What Google's New GSC AI Performance Report Actually Tells You

Google launched a dedicated AI Performance report inside Search Console on June 3, 2026. It surfaces impressions from AI Overviews and AI Mode -- the first time Google has provided publisher-accessible, first-party data on how often its AI systems surfaced your business in a response.

For businesses trying to understand their AI visibility, this is meaningful. But the report has a specific scope, one significant gotcha, and a set of limitations that determine how much it actually tells you.

What the GSC AI Performance report actually shows

The report surfaces AI Impressions: how many times Google's AI systems presented a response that included your content, across AI Overviews (the summary block at the top of Google results) and AI Mode (Google's conversational search experience). It lives inside Search Console under Search Results and updates continuously.

What it does not show: click-through rate on those impressions, who else appeared in the same AI responses alongside you, or any data from ChatGPT, Perplexity, Claude, or Microsoft Copilot. This is Google's surfaces only.

The comparison to Bing is instructive. In our June 2026 methodology work on the Bing AI Performance report (session 38, filed 2026-06-05), we documented that Microsoft's tool -- which launched in February 2026 for Copilot-specific grounding data -- carries the exact same limitation in the other direction: Bing covers its own surfaces, Google covers its own. Neither tells you about the other, and neither touches the platforms where most AI brand recommendations actually happen. Used together, they cover a real slice of AI search. But ChatGPT, Perplexity, and Claude remain outside both.

The opt-out trap that's silently zeroing your numbers

There is a setting in Google Search Console that has no obvious connection to AI visibility, but can produce a complete zero in the AI Performance report: the AI Overviews opt-out toggle.

Google shipped this opt-out alongside the AI Performance report, with the opt-out becoming enforceable June 17, 2026. The toggle lives in GSC Settings > Search Appearance > AI Overviews. If it's enabled, your content will not appear in AI Overviews or AI Mode -- and your AI Impressions count will show zero regardless of your schema, your directory presence, or your entity health.

In our June 14 methodology rec on the GSC report (2026-06-14-gsc-ai-impressions-baseline.md, session 47), we flagged this as the highest-priority check before interpreting any GSC AI data: "A client who has opted out will show zero AI impressions and will not appear in live query tests -- not because they lack presence, but because they have blocked AI features. This is an easy false negative to produce and an easy fix to recommend."

A zero AI Impressions count doesn't mean your business is invisible to Google's AI. It might mean someone checked a box, possibly during a period of experimentation with privacy or content controls, and never reversed it.

We've added this check to our pre-audit workflow. Before running live query tests on any Google surface, we now verify whether the client has the opt-out enabled. If the toggle is on and the exclusion wasn't intentional, reversing it may be the single highest-leverage action in the entire fix plan -- no schema work, no directory additions, just a settings change.

Using the report as a measurement baseline

The more durable use of the GSC AI Performance report is not diagnosis -- it's measuring improvement over time.

AI visibility fix plans have always had a measurement problem. How does a business know whether implementing schema changes and directory additions is actually moving the needle? A pre- and post-audit comparison answers that question at a point in time. The GSC AI Impressions report adds a continuous signal the business can check independently, between audits, without needing to run the full query suite again.

In the June 14 rec, we outlined what realistic timeline expectations look like for clients watching this report after implementing fixes:

- Schema and entity fixes (structured data updates, LocalBusiness markup, entity disambiguation): AI Impressions should begin increasing within 4-8 weeks - Directory additions (Bing Places, Foursquare, industry-specific listings): typically 6-12 weeks to surface in GSC AI data, since the directories themselves need to be crawled and re-indexed first

The longer window for directory work is the piece most businesses underestimate. A directory listing you add in July may not show up as a GSC AI signal until September. Setting that expectation upfront prevents clients from concluding the fix plan didn't work when the signal is simply still propagating.

This timeline matters specifically right now. Google AI Mode and Gemini Chrome auto-browse are rolling to US users this summer, with the agentic booking surface expanding through July and August 2026. Our June 11 and June 12 research on that rollout (agentic-browser-booking-layer.md, session 44) identified July 4 and July 15 as practical deadlines for schema and booking endpoint readiness. Businesses implementing those fixes in July can use the GSC AI Performance report to track whether Google's AI systems are responding -- with an expectation of early signal in September for schema work, later for directory changes.

One important limitation to state clearly: the report shows impressions only. No clicks, no CTR, no downstream traffic data. You cannot yet determine whether an AI Overviews appearance that included your business actually sent anyone to your site. The citation halo effect -- documented across multiple studies showing that AI citation presence correlates with downstream branded search volume increases -- suggests impressions and traffic are related. But GSC data doesn't currently confirm it directly.

What this doesn't replace

The GSC AI Performance report is worth setting up. It's free, it's first-party, and it's the only continuous Google-issued signal on AI citation activity available to publishers. For tracking improvement after implementing a fix plan, it's the right tool.

What it doesn't replace is answer-level query testing across platforms. The question "is AI recommending my business" cannot be answered by an impressions count alone, or by a tool that only covers Google's surfaces.

We've audited businesses with healthy GSC AI impressions that are being misattributed on ChatGPT, described incorrectly on Perplexity, and absent from Gemini's category recommendations in direct query tests. Those are not contradictions -- they reflect the structural reality that Google AI Overviews, Google AI Mode, ChatGPT, Perplexity, and Claude are drawing from different source architectures and producing different citation outcomes for the same business. The cross-platform divergence data in our knowledge base (platform-citation-behaviors.md, session 34) puts the domain overlap between ChatGPT and Perplexity citation pools at roughly 11% for the same queries. These are not slightly different versions of the same picture.

The GSC AI Performance report tells you what Google's AI systems are doing with your content on Google's surfaces. That's a real and useful signal. The full picture requires testing what each platform actually says when someone runs the queries your customers run -- which platform names you, which ignores you, which describes you accurately, and which platform sends them to a competitor instead.

Signal Check at sourcepull.ca runs that cross-platform query set across ChatGPT, Perplexity, Gemini, and Claude and returns per-platform scores with accuracy classification in under three minutes. Set up the GSC AI Performance report for continuous Google-side monitoring. Use Signal Check when you want to understand what's happening across all four platforms -- including the ones the GSC report can't see.