AI search statistics 2026: the numbers behind AEO, ChatGPT, Perplexity, and AI Overviews

Here is the short version. AI search has crossed from novelty to default, and the data now backs that up from every angle: adoption, traffic, citations, conversions, and money. This report pulls together the numbers that actually matter for getting your brand found inside AI answers in 2026, grouped so you can use them. One honest caveat up front, and it is the reason most stat roundups are misleading: the figures vary a lot by source and method, because everyone is measuring a fast-moving target differently. So we have cited every number to its source and shown ranges where studies disagree, rather than cherry-picking the scariest one. If you want the strategy behind the data, start with our complete guide to answer engine optimization. This is the evidence underneath it.

AI search adoption and usage

The engines are now mass-market utilities, not experiments.

• ChatGPT sits near 900 million weekly active users and handles on the order of 2.5 billion prompts a day (OpenAI, early 2026).
• ChatGPT Search alone processes roughly 250 to 500 million queries a week, making it one of the top search properties in the world by volume (Similarweb 2026 AI Search report).
• Google AI Overviews reach about 1.5 billion monthly users (Search Engine Land, January 2026).
• Gemini reached around 450 million monthly active users by mid-2025, with about 50 percent quarter-over-quarter growth in daily requests, helped by its integration across Gmail, Docs, and Workspace (industry reporting).
• Perplexity has roughly 45 million monthly active users and passed 780 million queries a month in 2025, with its CEO targeting a billion queries a week (company statements and Bloomberg).
• By share of the AI chatbot market in early 2026, ChatGPT held roughly 80 percent, with Perplexity and Gemini each in the high single digits (Statcounter).
• Global adoption of generative AI tools reached about 16.3 percent of the world's population by the end of 2025, up from 15.1 percent in the first half of the year (as compiled by Omnibound).
• By 2031, more than 1.1 billion people are expected to use AI tools globally (Statista).
• Roughly 65 percent of AI users are Millennials or Gen Z, the digital-native cohort driving adoption.

The takeaway: whichever single figure you trust, the direction is identical. Hundreds of millions of people now ask AI engines the questions they used to type into Google, and the audience skews young and high-intent.

The behavior shift: zero-click and the leaking click

As AI answers expand, the click that traditional SEO depends on is thinning out.

• AI Overviews appeared in about 25 percent of Google searches by early 2026, up from roughly 13 percent in March 2025, based on Conductor's analysis of 21.9 million queries. Other studies put prevalence higher depending on the query mix, so treat this as a measured floor rather than a ceiling.
• About 30 percent of keywords trigger an AI Overview in U.S. results (SE Ranking).
• When a Google AI summary appears, people click a traditional result in just 8 percent of visits, versus 15 percent when no summary is present, and they click a link inside the summary only 1 percent of the time (Pew Research Center, based on real browsing behavior).
• AI Overviews correlate with a 58 percent lower click-through rate for the top-ranking organic page (Ahrefs, late 2025 data).
• Only about 38 percent of URLs cited in AI Overviews also rank in the top 10 for that query, with roughly a third coming from beyond the top 100 (Ahrefs).
• Gartner has projected that traditional search engine volume will fall about 25 percent by 2026 as people shift to answer engines.
• AI search visits grew 42.8 percent year over year, from 15.6 billion to 27.4 billion between the first quarter of 2025 and 2026, while Google search visits grew just 2.4 percent. The ratio of Google to AI-search users narrowed from 4.9 to 1 down to 3.5 to 1 (as reported by Contently, drawing on Semrush data).
• AI referral traffic now makes up about 1.08 percent of all website traffic and is growing roughly 1 percent month over month, with ChatGPT driving about 87 percent of it (Conductor 2026 benchmarks).
• About 42 percent of people say they prefer an AI chatbot over a search engine for multi-step research, and roughly 29 percent of U.S. adults encounter AI summaries in daily browsing, rising above 40 percent among high-income shoppers (survey data compiled by Sedestral).

The takeaway: ranking number one means less every quarter, because a growing share of answers resolve before anyone clicks. Two facts coexist. AI referral traffic is still a small slice of total traffic today, and it is growing faster than anything else while the clicks from classic results shrink. Both things are true, and planning for only one of them is the mistake.

How AI engines cite sources

Being cited is the new being ranked, and the citation patterns are surprisingly consistent across studies.

• Only about 18 percent of ChatGPT conversations trigger a live web search at all, and commercial-intent questions trigger one far more often than informational ones (Profound, from an analysis of more than 700,000 conversations).
• When ChatGPT does search, it cites around 6 unique sources per conversation, and about two thirds of cited turns draw on just one to four (Profound).
• Citations concentrate on the opening question: the first turn was roughly 2.5 times more likely to be cited than the tenth (Profound).
• The field of cited domains is wide but unequal. The top 10 domains capture only about 12 percent of all citations, with a long tail beyond them (Profound).
• Wikipedia appears in roughly one in six cited ChatGPT conversations, and Reddit in about 13 percent (Profound).
• Between late 2025 and early 2026, LinkedIn climbed from outside the top 20 to the single most-cited domain for professional queries across AI search platforms (industry analysis).
• Earned media makes up the majority of AI citations (Muck Rack dataset).
• Content that AI engines cite tends to run meaningfully fresher, around 25 percent fresher, than the pages Google ranks (industry analysis).

The takeaway: no single site owns AI answers, but a small set of trusted, fresh, third-party sources punch far above their weight. If you want the engine-by-engine detail, we break down ChatGPT in our guide to ranking in ChatGPT and Perplexity in our guide to ranking in Perplexity.

