2 times more likely to be found by AI tools than sites with little or no organic traffic.

Position Digital

Position Digital sells listicle link building, which these findings support.

• r = +0.64. Across 95 B2B SaaS brands, appearing in more category lists correlated with how often ChatGPT named the brand at r = +0.64, ahead of branded web mentions (+0.58).
• 66%. 66% of B2B SaaS brand recommendations in ChatGPT happened without ChatGPT citing the brand’s own site.
• 58%. 58% of pages in Google’s top 10 were also cited by ChatGPT in its B2B SaaS prompt set.

Seer Interactive

• 87%. In early 2025, 87%+ of ChatGPT search citations matched Bing’s top 20 results for the same question, against 56% for Google. This is the original source of the “87% matched Bing” claim, and it said top 20, not top 10. It used about 100 questions and predates ChatGPT’s own index.

What it means for B2B SaaS. most of the brand mentions in AI answers come from other people’s pages, especially lists, comparisons and review sites. Earn a place on those pages, keep your own comparison pages strong, and expect citations to shift within weeks.

How B2B buyers use AI search

G2

G2 runs a software review marketplace and sells AI visibility and review products. Its surveys are self-reported.

• 51%. 51% of B2B software buyers start research with an AI chatbot more often than Google, in a March 2026 survey of 1,076 buyers.
• 29%. In April 2025, the share was 29%, in a separate survey of 1,169 buyers.
• 71%. 71% rely on AI chatbots somewhere in software research, up from about 60% seven months earlier.
• 54%. AI chatbots were the #1 influence on software shortlists in March 2026 (54%), ahead of review sites (43%).
• 69%. 69% say an AI chatbot led them to a different vendor than planned, and 33% bought from a vendor they’d never heard of.
• 85%. 85% think more highly of a vendor that AI cites in its answer.
• 64%. 64% run into chatbot inaccuracies often or very often.
• 80%+. In June 2026, more than 80% had sourced software recommendations from a chatbot in the last two years.
• 38% vs 37%. In the same June survey, review sites (38%) and chatbots (37%) were the top two shortlist influences. The wording and sample differ from March, so the two aren’t a trend.
• 80% vs 65%. Buyers who used chatbots were more likely to buy from their initial shortlist (80% against 65%).
• 9%. Over 60% use or plan to use AI agents in buying, but only 9% are comfortable letting agents make purchases within approved limits.

6sense

6sense sells account-based marketing software.

• 94%. 94% of B2B buyers used large language models during their buying journey, mainly to summarize reviews or analyze data, yet still averaged 16 interactions per person with the winning vendor, unchanged from the year before. The survey, published in November 2025, covered more than 4,000 buyers across North America, EMEA and APAC.
• 77%. Buying groups ranked preferred vendors before first contact and bought from that favorite 77% of the time.

Forrester

• 94%. 94% of business buyers use AI in their buying process, up from 89%, in Forrester’s 2025 Buyers’ Journey Survey. Twice as many named generative AI or conversational search as more meaningful than any other source.

Gartner

• 45%. In a 2025 survey of 645 B2B buyers, 45% used generative AI during a recent purchase, and 69% prefer to check AI-generated insights with a sales rep.

TrustRadius and HG Insights

TrustRadius sells buyer-intent and review products.

• 63%. 63% of technology buyers used AI in their purchase journey, in a January 2026 survey of 1,862 buyers. Of those, 94% fact-check its answers at least some of the time.
• 2%. Only 2% of buyers trust AI output “always”, while 54% say it made research easier, up from 40%.

Responsive

• A quarter. In a 2025 survey of 350 B2B buyers, generative AI had overtaken search for a quarter of buyers, and nearly two-thirds used it as much as or more than search when researching vendors.

Wynter

• 84%. 84% of mid-market B2B SaaS CMOs use AI tools for vendor discovery, up from 24% in 2025, and 68% start with AI before search engines. The sample was 101 CMOs, and Wynter calls the year-on-year comparison directional.

Semrush

All three come from Semrush’s March–April 2026 survey of 519 US B2B professionals who use AI at work. Leaving out non-AI users raises the percentages compared with all-buyer surveys.

• 92% say AI shaped their vendor shortlist, and 83% say it influenced the final decision.
• 41% start with AI and check with search, while 35% start with search and turn to AI. 75% still use Google in vendor research.
• 7% say brand recognition decides whether they notice a vendor in an AI answer, while 53% say a close match to their use case does.

