AI Search Statistics 2026

The search box is being replaced by the answer box. AI Overviews now sit on roughly half of US Google queries, most informational searches end without a click, and a new discipline — generative engine optimization — is the fastest-growing subject in marketing. This index tracks the numbers behind that shift: how often AI answers appear, what they do to clicks, who is winning the citation market, and what publishers are doing about it. Every figure is attributed, dated, and rated for confidence.

Thank you for reading this post, don't forget to subscribe!

The state of play: search after AI

For twenty-five years, the deal between Google and the open web was simple: publishers made content, Google sent clicks. In 2026 that deal is functionally over for informational queries. When Google shows an AI Overview — which it now does on roughly half of US searches — users click a traditional result barely half as often as they otherwise would. Pew Research Center’s analysis of nearly 69,000 real user queries found a click rate of just 8% when an AI summary appeared, versus 15% without one, and only 1% of AI Overviews produced a click on the source the AI actually cited.

The traffic is not coming back — it is moving. ChatGPT now processes over a billion queries a day, and referral traffic from AI assistants to websites grew roughly 527% year over year. That referral stream is still small next to Google’s, but it behaves differently: visitors arrive pre-qualified by a recommendation, and multiple analyses find they convert several times better than organic search visitors. The scarce asset is no longer the ranking; it is the citation.

That is why generative engine optimization (GEO) has gone from jargon to job title in eighteen months. Ranking #1 on Google no longer guarantees an AI engine will cite you — the overlap between top Google rankings and AI citations has collapsed from about 75% to somewhere between 17% and 38% depending on the engine. Winning both games now requires two distinct strategies, run in parallel.

• The click market is shrinking. AI summaries roughly halve click-through, and Google’s AI Mode resolves ~93% of sessions with zero clicks (Semrush).
• The citation market is growing. AI referrals are up 527% YoY and convert up to 9× better; brands cited in AI Overviews see a 35% organic CTR lift (Semrush).
• SEO and GEO have diverged. Google-rank-to-AI-citation overlap fell from ~75% to 17–38% — visibility in AI answers must be earned separately.

The zero-click era, measured

Independent studies using different methodologies now converge on the same picture: when an AI answer appears, clicks fall by half or more. The table below collects every major published measurement.

Note on the range: “zero-click” estimates vary from ~60% (all queries) to 93% (AI Mode only) because they measure different surfaces. The consistent signal across all of them is direction and magnitude: AI answers absorb the majority of informational clicks.

What it is doing to publishers

The aggregate CTR numbers translate into real losses for content businesses. The UK’s Daily Mail reported CTR declines of up to 89% on queries where AI Overviews appear. Business Insider lost roughly 55% of its organic traffic between 2022 and 2025 and cut 21% of staff in 2025, citing the search shift. Across the industry, analysts put the decline in traditional search-driven traffic at 15–25% and falling.

The pattern of who loses is uneven. Commodity content — definitions, how-tos, listicles, anything an AI can synthesize from many sources — loses hardest. What retains clicks: original data, tools, strong brands users search for by name, and transactional queries where the user must ultimately visit a site to act. This is the economic logic behind our own format — primary-sourced statistics are among the content types AI engines must cite rather than absorb.

AI assistants are the new search engines

The queries leaving Google are landing in assistants. ChatGPT reached roughly 900 million weekly users and processes more than 1 billion queries per day — against Google’s estimated ~14 billion daily searches, that is already a high-single-digit share of the world’s question-answering. Perplexity, the largest dedicated answer engine, crossed 170 million monthly visits. Our Generative AI Statistics index tracks the full user-growth series; our How People Use AI index breaks down what those sessions are for — information-seeking and guidance make up the largest share.

The commercial significance is in the quality, not just the quantity, of this traffic. AI-search referral traffic to websites grew about 527% year over year, and visitors referred by ChatGPT convert up to 9× higher than Google organic visitors: by the time an AI recommends a destination, the user has already had their comparison questions answered. For publishers and brands, one AI referral is worth many search clicks — which changes what “winning” search means.

GEO: the new discipline, by the numbers

Generative engine optimization is the practice of earning citations in AI answers. Its rise is measurable in search demand itself — Ahrefs’ keyword data shows the fastest growth curve of any marketing topic on record:

What actually earns AI citations

Citation studies and engine documentation converge on a consistent set of factors. AI engines preferentially cite pages that are:

• Primary or clearly-attributed sources — original data, named studies, dated figures. Synthesized commodity content gets absorbed, not cited.
• Structured for extraction — question-shaped headings, direct first-sentence answers, tables, and schema markup (Article, Dataset, FAQPage) that machines parse without guessing.
• Fresh and visibly dated — engines favor sources with explicit “updated” dates when answering time-sensitive questions.
• Open to AI crawlers — GPTBot, ClaudeBot, PerplexityBot and Google-Extended must be allowed in robots.txt; an llms.txt file signals which pages matter most.
• Corroborated across the web — being referenced by other sites remains a trust input for AI engines, much as links were for PageRank.

None of this replaces traditional SEO — Google organic still delivers the larger absolute volume. The 2026 playbook is to run both: rank for the shrinking click market, and be citable for the growing citation market.

How we got here: a short timeline

Definitions

AI Overview — the AI-generated summary Google places above traditional results, synthesized from multiple sources with citation links.

Zero-click search — a query resolved entirely on the results page, with no visit to any website.

GEO (generative engine optimization) — the practice of making content easy for AI engines to find, trust, and cite; the citation-market counterpart to SEO’s click market.

Answer engine — a system whose product is the answer itself rather than a list of links: ChatGPT, Perplexity, Google AI Mode. See also our glossary entries on LLMs and RAG, the technology behind most answer engines.

FAQ

What share of Google searches show an AI Overview in 2026?

Roughly 48–50% of US queries, per 2026 rank-tracker data — up from about 15% at launch in May 2024. Long, question-style queries trigger them most often.

How much traffic do AI Overviews take from websites?

The best behavioral evidence (Pew, 68,879 queries) shows clicks fall from 15% to 8% when a summary appears; Seer Interactive measured organic CTR down 61% across 25M impressions. Individual publishers report losses from 30% to as high as 89% on affected queries.

Is traditional SEO dead?

No — but it is no longer sufficient. Google-ranking-to-AI-citation overlap has fallen to 17–38%, so AI visibility must be earned separately through GEO, while organic still carries the larger absolute volume.

Is AI referral traffic actually worth anything?

Yes — disproportionately. It grew ~527% year over year and converts up to 9× better than Google organic, because the AI has pre-qualified the visitor before sending them.

What is the single highest-leverage GEO action?

Publish original, dated, clearly-sourced information that AI engines must cite rather than absorb — then make sure AI crawlers can read it (robots.txt, llms.txt, schema markup).

Figures are drawn from Pew Research Center (behavioral study, Jul 2025), Seer Interactive (CTR analysis, Sep 2025), Semrush (AI Overviews and AI Mode studies, 2025), Ahrefs (keyword-demand data, 2026), Gartner (search-volume forecast, 2024), and Think with Google (2026), with industry aggregation from SEO trade research. HIGH confidence marks primary research with published methodology; MEDIUM marks vendor or aggregated estimates not yet independently replicated. Corrections: see our methodology and corrections policy.

How People Use AI · Generative AI Statistics · AI-Generated Content · AI Slop: The Synthetic Flood