On June 3, 2026, Google added a Generative AI performance report to Search Console. The report shows how many impressions a site’s URLs received in AI Overviews and AI Mode, broken down by page, country, device, and date. It’s currently live for only a limited number of accounts, and click and CTR data aren’t in the report yet.
What Does the Report Measure?
The report’s single metric is impressions: an impression is counted whenever a link to your site is shown to a user in an AI Overviews or AI Mode result. It’s recorded whether or not it’s clicked.
You can break the data down across four dimensions:
| Dimension | Description |
|---|---|
| Pages | Impressions per URL (based on the canonical URL) |
| Countries | The country the search was made from |
| Devices | Desktop, tablet, mobile |
| Dates | By day, week, or month (Pacific time) |
What Doesn’t the Report Measure?
The official documentation is explicit about these limits:
- No clicks or CTR. Google says it plans to add further metrics down the line, but hasn’t officially said whether those will include clicks or CTR.
- Search Labs experiments aren’t included. Features still in active development aren’t covered.
- Discover has its own report. The Search Generative AI report only covers search results; Discover’s generative AI features are shown in a separate report.
- No query data. There’s no way to see which search query triggered an impression.
The Data Was Already There, Buried in the Web Total
An important technical note from the SEO community: AI Overviews and AI Mode impressions were already included in the Web totals in the existing Search Console performance report. The new report isn’t generating this data from scratch — it’s unbundling it and making it visible.
That distinction matters. AI visibility isn’t starting from zero; it’s been mixed in with organic traffic data all along. The new report’s real value isn’t a numeric jump — it’s the ability to break the data out by surface.
The Difference Between the Chart and the Table
The report has two sections, and their totals can look inconsistent:
Chart: Aggregates data at the property (site) level. If two URLs from the same site appear in the same search result, the chart counts that as a single impression.
Table: Groups by whichever dimension you select. With the page dimension selected, it counts per URL. That’s why the chart total and the table total can differ — it’s not a bug, it’s a difference in aggregation method.
The standard Search Console performance report’s 1,000-row limit and other data restrictions apply to this report too.
How the SEO Community Reacted
The reaction has been polarized:
Positive: Search Engine Journal called it “one of the most requested features in Search Console history.” In his live preview, Brodie Clark wrote that this is the most concrete site-level data Google has given for AI surfaces so far.
Critical: The entire conversation has centered on “are AI answers reducing clicks?” — and that’s exactly the piece that’s missing. Also, these impressions were already counted in the Web total; the report isn’t generating new data, it’s just unbundling existing data.
Both takes have merit. The lack of click data is a real gap, but until now it was impossible to even separate AI Overviews behavior from classic organic traffic behavior — that alone was a blocker.
How to Access It
In Search Console, go to Performance → Generative AI in the left-hand menu, or open the URL directly:
search.google.com/search-console/performance/search-analytics/ai
If the Report Isn’t Showing Up
The official documentation lists three possible reasons:
- Your account may not yet be included in the phased rollout.
- Your site may not have hit the impression threshold required for AI features.
- Your site may have been excluded from Search generative AI features. You can check this in Search Console’s content settings.
Where’s the Discover Report?
The Search Generative AI report only covers AI Overviews and AI Mode visibility in search results. Impressions from Google Discover’s generative AI features are shown in a separate report. It’s important not to conflate the two — they involve different user intent and different surface dynamics.
Reading It Alongside the GA4 AI Assistant Channel
The Search Console report measures impressions, while GA4’s AI Assistant channel tracks the clicks and sessions that actually reach your site. The two sources complement each other:
| Question | Source |
|---|---|
| How many times did we appear in an AI feature? | Search Console Gen-AI report |
| How many of those impressions turned into a click? | GA4 AI Assistant channel |
| What did the resulting visitors do — did they convert? | GA4 (engagement + conversion reports) |
Once click data lands in Search Console, this gap between the two tools will close. Until then, tracking both separately is the safest approach.
What Does an Impression Without a Click Mean?
Getting an impression means your content was shown to a user on an AI surface — the user may have gotten their answer right there and never come to your site at all. Sometimes that’s a failure, and sometimes it’s clickless brand visibility. Making content decisions without distinguishing between the two is risky.
In practice, there are three scenarios:
| Scenario | Possible interpretation | Action |
|---|---|---|
| Impressions present, no GA4 AI Assistant activity | The AI answer may be satisfying the user on its own, or the link simply isn’t being clicked. | Manually test brand name mentions, source URL accuracy, and answer correctness. |
| Impressions present, Direct traffic is rising | May be AI-driven clicks arriving without a referrer; not conclusive. | Compare Direct landing pages and conversion quality over the same date range. |
| Impressions present, the wrong page is being selected | The model or Google may be picking a different URL it judges to answer the query more clearly. | Clarify internal links, headings, and on-page answer blocks. |
Content language shouldn’t be dictated purely by this report either. Rewriting all your copy into dry definitions and bullet lists just to appear on an AI surface erodes a page’s distinctiveness over time. The better path is to sharpen the key answer, source your claims, spell out your experience and its limits clearly, and then reinforce that structure with schema and internal links.
Related Coverage
- The GA4 AI Assistant traffic channel: how do you measure ChatGPT, Gemini, and Claude?
- What are GEO and AEO? Google’s approach to AI search and the counterarguments