Google’s generative AI reporting in Search Console is live, but the numbers may not mean what many SEOs assume. John Mueller recently replied to a Reddit breakdown of the new data and agreed that the current approach has serious limitations.
What the new report shows
Google began rolling out the Search Generative AI performance report in June 2026 and opened it up globally on August 31. The view focuses on impressions: when a URL appears in AI search surfaces like AI Overviews and AI Mode. Importantly, this is a filtered slice of the regular search performance report, not an extra layer of visibility on top.
Why the metrics can mislead
A Reddit user outlined several issues with the data, and Mueller confirmed that the assessment was essentially correct. The main problems include:
- Impressions don’t require a view: A link inside an AI Overview can count as an impression even if the user never scrolls far enough to see it.
- “Show More” links are undercounted: If a citation is hidden behind an expandable section, it only counts once the user expands it.
- Position is not your link’s position: Every link inside an AI Overview gets assigned the position of the AI block itself, not where it sits among the cited sources.
- Do not add the numbers together: The AI report is already included in your regular search data, so treating it as incremental will double-count.
Mueller’s response
Mueller said Google tried to document the nuances clearly, but conceded that position tracking is especially hard for AI surfaces. He explained that Search Console currently tracks position “as a block” for many search features, and that approach is not separated out in the generative AI report.
He also pushed back on relying on legacy rankings. As he put it, search results pages now have many ways for users to interact, so the old “position 1–10” model is hard to map or make useful for site owners.
Why this matters now
AI search visibility is becoming a core acquisition channel for many D2C and content brands, but the reporting still borrows from a ten-blue-link era. If you optimise around a flawed position metric, you can misjudge which pages are actually earning AI citations and which are merely appearing inside a block. That can distort budget choices for SEO, content refresh and performance campaigns.
For example, a product page may show a high AI impression count because the AI Overview block appeared near the top, but the actual link sat fourth among citations and was never seen. That is not a signal to scale the page until you confirm clicks or referral behaviour.
What SEOs should do instead
Don’t treat AI impressions as equivalent to click-through visibility. Instead, pair the report with other signals: query-level data, landing page engagement, conversions and your own on-site search or chatbot logs. If a brand appears in AI Overviews but never gets clicked, the new report may still paint an overly positive picture.
A practical framework: for each important URL, compare AI impressions with regular search clicks and CTR. Then bucket pages into three groups: visible and clicked, visible but ignored, and cited only behind “Show More.” That gives you a better sense of which content actually earns attention inside AI surfaces.
Mueller also invited ideas on what position tracking should look like for modern search. That’s a useful opening: SEOs who document what they need now may help shape the next version of this report.
Source: Search Engine Journal
