20 AI search statistics that explain 2026 (with sources)

Statistics about AI search age fast and get repeated without sources until nobody can find the original. So here's a curated set we'd actually stake a claim on - each number links to its source, vendor research is labeled as such, and where studies disagree (they often do), we say so instead of cherry-picking the scariest figure. These are third-party findings, not Zene's own data; we just collected and sourced them. Last updated October 7, 2026: we replaced figures we could no longer trace to a primary source and added the August 2026 changes in how ChatGPT picks and cites brands.

The shift away from classic search

• 1. Traditional search volume was projected to drop 25% by 2026 as users move queries to AI chatbots and other virtual agents - Gartner (a February 2024 prediction, not a measurement).
• 2. When an AI summary appears, users click a result just 8% of the time - versus 15% when there's no summary, and only 1% click a link inside the summary itself. From the real browsing of 900 US adults, studied by the Pew Research Center (July 2025).
• 3. AI Overviews went from 6.49% of Google queries in January 2025 to 13.14% in March, then settled at around 16% later in the year, across more than 10 million keywords - Semrush's 2025 study (vendor). Other trackers report different shares depending on query mix and device, so treat any single figure as method-dependent.
• 4. AI Overviews have over 2.5 billion monthly users, and AI Mode passed 1 billion within a year of launch - Google, at I/O in May 2026. Google's own search results page is now the biggest AI answer surface there is.

How big the engines actually are

• 5. ChatGPT reaches 1.2 billion people a week - OpenAI, DevDay, September 29, 2026. That's up from 900 million weekly active users in February 2026 (OpenAI, via TechCrunch).
• 6. ChatGPT had 50 million paying subscribers in February 2026 - the same OpenAI disclosure. These are people who've made the assistant a paid habit, not casual triers.

Buyers are deciding inside the answer

This is the cluster that should reframe how you think about the channel. From G2's March 2026 survey of 1,076 B2B software buyers (G2 runs a software review marketplace):

• 7. 51% now start software research with an AI chatbot more often than with Google - up from 29% in April 2025.
• 8. 71% rely on AI chatbots for software research, up from 60% just seven months earlier.
• 9. 69% chose a different vendor than they'd originally planned, based on what the AI chatbot told them.
• 10. One in three (33%) bought from a vendor they weren't familiar with before the assistant surfaced it - the clearest evidence that a recommendation can manufacture a shortlist seat from nothing.
• 11. ChatGPT is the dominant research assistant among these buyers at 63%, though the field is fragmenting fast.
• 12. 53% say AI-chatbot research feels more productive than traditional search - up from 36% seven months prior, so the preference is hardening, not a novelty.

Where the engines get their answers

• 13. Reddit is still ChatGPT's most-cited domain at 16.8% of citations among its top sources, across a broad sample of US queries - Ahrefs Brand Radar (vendor; updated September 2, 2026, refreshed monthly).
• 14. ...or Reddit's share of ChatGPT Search citations fell 86% in August 2026, from a 3.83% average (July 18 - August 7) to 0.52% (August 14-17), in live-interface data from Promptwatch (vendor; it calls the figure provisional). Different methods, overlapping dates - we don't pick a winner; more in why Reddit keeps showing up in AI answers.
• 15. ChatGPT's share of site: searches jumped from 0.37% to 16.8% on August 8, 2026 - it started checking specific websites, mostly brands' own, before answering. Source: Promptwatch (vendor); a separate Peec AI analysis (vendor) found 84% of the targeted domains were branded.
• 16. AI assistants cite fresher pages than Google ranks: the average URL cited by AI assistants was 1,064 days old versus 1,432 days for organic results - 25.7% "fresher", across ~17 million citations (Ahrefs, vendor study, July 2025).

The engines also weight sources very differently from each other, which is the structural reason two engines name different brands for the same question. In our own AI Visibility Index (July 2026 answers), Wikipedia made up 9.5% of ChatGPT's citations and none of Perplexity's.

What gets a brand named

• 17. ChatGPT often decides before it searches. In 21 of 27 conversations, its first search already contained brands the user never typed; a brand it named that way reached the answer 68.9% of the time, versus 2.1% for brands it only found during the search - an independent analysis of one ChatGPT Plus account (July 2026). Directional, by the author's own account.
• 18. YouTube mentions are the strongest correlate of AI visibility (~0.737) across ChatGPT, AI Mode and AI Overviews, in a study of 75,000 brands; branded web mentions follow at 0.66-0.71, while link metrics correlate only weakly - Ahrefs (vendor study, December 2025). Correlation, not causation.
• 19. 97% of llms.txt files got no requests at all in May 2026, across ~38,000 domains that had one - Ahrefs (vendor study). Google, for its part, says its search ignores the file. More in does llms.txt work.

Why a mention is worth winning

• 20. AI-referred visitors punch far above their volume. Ahrefs found that AI search drove 0.5% of its own site's traffic over 30 days but 12.1% of its sign-ups (June 2025) - one company's data, but a striking ratio.

Numbers about the whole market are interesting; the only one that changes your roadmap is your own. A free Zene audit shows which of these engines name your brand today - and which name your competitor instead.