AI Search in 2026: The Verified Numbers, and Where They Disagree

There is no shortage of statistics about AI search in 2026 — the problem is that a lot of them contradict each other. Different research firms measure different things (installed-base usage vs. referral traffic vs. revenue vs. regional samples), and the headline numbers you see quoted often depend entirely on which of those four things is actually being measured. This is an attempt to lay out the most-cited figures honestly, including the places where they conflict, rather than picking whichever number tells the cleanest story.

How many people are actually using this

The underlying adoption numbers are not in serious dispute, even if market-share percentages built on top of them are. ChatGPT crossed 900 million weekly active users in early 2026 (SERPs.io / DemandSage, 2026), recording 5.72 billion visits in January 2026 alone. Its mobile app has roughly 557 million monthly active users across iOS and Android — nearly eight times Google Gemini's approximately 70 million app users (DataReportal, cited in SERPs.io, 2026). Perplexity, while much smaller in absolute terms, has grown the fastest: over 45 million monthly active users, roughly 170 million monthly visitors, and an estimated 1.2–1.5 billion search queries processed per month as of mid-2026, alongside reported 800% year-over-year usage growth (SEOProfy / DemandSage, cited in SERPs.io, 2026). At the platform level, Google itself reported at its I/O 2026 conference that AI Mode had surpassed 1 billion monthly users, with query volume more than doubling each quarter (Search Engine Land, 2026).

Market share: pick your methodology, get a different answer

This is where the numbers genuinely diverge, and it's worth understanding why rather than just picking a favorite. Depending on the source and what's being measured, ChatGPT's share of the AI-chat category in 2026 has been reported as: 60.7% (Similarweb, January 2026, cited in Presenc AI), 76.85% (StatCounter, April 2026, cited in ReporterOutreach), 64.5–68% (FirstPageSage/Graphite/Similarweb composite, March 2026, cited in Stackmatix), and 81% (Semrush, cited in SERPs.io). One analysis (Goodie, May 2026) even documents a specific before/after: measuring the same 41-brand referral panel eight months apart, ChatGPT's share of AI referral traffic fell from 89.1% to 63%, while Claude grew from 1.4% to 18.5% and Gemini quadrupled — concluding that "the Big 1 has become the Big 4."

These aren't necessarily contradictions so much as different rulers. Usage share (how many people open the app), referral-traffic share (how many clicks a platform sends to outside websites), and revenue share all tell different stories, and most public "market share" charts don't specify which one they mean. The one thing every source agrees on directionally: the market has fragmented from a near ChatGPT monopoly in 2024–2025 into a genuine multi-player field by mid-2026, with Claude showing the fastest quarterly growth of the major players in several of these datasets.

The number that should recalibrate the whole conversation

Here's the figure most "AI is eating search" narratives leave out. Cloudflare Radar's own referral-traffic data for May 2026 found Google still sent 87.63% of all search referral traffic it observed, with Bing at 3.45%, TikTok at 3.26%, and DuckDuckGo at 1.38% — while every AI chatbot combined (ChatGPT, Gemini, Claude, and Perplexity together) sent just 0.29% (TechnologyChecker.io, 2026). The same analysis notes a striking asymmetry in the other direction: Anthropic's ClaudeBot crawled 11,122 pages for every one human visit it referred back to a site, OpenAI's ratio was 857 crawled pages per referred visit, and Google's was 5:1. In plain terms — AI companies are reading the open web at enormous scale while sending back only a small fraction of the traffic in return.

That single data point doesn't cancel out everything else in this article. It does mean two things can be true at once: AI usage is genuinely exploding, and AI's current direct contribution to a typical website's traffic remains, as of mid-2026, still small relative to Google. What's shifting fastest isn't total traffic — it's where the decision gets made, even when the click still doesn't happen.

What's happening inside Google itself

Google's AI Overviews are the biggest single driver of behavior change, and here the data is both more consistent and more alarming for publishers. AI Overviews appeared on roughly 25–30% of U.S. queries in Q1 2026 by one estimate (Presenc AI) and as high as 48% by another (BrightEdge, February 2026, cited in Omnibound), up from about 8% in early 2024 — a difference likely reflecting query-type sampling rather than a real disagreement about direction. Sector-specific triggering is notably uneven: B2B technology queries trigger AI Overviews 82% of the time, healthcare queries 88% (Omnibound, 2026).

The most methodologically rigorous study available is Pew Research Center's July 2025 report, which tracked 68,879 real Google searches from 900 U.S. adults in real time (not estimates or keyword-tool modeling). It found users clicked through to a traditional result just 8% of the time when an AI Overview appeared, versus 15% without one — a 46.7% relative decline — and that only 1% of users click a link inside the AI summary itself. Google publicly disputed Pew's methodology, saying the study period overlapped with unrelated algorithm testing (PPC Land, 2026). Separately, Seer Interactive's longitudinal analysis of 25.1 million organic impressions across 42 client organizations found organic CTR fell 61% on queries where AI Overviews appear (Omnibound, 2026).

Zero-click search — queries that end without any click to an external site — has climbed from roughly 45% a decade ago to a range reported between 58.5% and 69% in 2026, depending on the panel (SparkToro/Datos vs. Similarweb, both cited across multiple 2026 reports). SparkToro's Rand Fishkin, whose clickstream panel has tracked this metric for years, found the share of searches ending in another Google search (rather than a click anywhere) rose 7.2 percentage points between 2024 and 2026 — evidence, per that analysis, that Google is increasingly keeping users answering questions inside its own ecosystem rather than sending them anywhere at all.

