On June 3, 2026, Google quietly shipped something the SEO and GEO communities have been asking for since AI Overviews launched: a dedicated performance report inside Search Console that tracks your site’s visibility inside AI-generated answers. No more guessing. No more proxy metrics. Actual impression data, per page, for generative AI features.
The problem is that the report is currently rolling out only to UK-based website owners, the data it provides is narrower than most people expect, and the decisions it forces you to make — especially around the new content-blocking toggle — have long-term implications most brands aren’t thinking through carefully.
Here’s a complete breakdown of what the report shows, what it’s still missing, how to read the numbers, and what actions to take on the pages where AI impressions are low or zero.
What the Gen AI Performance Report Actually Tracks
Inside Google Search Console, under the Performance section, there is now a separate “Search generative AI” view alongside the standard organic search report. The report covers two surfaces: Search (AI Overviews and AI Mode) and Discover (generative AI features in the Discover feed).
The data dimensions available are:
- Impressions — how many times a URL from your site appeared in a generative AI feature
- Pages — which specific URLs are getting AI impressions
- Countries — geographic breakdown of where your AI visibility is coming from
- Devices — mobile vs. desktop (available for Search; Discover only breaks out by country and date)
- Dates — daily, weekly, or monthly views going back to when the feature launched
That’s the full scope. Five dimensions, all focused on impressions.
What’s Not in the Report (And Why It Matters)
The data the report omits is more significant than what it includes. There are currently no clicks, no click-through rate, no average position, and no query-level data. Google has confirmed it is “continuing to work with website owners to understand what insights will be most helpful” but has given no timeline for when query or click data will appear.
This creates a specific analytical problem: you can see that a page appeared in AI answers, but you cannot see what question triggered that appearance. A product page might show 3,400 AI impressions, but you have no way to know whether users were asking about your brand specifically, comparing it to competitors, or asking a category question where you happened to get cited.
For now, the workaround is prompt tracking. Tools that systematically query AI platforms with a fixed set of prompts relevant to your category and record whether you appear — run at regular intervals — give you the query-level context that GSC currently can’t. Without this layer, the impression data tells you that you’re appearing but not why, which makes it difficult to reproduce the result or fix a drop.
The Content Blocking Toggle: What It Is and Why You Probably Shouldn’t Use It
Alongside the new report, Google added a toggle in Search Console that lets you opt your entire site out of appearing in AI Overviews, AI Mode, and related generative AI features. If you opt out, your content will not appear in AI answers — and you will receive no impressions or traffic from those features. Your standard organic search rankings are not affected.
This toggle exists partly because Google is navigating publisher pushback around content being used in AI responses without clear traffic attribution. But for the vast majority of website owners, opting out is the wrong call, for three reasons:
- AI Mode traffic is real and growing. Google AI Mode crossed 1 billion users in May 2026. Opting out removes your brand from answers seen by a user base larger than most countries’ entire internet populations.
- Cited brands build recall even without clicks. Research from Semrush’s Ghost Citations study found that 72% of AI brand appearances generate zero direct clicks — but brand recall and subsequent direct searches increase significantly when users see your brand name in a confident AI answer. Opting out eliminates even the zero-click visibility.
- You can’t opt back in selectively. The toggle applies to the entire site, not individual pages. If your homepage and product pages are getting cited but your blog is the source of a problematic topic, the only current option is to remove specific pages from Google’s index entirely — which has much larger SEO consequences.
The only scenario where opting out makes tactical sense is for highly regulated industries (legal, financial, medical) where an incorrect AI-generated summary of your content could create liability — and even then, the risk management conversation belongs with legal, not marketing.
How to Read Page-Level Impression Data
Once you have access, the most actionable view is the Pages tab sorted by impressions. Run the report for the last 90 days. What you’re looking for is the distribution pattern, not just individual page performance.
A healthy AI-visible site typically shows a small number of pages with very high impressions and a long tail of pages with low but nonzero impressions. If your report shows one or two pages accounting for more than 60% of all AI impressions with everything else at zero, that is a concentration problem — your site is likely appearing only for brand-name queries, not category or comparison queries where new users can discover you.
Three patterns to flag immediately:
- High-traffic pages with zero AI impressions — Pages that rank in the top 10 organically but show no AI impressions indicate a content structure problem. These pages are findable by Google’s crawler but are not being surfaced as answers to conversational queries. The fix is usually restructuring the content to answer a specific question directly, near the top of the page, with supporting evidence in the subsequent paragraphs.
- High AI impressions on pages with thin content — If Google is showing a page in AI Overviews but the page has fewer than 600 words or lacks specific data, you are likely appearing in summary form without your key claims being accurately represented. Expand the page with structured, numbered data points and add an explicit answer to the primary question within the first 150 words.