What actually earns citations

The research on which content traits move citation probability is unusually clear, and it is the most actionable data here.

• Adding statistics, quotations from credible sources, and citations boosted a page's visibility in generative answers by more than 40 percent, with gains up to 37 percent on Perplexity, in the foundational GEO study from Princeton, Georgia Tech, the Allen Institute for AI, and IIT Delhi (presented at KDD 2024).
• About 44.2 percent of ChatGPT citations came from the first 30 percent of a page, so front-loading your answer matters (Zyppy analysis of thousands of citations).
• For commercial and evaluation-stage queries, 83 percent of AI citations came from pages updated within the past 12 months, and more than 60 percent from pages refreshed within six months (AirOps 2026 State of AI Search).
• Content updated within the last 30 days earned roughly 3.2 times more citations than stale pages (industry analysis).
• The typical AI summary runs about 67 words, a useful length target for the direct answer you put at the top of a section (Pew Research Center).

The takeaway: the levers are knowable and mostly within your control. Lead with the answer, back it with sourced data, structure it cleanly, and keep it fresh. Our answer engine optimization guide turns these findings into a step-by-step framework.

The business impact and the money

This is the part that turns AEO from a marketing curiosity into a budget line. AI traffic is small but unusually valuable, and it is starting to drive real revenue.

• AI-search visitors are about 4.4 times as valuable as the average traditional organic visitor, because they tend to arrive later in the decision (Semrush).
• ChatGPT referral traffic converts at about 7.1 percent, second only to paid search and ahead of organic, social, and email (Similarweb). Other datasets report even higher conversion for AI-referred visitors, so the exact number varies, but the direction is consistent: this traffic converts well.
• AI-referred visitors spend roughly 68 percent more time on site than visitors from traditional organic search (SE Ranking).
• On May 7, 2026, ChatGPT began embedding clickable links to brand homepages directly inside answers. The share of answers carrying an inline brand link jumped roughly 14 times almost overnight, ChatGPT referral traffic rose about 157.7 percent week over week, and homepage referrals climbed more than 350 percent (independent analyses and Similarweb clickstream data).
• Retail traffic referred by AI grew nearly 393 percent year over year in early 2026, and AI-referred orders on Shopify grew roughly 13 times over the same period.
• The AI search engine market was valued around 18.8 billion dollars in 2025 and is projected to exceed 50 billion by 2033 (market research compiled by Omnibound).
• McKinsey projects that 750 billion dollars in consumer spend will flow through AI-powered search by 2028.

The takeaway: a citation is no longer a vanity metric. With ChatGPT now linking homepages directly, an AI mention can be a high-converting referral, and for product brands it increasingly translates into orders. If you sell online, our view on the ecommerce angle covers what that means in practice.

The measurement gap, which is the opportunity

The most telling statistics in this entire report are about how few brands are actually doing anything. The audience has moved, but the response has not.

• Only 16 percent of brands systematically track their AI search performance (McKinsey).
• About 54 percent of teams say they plan to act on generative engine optimization, but only 23 percent are currently measuring it, a gap that defines the category's maturity in 2026 (as compiled by Omnibound).
• The GEO market itself was valued around 848 million dollars in 2025 and is projected to reach 33.7 billion by 2034, a roughly 50 percent compound annual growth rate (market research compiled by Omnibound).

The takeaway: this is what an early-stage channel breakout looks like, where first movers compound an advantage that becomes hard for late entrants to close. The reason so few brands measure is partly that it is genuinely harder than SEO, since zero-click answers leave no referral session and the most valuable influence happens before any trackable touchpoint. That is exactly the problem a dedicated tracker solves, and it is the heart of our AI search visibility guide.

What all of this means

Step back from the individual numbers and a clear picture forms. AI search is growing many times faster than traditional search, its visitors are worth several times more and convert better, and the buying journey increasingly starts and sometimes ends inside an answer. The mechanics of winning are well documented: get cited by leading with a clear answer, backing it with fresh sourced data, structuring it cleanly, and building a trusted off-site footprint. And almost nobody is measuring any of it yet, which is precisely why acting now pays off.

The data also argues against two common overreactions. This is not the instant death of Google or SEO, since classic search still drives the majority of traffic and AI engines lean on the same indexes and authority signals. And it is not a passing fad, since adoption, citations, and revenue are all climbing on independent measurements. The right posture is parallel: keep earning clicks where clicks still flow, and start earning citations where they do not.

Where outwrite.ai fits

Every statistic in the measurement section points to the same gap, and outwrite.ai is built to close it without an enterprise budget. It tracks your mentions and citations across ChatGPT, Gemini, and Perplexity, scanning the prompts your customers actually ask every day and showing your share of voice against competitors, which models cite you most, the sentiment of each mention, and the full answer behind every one, so you can see exactly where a competitor is cited and you are not. On the execution side, its content creation workflow produces the answer-first, sourced, citable structure the research rewards, and its topic discovery surfaces the questions worth targeting. For the strategy and tactics behind these numbers, the AEO section of our blog goes deeper. outwrite.ai is the practical choice for founders and lean teams who want to measure their AI visibility and act on it in one place.

The bottom line

The data on AI search in 2026 tells one story from many directions. The audience has moved, the click is thinning, citations are the new ranking, AI traffic converts better than almost anything else, and only a small fraction of brands are tracking any of it. Numbers will keep shifting as the space matures and as studies refine their methods, which is exactly why you should anchor on the trend rather than any single figure. The brands that start measuring and optimizing for AI answers now will build a lead that compounds.

If you want to see what AI engines say about your brand today, and then move those numbers in your favor, that is exactly what outwrite.ai was built to do.