GSC Generative-AI Performance Report: What It Shows

By Amit Patel

For two years, sites appeared inside AI Overviews with no way to measure it. That changed on June 3, 2026, when Google launched the Search Console Generative-AI performance report — dedicated reporting for your visibility inside AI Overviews, AI Mode, and generative features in Discover, plus a control to opt your content out entirely. This guide covers what the report measures, what it deliberately leaves out, and how to build a useful routine around it while the data is still impressions-only.

What the Report Shows

The Generative-AI reports show impressions your URLs earned inside generative-AI features — AI Overviews and AI Mode in Search, and generative features in Discover — broken down by page, country, device, and date, with granularity from hourly to monthly. Data begins May 18, 2026; there is no historical backfill before that date.

Structurally it sits alongside the classic Performance report rather than inside it: separate views for Search and Discover, each isolating the generative surfaces. The pages dimension is the immediately useful one — it names which of your URLs are being pulled into AI answers, which is information no third-party tracker could give you reliably.

Google announced the launch on its Search Central blog. Rollout is staggered — UK properties first (a consequence of regulatory pressure there), with a global rollout following — so if your property does not show the report yet, that is a queue position, not a problem with your site.

What It Does Not Show (Yet)

Three deliberate gaps: no clicks, no CTR, and no query data. You can see that a page appeared in AI features and how often — but not which prompts triggered it, whether anyone clicked through, or what the visit was worth. Google has said more metrics will come, without committing to a timeline.

That makes the report a visibility instrument, not a traffic instrument. Treat impression counts as a share-of-surface signal: they tell you whether Google’s generative systems consider your pages source material. For click-side truth you still need referral analysis in GA4 and direct testing — the method in our AI citation study framework — because Google surfaces are only one of the assistants quoting you. Coverage of the launch by Search Engine Land catalogs the gaps in more detail.

How to Read It Without Overreacting

Record a baseline month before drawing any conclusion: total AI impressions, your top 10 pages in the report, and the ratio of AI impressions to classic impressions for those pages. From then on, judge direction — rising, flat, falling — rather than absolute numbers nobody can benchmark yet.

Two comparisons matter early. Pages with high classic impressions but zero AI impressions are your optimization backlog — content Google ranks but its generative systems skip, often a structure and extractability problem covered in AI content formatting. Pages with strong AI impressions relative to their classic ranking are your templates: study what their structure does right and replicate it. Run both lists monthly, the same cadence as your indexing review.

The AI Opt-Out Control

Alongside the reports, Google shipped a control to block your content from AI Overviews, AI Mode, and Discover AI features. Opting out removes you from those surfaces — and their impressions — without affecting your classic Search rankings. For most publishers, opting out means forfeiting the fastest-growing answer surface.

The choice should be deliberate, not reflexive. AI Overviews now appear on a large share of queries, and being cited inside them correlates with stronger click-through for the cited brand — the evidence is assembled in our AI Overview optimization guide. The parallel decision for non-Google crawlers — GPTBot, ClaudeBot, PerplexityBot — still lives in robots.txt, covered in the robots.txt guide, and you can audit your current stance in seconds with the free AI Crawler Access Auditor.

Turning Impressions Into Action

The report tells you where you stand; changing the numbers is GEO work: direct-answer openings, extractable structure, cited statistics, and topical depth. Use the report as the measurement layer and the GEO playbook as the treatment.

Start with the strategy in the complete GEO & AEO guide, check your foundations with the AEO Readiness Analyzer, and track non-Google assistants with the AI Citation Tracker — GSC covers Google’s surfaces only. If you want the whole loop run for you, that is exactly what our services cover. And keep the classic Search Console routine running: AI visibility is built on the same indexed, crawlable foundation.

Frequently Asked Questions

Where do I find the Generative-AI report in Search Console?

It appears as dedicated performance views for generative-AI features under the Performance section, with separate reporting for Search (AI Overviews and AI Mode) and Discover. If you do not see it yet, the staggered rollout has not reached your property — UK properties received it first, with global availability following.

Why does my AI report show zero impressions?

Three common reasons: the rollout has not fully activated for your property, your pages are not yet being surfaced in generative features, or your content sits in topics where Google shows few AI Overviews. Check that your strongest pages are indexed and structured for extraction before assuming the worst.

Does appearing in AI Overviews reduce my clicks?

Industry studies point both ways: sites that are merely ranked under an AI Overview tend to lose clicks, while brands cited inside the Overview tend to gain them relative to non-cited competitors. The strategic goal is citation, not avoidance — measured now, for the first time, with first-party data.

If I opt out of AI features, do my rankings change?

Google states the opt-out affects generative-AI features only — AI Overviews, AI Mode, and Discover AI — and does not change your classic Search ranking. What you lose is presence on those surfaces and the impressions they generate, which for many queries is now the first thing searchers read.

Does this report cover ChatGPT or Perplexity?

No — it covers Google surfaces only. Visibility in ChatGPT, Perplexity, Claude, and other assistants requires direct testing and citation tracking, which is why we pair GSC’s report with an AI citation study and the free AI Citation Tracker tool.

The Bottom Line

The Generative-AI performance report is the first official measurement of something SEOs have been optimizing blind: presence inside Google’s AI answers. It is impressions-only and unevenly rolled out, but a baseline you start recording now becomes the “before” picture for every GEO improvement you ship next. Set the baseline, work the gap list, and revisit monthly — the metrics will deepen; your history should already exist when they do.