Google finally put numbers on AI search and gave site owners data to measure it. Here is what changed in 2026, with dates — and what it means for your visibility.
The numbers Google confirmed at I/O 2026
For two years, AI search adoption was argued over with third-party estimates. At I/O 2026 in May, Google published its own figures:
- AI Overviews: more than 2.5 billion monthly users.
- AI Mode: more than 1 billion monthly users.
- AI Mode query volume: more than doubling every quarter since launch.
Google also made Gemini 3.5 Flash the default model in AI Mode globally, and merged AI Overviews and AI Mode into a single continuous AI Search experience — a user can move from a question, to a results page with an AI Overview, to a follow-up in AI Mode without breaking stride. That experience went live worldwide across desktop and mobile.
The strategic point for site owners is not the size of the numbers. It is that being cited inside the answer has become a visibility channel in its own right, running parallel to the ranked link — and until June, one you had no native way to measure.
June 3: Search Console starts reporting AI visibility
Google introduced Search Generative AI performance reports in Search Console on June 3, 2026. They cover AI Overviews, AI Mode and AI features in Discover, broken down by page, country, device and date.
Two limitations decide how you can actually use this data, and both are easy to miss:
- Impressions only. There are no clicks, no CTR and no query data in this version. You can see that a page surfaced in an AI answer; you cannot see what was asked, or whether anyone clicked through.
- No historical backfill. Data starts on May 18, 2026. There is no way to reconstruct what your AI visibility looked like before that date.
It rolled out first to a subset of UK sites before widening. Alongside it, Google shipped a control letting owners keep their content out of generative AI features — and stated plainly that this control is not used as a ranking signal for search results outside those generative AI features. Opting out costs you AI traffic; on Google's own account, it does not cost you organic rankings.
What "optimising for AI answers" actually means in 2026
Google's position is consistent and worth taking at face value: optimising for generative AI features is still SEO. Its guidance for AI features points at the same fundamentals — useful content, a clear technical structure a crawler can parse, accurate details — and states there are no additional technical requirements and no special AI-specific files needed to appear in those features.
That last point is worth being blunt about, because a large amount of 2026 commentary claims otherwise. There is no Google-sanctioned llms.txt requirement. We cover the actual state of AI crawler control in our technical SEO news deep-dive.
What does hold up in practice:
- Answer the question in the first two or three sentences under the heading, then expand. Extractable, self-contained passages are what a synthesised answer can lift and attribute.
- Publish things that only you can publish. Original data, first-hand testing, real case results. A model summarising five sources that all say the same thing has no reason to cite any one of them; a source with a number nobody else has is citable by necessity.
- Make entities unambiguous. Consistent organisation and author identity, clean internal linking, and supported structured data — note that FAQ markup no longer earns anything, as covered in our technical roundup.
- Keep pages crawlable and fast. The AI surfaces are fed by the same index as everything else.
The honest state of measurement
Being straight about what cannot currently be measured is more useful than pretending otherwise:
- Google does not separately attribute AI Mode or AI Overviews referrals in analytics — those clicks arrive bundled into
google / organicalongside ordinary search traffic. - Search Console's new reports give impressions but no clicks, so AI-surface CTR is not calculable from Google's own data.
- Third-party AI visibility trackers sample prompts rather than observing your actual users, so they estimate share of voice — a genuinely useful directional signal, but not a traffic measurement. See our SEO tools news roundup for where that tooling stands.
Anyone quoting you a precise "AI traffic" figure for your site today is modelling, not measuring. That will change as Google extends the reports; it has not changed yet.
← Back to the SEO News hub for the running algorithm tracker and every other deep-dive.
Sources
- Google Search's I/O 2026 updates: AI agents and more — Google Blog
- Introducing Search Generative AI performance reports in Search Console — Google Search Central
- New opportunities, control and insights for website owners — Google Blog
- Google Search's AI features and your website — Google Search Central
Mubashar Shahzad
Dedicated to tracking and decoding the latest Google algorithm updates and SEO trends.
