On 3 June 2026 Google added a dedicated generative AI performance report to Search Console. It is the first time site owners have been given a direct view of how often their pages appear inside AI Overviews and AI Mode, separated from ordinary organic data.
It is a real and useful addition. It is also much narrower than the reaction to it suggested, and the narrowness is the part worth understanding before anyone builds a report around it.
What the report actually contains
One metric: impressions. Google's documentation defines an impression here as a link to your site being shown to a user in a generative AI feature on Google Search. The report covers AI Overviews and AI Mode. Generative AI features in Discover sit in a separate report.
Four dimensions: pages, countries, devices and dates. Pages group by the final URL after redirects, assigned to the canonical, exactly as the ordinary performance report does. Dates run at hourly, daily, weekly and monthly granularity.
The usual Search Console limits apply, including the 1,000 row cap and the preliminary-data window on the most recent days. Data from Search Labs experiments is excluded.
What is missing, and why it matters
There are no clicks. There is no click-through rate, no average position, and no query dimension. You can see that a page was shown inside an AI answer. You cannot see whether anyone came, or what they had asked.
That gap is the whole measurement problem in one place. An impression in an AI answer is not a visit, and for a lot of informational content it is specifically the thing that replaces a visit. Counting impressions as a success metric would reward exactly the pattern that costs you traffic.
Google has said it expects to add metrics over time. Until it does, the honest description of this report is that it confirms presence and says nothing about value.
How to use it without misreading it
Use it as a coverage check. Which pages are being pulled into AI answers at all, and are they the ones you would have chosen? A commercial page that never appears is a relevance problem worth investigating. A thin informational page that appears constantly is a page whose traffic is likely to keep falling.
Use it to explain a drop, not to declare a win. If organic clicks to a page fell while its AI impressions rose, you have a plausible mechanism rather than a mystery, and you can decide whether that page still earns its keep.
Do not build a target on it. A number with no denominator and no outcome attached is not something to manage against. The scoreboard for whether search is working stays where it was, on qualified leads and revenue.
If the report is not there, that is usually rollout rather than failure. Google is releasing it to a subset of properties. It can also be empty because the site has too few impressions in AI features, or because the site has been excluded from them in the Search generative AI control.
What this means for an operator
Pull the report to see which pages are being surfaced in AI answers, then go back to qualified leads to decide whether it mattered. Do not set a target on AI impressions.