This is a living, fully-sourced reference of AI search statistics for 2026. Every figure below is tied to its named source and date, hedged where the number is volatile; anything we couldn't attribute to a primary source has been left out. AI-search data is early and fast-moving, so treat these as directional reference points — and check the original before you quote them.
Last reviewed: 15 June 2026. Figures are vendor- or platform-reported unless noted; citation shares in particular swing month to month.
How big is AI search in 2026?
Adoption is climbing fast across the major engines. The platform-reported figures, hedged:
| Statistic | Source | As of |
|---|---|---|
| Google AI Mode passed 1 billion monthly active users | Google (I/O 2026) | May 2026 |
| Google AI Overviews reach well over a billion users monthly | Google (reported) | 2026 |
| OpenAI reported ~800 million weekly active users of ChatGPT | OpenAI (reported) | late 2025 |
| Traditional search engine volume forecast to fall ~25% by 2026 as users shift to AI | Gartner (forecast) | 2024 |
The direction is unambiguous even where the exact numbers move: the major engines now serve AI answers at the scale of the open web. (For how those answers are assembled, see how Google AI Mode works and the AI Overviews guide.)
How much discovery is now zero-click?
A growing share of searches end inside the answer, with no click to a website — the zero-click reality that reshapes how brands get found.
- ~68% of Google searches ended without a click in early 2026, per a Datos/SparkToro analysis reported by Search Engine Land. The rise of AI summaries is a major driver.
- Engines increasingly answer the question on the results page, so visibility shifts from earning the click to being cited in the answer.
The practical read: if your only metric is sessions, AI search will look like a loss even as your brand is being shown. Measure citations and recommendations alongside traffic.
Which sources do AI engines cite most?
Community-edited and experience-rich sources dominate AI citations — often over established news and brand pages. Numbers vary by engine and category, so these are directional:
| Statistic | Source | As of |
|---|---|---|
| Wikipedia ~13.15% + Reddit ~11.97% of US ChatGPT citations (together >25%); WSJ, NYT, Bloomberg absent from top 20 | 5W Research | June 2026 |
| Reddit leads AI citation share; LinkedIn cited in ~14.3% of ChatGPT Search, ~13.5% of Google AI Mode, ~5.3% of Perplexity responses | Semrush (325K-prompt study) | 2026 |
| "Best X" blog lists were 43.8% of all cited page types in a ChatGPT analysis of ~750 prompts / ~26,283 URLs | Glen Allsopp (practitioner study) | 2026 |
| Claude–ChatGPT citation domain overlap just 8%; Claude–Google overlap ~64% | Profound, "The State of AI Search" (Zero Click conf) | June 2026 |
Two takeaways: citation sources skew toward corroborated community content (a reason to take Reddit visibility seriously), and the engines barely agree with each other — so being cited on one is no guarantee on another. The dominance of independent roundups also confirms the listicle effect and the case for earned placement.
How fast is AI referral traffic growing?
When citations are clickable, AI engines are starting to send real referral traffic — though it's still a small share of the total.
- Adobe Analytics reported a roughly 693% year-over-year surge in AI-source traffic to US retail sites over the 2025 holiday season — a large jump off a small base.
- Profound reported ChatGPT referral traffic jumped ~60% in May 2026 when its citations became clickable, with about a quarter of those clicks landing on homepages (up from ~3.6%).
These are directional and vendor-reported, but the pattern is consistent: AI referral volume is growing quickly off a small base, and it behaves differently from search clicks — which is why it often shows up as direct traffic in analytics.
The numbers move every month, but the direction doesn't: discovery is shifting into AI answers, the sources engines cite skew toward community and independent content, and the engines disagree with each other. Optimise for citations, corroborate across engines, and measure the trend — not the headline.
How should you use these statistics?
Treat this page as a reference, not gospel. Cite the named primary source when you reuse a figure, note its date, and prefer a measured trend over a single snapshot. Because live retrieval favours recently-updated pages, we keep this reference on a refresh cadence and update figures substantively as new studies land — the same freshness discipline that keeps any competitive page citable.
The one stat that matters most is your own: whether AI engines actually surface, cite, and recommend your brand, tracked over time across every engine. That measurement is exactly what Buffy Intel is built to provide.
