Google AI Mode hits 75M daily users — citation decouples from ranking
Nick Fox, Google’s VP of Search, publicly confirmed in April 2026: AI Mode has reached 75 million daily active users — a 4× increase since the US launch in May 2025. AI Mode is now available in 53 languages and 40+ markets; in the US it has been rolled out to all users since March 2026. In parallel, an Ahrefs study of over 863,000 keywords surfaced a striking detail: only 38% of URLs cited in AI Overviews still rank in the top 10 for the same query — seven months earlier the figure was 76%.
What concretely changed
- AI Mode at 75M daily users, +4× since launch (May 2025); scaled to 53 languages and 40+ markets.
- 93% of AI Mode queries end without an outbound click (Seer Interactive analysis, 25.1M impressions).
- AI Overviews now cite 13.34 sources per answer — nearly double 2024 (6.82).
- Citation-ranking decoupling: Ahrefs measures only 38% overlap between AIO citations and top-10 ranking — vs. 76% in the prior year.
- Ads in 25.5% of all AI results — a 394% increase year-over-year.
What used to be
Until summer 2025, citation in AI Overviews was tightly coupled to classic ranking. Ranking in the top 10 gave you good odds of being cited — and vice versa: getting cited usually meant ranking conventionally too. The Ahrefs sample for that period showed 76% overlap. For SEOs the practical consequence: classic measures (technical hygiene, backlinks, on-page) paid off twice — on ranking and on citation.
AI Mode itself — Google’s full AI search interface, similar to Perplexity — was a US experiment with modest reach. The public beta launched in the US in May 2025; further markets followed during the year, but user numbers stayed well below those of classic search.
What applies now
1. AI Mode has hit the mass market. 75M daily users is the threshold beyond which a search product can’t be ignored. For comparison: Perplexity reported 22M monthly actives in the same period. The speed matters: 4× growth in eleven months, driven by the US general rollout in March 2026 and the language expansion. For German sites this means: AI search’s share of total query volume is growing visibly — even if Germany is still trailing in the AI Mode rollout.
2. Citations and rankings are drifting apart. The Ahrefs number is the central shift. If only 38% of cited URLs still rank in the classic top 10, Google is increasingly selecting for AI answers by criteria that differ from the classic SERP. Which exact criteria isn’t disclosed in detail, but the signals point at: topical specificity, primary-source character, clear factual density, an authoritative domain without tabloid mixing.
Practical consequence: optimizing for classic ranking while disappearing from AI Overviews means optimizing for the shrinking field. Conversely there are sites that get cited regularly in AIO without ranking classically at the top — a sign of sites that work journalistically densely and factually cleanly.
3. More sources per answer, lower per-source click probability. With 13.34 sources per AI-Overview answer on average (vs. 6.82 in 2024), Google distributes attention across twice as many sites. That lowers the chance that any single cited source captures the click — at 93% zero-click queries it’s marginal anyway. Citation becomes a brand-visibility playing field, no longer a traffic source.
4. Ads are entering the AI answer. 25.5% of all AI results now show ads — vs. 5.1% a year ago (+394%). That makes AI Mode relevant for SEA. The ad logic differs from classic search ads: ads appear contextually inside the answer, marked as “Sponsored,” but visually harder to separate from the AI response.
Takeaway
The Ahrefs study is methodologically sound (863,000 keywords) and shows, with the 76→38% shift in just seven months, an unusually fast structural change. Anyone ignoring it keeps optimizing for a world that already doesn’t exist for more than half of the relevant queries.
The 75M mark suggests in turn that Google is building AI Mode not as a beta but as a strategic default product. The question for German marketing leads is no longer “when is this coming?” but “how much head start can we use before it arrives here?” The head start is finite — language expansion to 53 languages is a clear signal.
What you can do now
If you do classic SEO: alongside ranking position, track citation performance in AI Overviews. Tools like Profound, Otterly, ChatRank, or a manual sample show whether your site gets cited, where it doesn’t, and which competitors hold the citation slot.
If you own content strategy: read the Ahrefs study and ask yourself: which content on my site would be citable if I read it from an AI-Mode perspective? Factual density, primary sources, clean definitions, unambiguous data points — those are the factors that survive in the 38% world.
If you handle SEA: check Google Ads reports for whether impressions from AI Mode contexts are reported separately — the reporting logic is currently shifting. 25.5% ad share inside AI means competitive pressure there is rising fast; early tests pay off.
If you manage client expectations: explain the 76→38% shift. It’s the clearest single number to ground the structural move from ranking to citation.
Entdecke mehr
Three Days with Claude Fable 5 — and Why Opus Stays My Daily Driver
Fable 5 fixed a bug in one shot that Opus failed at twice. I'm still not switching. My honest impressions from the trenches after three days.
GlossarGEO — Generative Engine Optimization
GEO is the discipline of shaping content for visibility in AI answer engines — AI Overviews, Perplexity, ChatGPT Search, Claude. The goal is not the classic SERP click, but appearing as a cited source inside the generated answer.
LexikonGEO — Generative Engine Optimization Explained
What GEO is, how generative engines cite sources, and which factors raise citation likelihood. Honestly framed, without the hype.