Search Generative AI performance reports are the AI Overviews and AI Mode reporting views Google added to Search Console on 3 June 2026 - the first first-party data any site owner has had on how often their pages surface inside Google's AI-generated search answers. The reports arrived almost two years after AI Overviews launched, and they finally answer a question AEO practitioners had only been able to guess at: which pages actually appear inside an AI-generated answer, and how often.
Before this rollout, measuring AI Overviews exposure meant inferring it indirectly - watching click-through rate erode on queries that used to send traffic, or manually retyping target queries into Google to see whether an overview appeared and whether your site was named. Neither method gave a reliable, page-level number. The new reports do, with real limits worth understanding before you act on them.
What the new reports actually show
The Search Generative AI performance reports sit alongside the standard Performance report in Search Console, with separate views for Search and for Discover. They break down impressions - how often a page was shown inside an AI Overview, AI Mode, or a generative AI feature in Discover - by page, by country, by device, and by date, with granularity down to the hour for recent data.
The rollout began in the UK before expanding to other markets, and the data itself only goes back to 18 May 2026. There is no historical backfill, so anyone hoping to compare current AI visibility against last year's baseline is out of luck. The clock effectively starts now.
What's missing: no clicks, no queries, no position
The single most important thing to understand about these reports is what they do not include. There is no click data, no click-through rate, no average position, and - critically - no query-level breakdown. You can see that a page appeared inside an AI Overview 4,000 times last week, but not which queries triggered it, and not whether anyone clicked through as a result.
That gap matters because impressions alone are a weak proxy for value. A page could rack up thousands of AI Overview impressions while sending zero traffic, because AI Overviews are built specifically to answer the query without a click. Without query-level detail, you cannot connect a spike in AI impressions back to a specific piece of content strategy, and without click data, you cannot tell whether that visibility is translating into anything measurable.
Treat the new AI performance reports as a visibility signal, not a performance metric. Impressions tell you that Google's AI surfaces know your page exists and consider it relevant enough to draw from. They do not tell you whether that citation is doing any commercial work.
The AI content opt-out toggle
The same release added a content control toggle, effective from 17 June 2026, that lets a site stop appearing in AI Overviews, AI Mode, and AI Overviews within Discover, while leaving standard organic search results and the regular Discover feed untouched. It is a narrower version of the old 'nosnippet' approach, aimed specifically at AI-generated surfaces rather than search as a whole.
For almost every commercial site, enabling this toggle would be counterproductive. Citation inside an AI-generated answer is exposure most brands are actively trying to win, not avoid - it is the entire premise of answer engine optimisation. The toggle exists for publishers with a specific reason to withhold content from generative reuse, such as subscription or licensing concerns, not for the typical site trying to grow AI visibility.
Where to find the reports in Search Console
- Sign in to Search Console and select the property you want to check
- Look for the new 'Search Generative AI performance' or 'AI performance' section in the left-hand navigation, separate from the standard Performance report
- Open the Search view first for AI Overviews and AI Mode impressions, then check the Discover view for generative AI features in Discover
- Filter by page to see which URLs are being drawn into AI-generated answers most often
- Cross-reference the pages showing up here against your standard Performance report to see whether the same URLs are also losing click-through rate on their top queries
How to fold this into an AEO measurement stack
Impressions-only data is still useful, provided you pair it with the layers that fill in what it leaves out. If you are tracking AI referral traffic and citation rate already, the new Search Console reports slot in as a fifth layer: a directional read on Google-specific AI visibility that complements, rather than replaces, GA4 referral tracking and manual citation testing across ChatGPT, Claude, and Perplexity.
- Check the AI performance report weekly for the first month to learn its normal range for your site, then move to a monthly cadence once you have a baseline
- Cross-reference pages with high AI impressions against the standard Performance report for the same URL - a page climbing in AI impressions while losing clicks on its best queries is a strong sign it is being answered inside the overview rather than driving a visit
- Do not use AI impressions alone to justify a content decision. Pair the trend with your manual citation testing protocol and with an AEO readiness check on the page, so you understand why a page is or is not being drawn into AI Overviews, not just whether
- Watch country and device splits, since AI Overviews rollout has varied by market and by mobile versus desktop, and a flat overall trend can hide a fast-growing segment
The limits are the point, for now
It is tempting to read the absence of click and query data as an oversight Google will fix quickly. It is more likely a deliberate first step. Google has historically rolled out Search Console features in stages - Discover performance reporting itself launched with far less granularity than it eventually gained. Expect the same pattern here: impressions first, richer dimensions later, if AI Overviews adoption continues to grow.
In the meantime, the practical answer is not to wait for a complete dataset. The reports are free, already live in your existing Search Console property, and directionally useful today. Combined with a citation testing routine and a schema and content structure review, they give you the closest thing to a real AEO dashboard that exists for Google's AI surfaces specifically.
If your property shows little or no data in the new AI performance report, that is itself informative. It usually means either your pages are not being retrieved for the queries that trigger AI Overviews, or your content is not structured in a way the retrieval layer favours. A CiteRank audit will tell you which.