Google launches generative AI performance reports in Search Console

Google has added dedicated generative AI performance reports to Search Console, giving site owners measurable insight into how their content appears within AI Overviews, AI Mode and other generative search features. For UK businesses running SEO programmes built around traditional organic visibility, this marks the point where measurement infrastructure catches up with the shift in how search results are presented.

The new reports, announced through Google Search Central Blog, include separate views for Search and Discover surfaces. Site owners can now see impression data specifically tied to generative AI features, separate from traditional organic results. The data shows how often a site’s content is surfaced within AI Overviews, AI Mode and generative features in Discover, broken down by page, country, device and date.

What the new reports measure

The reports separate performance into two areas, Search and Discover. Within Search, the data captures impressions generated when a query triggers an AI Overview or AI Mode response that includes your content. Within Discover, the reports show how often your content appears in generative AI feeds. You can break the data down by page, country, device and date, though device data applies to Search only and query-level data is not available for AI features.

One important limitation is that the reports do not include click data. They show impressions only, so you can see how often your content appears in AI features but not how often users click through to your site. Google has said more metrics may follow, with no timeline confirmed. Even so, comparing your AI impression share against your traditional organic visibility gives a clearer view of where generative features are surfacing your content.

Implications for UK search strategy

For UK businesses, particularly those in professional services, healthcare, technology and B2B sectors, this changes how search performance is evaluated. Organic rankings still matter, but they no longer represent the full picture of search visibility. A site can rank well in traditional results and still fail to appear in AI Overviews if its content lacks the structured, cited and contextually rich signals that generative features prioritise.

This has direct implications for technical SEO and content structuring. The same ranking factors that drive traditional visibility do not automatically transfer to AI-driven surfaces. Schema markup, entity clarity, content depth and authoritative sourcing become more influential. Sites that have treated structured data as optional now face a measurable disadvantage in AI Overview inclusion.

What Priority Pixels is tracking

Priority Pixels is already integrating AI Overview performance into client reporting frameworks. We are tracking impression share within AI Overviews and AI Mode, along with the relationship between AI visibility and traditional ranking positions. For clients running AI SEO strategies, this data provides a direct feedback loop on whether content optimisation is translating into measurable presence within generative features.

AI Overview performance is not a replacement metric. It sits alongside organic rankings, featured snippet inclusion and knowledge panel presence as part of a broader visibility picture. The difference is that this visibility now has its own measurement layer.

We are also monitoring how AI Overview inclusion varies by sector and query type. Informational queries with clear factual answers tend to generate higher AI Overview impression volumes. Transactional and navigational queries still lean toward traditional result formats. For B2B clients, this means content strategy needs to account for both. Thought leadership and educational content should be structured to support AI summarisation. Service and product pages still need to rank in traditional results to capture bottom-of-funnel intent.

Structural changes required

The introduction of AI performance tracking does not mean abandoning existing SEO fundamentals. It means layering additional structural requirements on top of them. Content still needs to rank. It also needs to be machine-readable in a way that allows generative models to extract, cite and summarise accurately.

This requires clear entity definitions, explicit relationships between concepts and unambiguous sourcing. Pages that rely on implied context or assume prior knowledge will struggle to appear in AI Overviews. Generative models prefer content that defines terms, explains relationships and states conclusions clearly. Nuance and subtlety are less reliably extracted than direct statements and structured comparisons.

According to Google’s structured data documentation, implementing schema markup for articles, how-to guides, FAQs and reviews increases the likelihood of content being surfaced in rich results and AI-driven features. This is not speculative. It is documented behaviour that now has a corresponding measurement layer in Search Console.

What UK businesses should do now

Access the new reports in Search Console. Compare AI-driven impressions to your traditional organic performance. Identify pages that rank well but do not appear in AI Overviews. Those pages are structurally or contextually deficient in ways that matter to generative retrieval. Review schema implementation, entity clarity and content depth on those pages. Rebuild them with AI summarisation in mind.

Track changes over time. AI Overview inclusion is not static. As generative models evolve and as Google refines what content qualifies for summarisation, inclusion rates will shift. Sites that monitor this data quarterly will spot performance drops before they compound. Sites that ignore it will lose visibility without understanding why traditional metrics still look stable.

For UK businesses operating in regulated sectors such as healthcare, financial services or public sector delivery, this also introduces a compliance consideration. Content that appears in AI Overviews is being interpreted and recontextualised by a generative model. If that model misrepresents or oversimplifies critical information, the source site may still be perceived as responsible. Monitoring what gets summarised and how it gets presented becomes a reputational and regulatory concern, not just a traffic question.