There is a genuine silver lining buried in this data, though: brands that do get cited inside an AI Overview see 35% more organic clicks and 91% more paid clicks than brands sitting in the position just below the AI answer on the same results page (Seer Interactive, cited in Bloggersideas, 2026). Being cited, not just ranking well, is now the variable that matters.

Where the citations actually come from

Across every independent measurement approach — Muck Rack's Generative Pulse (25+ million links analyzed across ChatGPT, Claude, and Gemini, three editions since July 2025), Fullintel/UConn's academic study presented at IPRRC in February 2026, and Golin's independently presented figures — earned media consistently accounts for the large majority of what AI systems cite: reported between 82% and 89% across studies, holding at 84% in Muck Rack's May 2026 edition specifically. Paid and advertorial content, by contrast, represents a negligible 0.3% of citations (Muck Rack, May 2026). Journalism alone makes up 25–27% of all cited links across editions of the same study. This pattern has reportedly held steady since measurement began in mid-2025, suggesting it reflects how these models were trained to weight source credibility, not a temporary quirk of one model version.

That has a specific, practical implication documented in the same research: the journalists PR teams most frequently pitch and the journalists AI engines most frequently cite overlap by just 2% (Muck Rack Generative Pulse, cited in AuthorityTech.io, March 2026) — meaning most current PR targeting effort and most actual AI-citation influence are aimed at almost entirely different people.

Separately, Ahrefs' study of 75,000 brands found that a brand's raw web mentions correlate roughly three times more strongly with AI visibility than its backlink count does (a 0.664 vs. 0.218 correlation coefficient) — a meaningful data point for anyone assuming classic link-building SEO tactics transfer directly to AI visibility (Ahrefs, 2025, cited in AuthorityTech.io, 2026).

Citation sourcing is also concentrated: BrightEdge and Ahrefs data show that roughly 40–55% of ChatGPT Search and Perplexity citations flow to fewer than 1,000 domains total (DigitalApplied, 2026) — meaning a small set of highly-trusted sites (Reddit, Wikipedia, Stack Overflow, and major news outlets are named specifically) are doing an outsized share of the work of informing what AI systems say across nearly every topic.

Who's actually using this, and for what

Adoption isn't even across age groups. One 2026 composite analysis found ChatGPT adoption at 13–17% among the 13–44 age bracket, versus just 5–8% for people 45 and older (Vertu, cited in Stackmatix, 2026) — a meaningful generational gap in who's currently forming habits around AI-first search behavior. Query type also matters: informational queries show a roughly 74% zero-click rate versus just 31% for transactional queries (Semrush Intent Study, cited in Omnibound, 2026), meaning research-heavy categories are being reshaped far faster than purchase-ready ones.

The honest takeaway

Every individual number in this space comes from a real study, a real panel, or a real analytics platform — but almost none of them agree on the exact magnitude, because they're measuring genuinely different things with genuinely different methods. What's consistent across all of them, regardless of source or methodology, is the direction: AI-mediated queries are a real and growing share of how people research, ask, and decide; being cited now matters more than being ranked; earned, independent coverage is what these systems trust; and despite the scale of usage, direct referral traffic to the open web from AI platforms remains, for now, a small fraction of what Google alone still sends. Anyone building a strategy on a single headline statistic from this space is building on a number that a different, equally credible study would put somewhere else entirely.

References

Presenc AI, "AI Search Engine Market Share 2026," May 2026 (Similarweb data)
ReporterOutreach, "AI Search Statistics 2026," June 2026 (StatCounter data)
Stackmatix, "AI Search Market Share 2026" and "AI-Powered Search Engine Statistics," March–April 2026 (FirstPageSage, Graphite, Similarweb, Vertu data)
Goodie, "2026 AI Search Traffic Report: ChatGPT Is Slipping," May 2026
SERPs.io, "AI Search in 2026: Every Stat You Need to Know," May 2026 (DemandSage, Semrush, SEOProfy, DataReportal, Sacra data)
TechnologyChecker.io, "Search Engine Market Share 2026," July 2026 (Cloudflare Radar data)
DigitalApplied, "AI Search Engine Statistics 2026," April 2026 (BrightEdge, Ahrefs data)
Bloggersideas, "80+ Google AI Overview Statistics 2026," June 2026 (Pew Research, Seer Interactive data)
Similarweb, "Zero-Click Marketing: What the 2026 Data Means," June 2026 (SparkToro/Rand Fishkin data)
Search Engine Land, "Google zero-click searches reach 68% in early 2026," June 2026 (SparkToro data)
PPC Land, "Researchers find Google AI Overviews cut publisher clicks 39.8%," 2026 (Pew Research, Seer Interactive data)
Omnibound, "Zero-Click Search Statistics (2026)" and "AI SEO Statistics (2026)" (SparkToro, Bain & Company, Pew, Gartner, Ahrefs, Seer Interactive, BrightEdge, Semrush data)
Muck Rack / Generative Pulse, "What Is AI Reading?," editions July 2025, December 2025, and May 2026
AuthorityTech.io, "82% of Journalists Now Research Stories in AI," April 2026 (Muck Rack, Ahrefs, Fullintel/UConn data)
Machine Relations, "Evidence Base — Independent Validation," June 2026