Demandbase

• 303%. Monthly ChatGPT referrals to B2B websites on Demandbase’s platform rose from about 645,000 to 2.6 million between June 2025 and June 2026. It’s measured traffic across 1,584 customer instances, not a survey.

Marketers

• 94%. 94% of enterprise CMOs and digital leaders plan to increase spending on answer engine optimization in 2026, in a Conductor survey of 250+ people. Conductor sells the platform, and the respondents work in the field.
• 37.7%. AI chatbots top the list of channels where marketers plan to increase investment in 2026 (37.7%) in HubSpot’s survey of 1,500+ marketers, just ahead of paid social (37.4%) and video (37.1%).

What it means for B2B SaaS. buyers use AI to build the shortlist, then check it with search, review sites and sales conversations. Being named in the AI answer gets you onto the list, and the page the buyer lands on has to close the case.

Where the studies disagree, and why

Five numbers vary widely depending on the study. Here’s what’s behind each spread.

How often AI Overviews appear: 6.5% to 88%

Counting keywords once (Semrush, Ahrefs) gives low numbers, because the long tail rarely triggers AI Overviews. Weighting by volume (Similarweb, First Page Sage) or tracking commercial keyword sets (BrightEdge, Semrush’s commercial panel) gives high ones. Timing plays a part too. Coverage peaked in July 2025, dipped, then expanded sharply in 2026. The full table is in the AI Overviews section above.

How big AI is next to search: 0.4% to 56%

Statcounter’s 0.4% counts clicks from chatbots to websites. SparkToro’s 3.2% counts US desktop searches. Wix’s 9.2% compares web visits. Graphite’s 56% compares all AI sessions, in apps and on the web, with search sessions. None of them is “AI’s share of search,” and quoting any of them that way is misleading.

How many B2B buyers start with AI: 25% to 68%

Responsive found a quarter of buyers had moved from search to AI. Semrush found 41% of AI-using buyers start with AI. G2 found 51% start in a chatbot more often than Google. Wynter found 68% of SaaS CMOs start with AI. Each asked a different question of a different group, and the narrower and more AI-friendly the group, the higher the number. How many buyers use AI at all varies the same way: Gartner 45%, TrustRadius 63%, G2 71%, Forrester and 6sense 94%.

Whether cited pages rank in Google’s top 10: 16.7% to 43%

BrightEdge (16.7%), Ahrefs (38%) and Conductor (43%) measure the share of citations that rank in the top 10. Semrush’s 82% overlap for Perplexity measures whether any cited URL also ranks, per query, which always comes out higher. seoClarity’s 56% uses the top 20. Mixing those measures causes most of the disagreement.

Whether AI visitors convert better

Vendor data says yes: Ahrefs measured 23x on its own site, Microsoft Clarity found 1.66% against 0.15% for sign-ups, First Page Sage found 4.7% against 1.9%, and Similarweb’s model estimates 11.4% against 5.3%. The one peer-reviewed study, Kaiser and Schulze’s analysis of 973 ecommerce sites, found AI traffic converted worse than organic search. Ahrefs also found AI visitors bounce more and view fewer pages. For B2B, there’s no independent study of pipeline from AI visitors yet.

What my own Search Console and AI tracking show

These are first-party numbers from my site, nathanojaokomo.com, which publishes B2B SaaS content and tool reviews. They cover the three months to October 23, 2026.

Search Console’s Generative AI features report

Search Console now has a Generative AI features report, still in beta, showing impressions in Google’s AI features.

• 741,000 impressions in AI features over three months, about 24% of my site’s 3.08 million total Search impressions.
• 209 pages appeared in AI features. Tool and agency roundups lead: my Rankshift review had 165,113 impressions, best AI visibility tools 93,897, best rank tracking tools 72,712, and best local SEO tools 62,598.
• The report shows impressions only, without clicks.

AI visibility trackers are inflating Search Console impressions

Queries of eight words or more brought my site 470,000 impressions and 5 clicks in three months. When I read the top 1,000 of those queries, a large share weren’t written by people.

• 120 of the top 1,000 long queries were prompts run by AI visibility tracking tools. They give themselves away with instructions such as “you must provide a forced ranking from best to worst,” often in Spanish or German, or “search the web and give me actual names.”
• 172,395 impressions, 53% of the impressions in those 1,000 queries, came from those 120 prompts.
• 0 clicks came from them.
• The prompts named tools such as HubSpot AEO, Rankshift, Profound, Peec AI, AthenaHQ, Otterly.ai, Scrunch, Semrush’s AI toolkit and Ahrefs Brand Radar, which suggests they come from tools tracking those brands.

