Google's generative AI performance reports, launched in June 2026 and rolled out globally by August 31, show impressions for AI Overviews, AI Mode, and Discover. They don't show clicks. They don't show queries. And Google's John Mueller, responding to a Reddit thread in September 2026, confirmed the reporting is inadequate and that Google doesn't have a better solution yet.
The company building AI search can't tell you whether AI search is sending you traffic.
What the Report Actually Measures (and What It Doesn't)
The generative AI performance report gives you impressions broken out by page, country, device, and date. No clicks, no CTR, no query or prompt-level data. The values are already included in your regular Web search performance report; the AI view is a filter, not a separate dataset. Add the two together and you're double-counting.
Three distortions make the data harder to trust than it looks:
- Phantom impressions. An impression counts when the AI Overview renders on the results page, whether or not the user scrolled far enough to see your link. Your "visibility" may be visibility to a server, not a person.
- Understated exposure behind "Show More." Links hidden behind the expansion toggle don't count as impressions until someone clicks to reveal them. If your content sits behind that fold, Search Console is understating your actual presence in AI answers.
- Position of the block, not the link. Every link inside an AI Overview inherits the position of the Overview itself. You see where the AI feature sat on the page, not where your URL ranked within it. For anyone trying to figure out citation prominence, this number is nearly useless.
Mueller's own words: "Position for these is hard to do in a way that makes it useful, so we're currently tracking it like we do for many search features (as a block), & it's not separated out in the Gen-AI performance report." He invited suggestions. That candor doesn't fix your dashboard.
Legacy Concepts in a Post-Legacy Format
Search Console was designed around ten blue links. Impressions, clicks, position, CTR. AI Overviews blow that model apart. A user might read your content synthesized into an AI answer, absorb your data point, form an opinion about your brand, and never click. That interaction has value. Search Console records it as an impression (maybe) and zero clicks.
Mueller acknowledged this: "Search results pages have a lot of ways for users to interact nowadays, so the old 'position 1–10' is hard to map, or to make useful for site owners." The old paradigm doesn't fit. Google hasn't shipped the new one. There's also no retroactive data. AI-related activity before June 2026 was blended into general Web data and can't be separated out. Treat the report as a forward-looking baseline, not a historical truth source.
What B2B Demand Gen Teams Should Do Instead
AI impressions are a signal, not a metric. They tell you which pages Google's AI is surfacing. They can't tell you whether that exposure converts to pipeline. The gap between those two things is where your measurement work lives.
Pair GSC AI impressions with GA4 and CRM data. Identify which URLs earn AI impressions in Search Console. Check GA4 for traffic patterns on those URLs. Then look at your CRM: are those pages associated with demo requests, assisted conversions, or pipeline? The answer won't be clean. It'll be directional. That's the best anyone has right now.
Run manual AI-answer audits. Search Console can't reliably distinguish whether your URL was cited inside an AI answer versus merely appearing elsewhere on the page. The only way to confirm citation is to search the queries yourself (or use a tool that does) and check. Pick your top 20 pages by AI impressions and audit them monthly. Tedious, but the only way to close this blind spot.
Strengthen decision-stage content. If AI is summarizing your top-of-funnel answers, the pages that still earn clicks are the ones AI can't fully replace: comparisons, pricing, integrations, alternatives. Prioritize refreshes on pages already earning AI impressions, but invest new production budget in content closer to the buying decision.
Don't opt out without a model. Google added a toggle for site owners to opt out of generative AI features. Opted-out sites won't receive traffic or impressions from those features. Before you pull that lever, build a rough estimate of what AI visibility contributes to your pipeline (even if directional). Opting out without knowing the trade-off is reckless.
The Measurement Gap Won't Close Soon
Google built AI search before building the instrumentation to measure it. Mueller's admission confirms what most operators suspected: the reporting is configured around concepts that don't fit the product. The interface caps at 1,000 rows per view. The API has its own ceilings. Long-tail AI queries disappear into the noise. And the company responsible for the gap is asking the community for ideas on how to fix it.
That's not a reason to ignore AI visibility. Pages showing up in AI Overviews may influence pipeline before a single click happens. But stop treating Search Console's AI report as a measurement system. It's a signal source. Build the measurement system yourself, with GA4, CRM attribution, and manual audits filling the gaps Google left open.