- Impression drops over a 30-day window — A sustained 20%+ drop in impressions for a specific page usually signals either a content freshness issue (the page hasn’t been updated and a competitor with newer data has displaced it) or a structural change in how AI Mode is handling queries in that topic cluster. Check whether the page’s publication date is more than 90 days ago and whether any cited statistics in the content are now stale.
The Platform Diversification Problem the GSC Report Won’t Solve
The new Search Console report covers Google surfaces only. It tells you nothing about your visibility in ChatGPT, Perplexity, Claude, or Bing Copilot — which collectively represent a large and growing share of AI-mediated queries.
As of June 2026, ChatGPT holds approximately 54.7% of worldwide AI chatbot web traffic. But Claude’s traffic share has tripled over the past 12 months, climbing from 1.6% to 8.9% of all generative AI traffic. In the United States specifically, Claude now accounts for 12.5% of AI chatbot market share. These platforms use different retrieval systems, prioritize different source types, and have different freshness thresholds for what they consider citable content.
A brand that scores well in Google AI Mode impressions (appearing frequently in your GSC report) can simultaneously be invisible in ChatGPT and Perplexity. Studies analyzing source overlap between platforms have found that ChatGPT and Perplexity share only 11% of their cited sources. This means that optimizing for one platform does not transfer to the others.
The practical implication: the GSC Gen AI Performance Report is your first real measurement instrument for Google AI visibility, but it is one instrument tracking one platform. It should be used alongside systematic prompt testing across ChatGPT, Perplexity, Claude, and Gemini — not as a replacement for cross-platform monitoring.
Five Immediate Actions Based on the Report Data
Once you have access to the Gen AI Performance Report, here is the diagnostic sequence that will produce the clearest picture and the fastest improvements:
- Export page-level impression data for the last 90 days. Sort by impressions descending. Identify the top 10 pages and the pages you expected to appear but show zero impressions.
- Run the zero-impression pages through a manual AI test. Ask ChatGPT or Google AI Mode a question that the page should answer. If the page doesn’t get cited, read what does get cited and compare content structure, word count, presence of specific data points, and author attribution.
- Check freshness on your top-impression pages. If any page in your top 10 has not been updated in the last 60 days and contains statistics from 2025 or earlier, schedule an update. Research from multiple sources points to 90 days as the threshold after which AI citation frequency begins to decline without content refreshes.
- Add a direct-answer block to zero-impression pages. The highest-cited content format across AI platforms is a direct question-and-answer structure in the first 200 words of a page. Rewrite the introduction of each zero-impression priority page to lead with the specific question the page answers and provide a 2-3 sentence direct answer before expanding with detail.
- Set up a monthly impression comparison. The report supports date-range comparison. Once the rollout reaches your account, bookmark a monthly comparison of the current 30 days vs. the prior 30 days. A consistent downward trend in impressions across multiple pages is an early signal of a content freshness problem or a shift in how AI Mode is handling your topic cluster — and catching it at 30 days gives you time to respond before the drop becomes structural.
What Google Still Hasn’t Said
Google’s official guidance from its June 2026 AI Optimization documentation continues to position generative AI visibility as an extension of standard SEO: high-quality content, crawlable structure, and experience-driven authorship. The official position is that llms.txt files, special schema, and content chunking are not required or explicitly helpful for AI features.
What the citation data shows, however, is more complicated. Studies analyzing thousands of AI Overview citations consistently find that 83% of citations come from pages outside the organic top 10, meaning search rank and AI citation frequency are partially decoupled. Domain authority correlates with AI visibility but does not determine it. The characteristics that most reliably predict AI citation — specific numerical claims, named authors with verifiable credentials, direct answer structures, and content freshness within 60 days — are not the same signals that have historically driven organic ranking.
The Gen AI Performance Report is a step forward because it gives you real measurement data. But the strategy gap between “appears in GSC impressions” and “appears across all AI platforms for the queries that matter” remains significant — and no single tool currently closes it in one view.
Getting Your Baseline Before Broader Rollout
If you are in the UK, you likely have access now. If you’re outside the UK, access is coming — Google has confirmed global rollout is planned, though no date has been announced. When it arrives in your account, the most important thing is to pull baseline data immediately and save it. The report does not retain historical data from before your rollout date, so whatever impressions were accumulating before you gained access are not visible.
The sooner you establish a baseline, the more diagnostic value you get from the month-over-month comparison. Teams that wait until rollout to start thinking about AI visibility measurement will be two or three months behind teams that have already built reporting workflows around the data.
If you want to see your current AI visibility across ChatGPT, Perplexity, Google AI Mode, and other platforms — before the GSC report rolls out to your account, and with the cross-platform query-level data that GSC doesn’t provide — run a free audit at ai-visibility.llmagnet.com. It scores your site across all major AI platforms and shows exactly which pages are being cited, for which query types, and where the gaps are.