I classified the prompts with a conservative pattern, so the real share is likely higher. If your pages cover AI search or marketing tools, your impressions and click-through rate are probably affected too. To check, add a query filter in Search Console, choose “Custom (regex),” set it to “doesn’t match,” and paste:

forced ranking|ranking forzado|erzwungen|you must provide|debes proporcionar|du musst|search the web|look online|search online

The same prompts score differently by engine and country

I track 100 buyer prompts about hiring B2B SaaS writers in SE Visible, in the US and the UK, across five AI engines. In the first check on October 24, 2026:

• 9% to 17%. My visibility ranged from 9% in Gemini to 17% in Google AI Overviews, with ChatGPT, Copilot and AI Mode at 15–16%.
• 80% vs 24%. The prompt “Recommend a freelance writer for B2B SaaS SEO content” named me in 80% of US answers and 24% of UK answers.

One brand and one check don’t make a benchmark. But it shows why your AI visibility number depends on the engine, the country and the tool, and why a fixed set of prompts tracked over time tells you more than a comparison with someone else’s figure.

What the numbers mean for a B2B SaaS team

These rules follow from the data above.

1. Start with commercial searches and prompts. Google shows an AI Overview on nearly every software alternatives and “best X” search (Semrush), so that’s where the AI answer is already in front of your buyers.
2. Judge AI search by visibility and assisted pipeline, not referral traffic. AI sends well under 1% of traffic to most sites (Ahrefs, Statcounter), even as ChatGPT referrals to B2B sites grew 303% in a year (Demandbase).
3. Get onto the pages AI cites about your category. 85% of brand mentions in commercial AI answers come from third-party sites (AirOps), and being #1 on a cited list raises the chance of a mention (Peec AI). Review site profiles help too (SE Ranking).
4. Keep comparison and alternatives pages current. Cited sources lose half their citations in about 4.5 weeks (Scrunch), so a page that was cited last quarter may not be now.
5. Read Search Console impressions on AI topics carefully. Tracking tools’ prompts can make up a large share of long-query impressions, which drags down click-through rate without any change in what real buyers do.
6. Expect AI visibility to differ by engine and country. Any two engines share only about 17% of cited domains (Writesonic), so a single “AI visibility score” hides a lot.

If you’re starting from zero, getting your brand into ChatGPT and Claude answers follows the same order, and AI visibility tools can track a fixed prompt set for you once checking by hand gets slow.

How I chose these statistics

I started with 156 statistics and kept 120. Each one links to the original publisher: the company or researcher that ran the study, not an article that repeated it. I checked the key numbers against the live source pages in October 2026.

I left out:

• forecasts with no published method
• claims with no source
• figures I could only find second-hand
• vendor statements with no data behind them

Where a company sells the product its finding supports, the publisher block says so. Dates refer to when the data was collected where the publisher gives it, and to the publish date otherwise.

If you spot a number that has changed or a study I should add, email me and I’ll update the page.

Update log

• October 2026: first version, 120 statistics.

AI search statistics FAQs

How many people use AI search?

OpenAI said in September 2026 that ChatGPT reaches about 1.2 billion weekly users, and Google says AI Overviews reach more than 2.5 billion people a month. In the US, 49% of adults use AI chatbots and 42% use them to search for information, in Pew’s February 2026 survey of 5,119 adults.

How often do AI Overviews appear in Google?

It depends on what’s counted. Estimates range from 6.5% of keywords counted once (Semrush, January 2025) to 88% of commercial keywords (Semrush, August 2026). Pew found AI summaries on 18% of real US searches in March 2025, and 60% of question searches.

Do AI Overviews reduce clicks?

Yes, in most studies. Pew found users clicked a result on 8% of visits with an AI summary against 15% without one, and Ahrefs measured a 58% lower click-through rate for position 1 on keywords with an AI Overview. Semrush is the exception, finding a slight drop in zero-click rates.

How much traffic does AI search send to websites?

Very little so far. AI assistants sent 0.43% of traffic across 115,254 sites in September 2026 (Ahrefs), and AI chatbots sent 0.4% of referral traffic worldwide (Statcounter). Referrals are growing fast, though: ChatGPT referrals to B2B sites rose 303% in a year (Demandbase).

Do AI search visitors convert better?

Vendor studies say yes, from 3x to 23x, but most measure one site or estimate the rate. The only peer-reviewed study, of 973 ecommerce sites, found AI traffic converted worse than organic search. There’s no independent B2B pipeline study yet, so treat conversion claims with care.

Get named when buyers ask AI

If you want your product named when buyers ask AI which tool to pick, book a call with me.