Google Search Console's Generative AI Report: What It Shows and How to Use It

The short answer: Google Search Console now includes a Generative AI features report, under Performance, showing how often your pages appear inside Google's AI surfaces - AI Overviews and AI Mode - as impressions, clicks and position. It is the first official measurement of AI search visibility. It is currently a Beta report in the Search Console interface only: the Search Analytics API does not yet expose the data, so any tool promising to pull it programmatically is ahead of what Google actually ships.

Where to Find the Report

Open Search Console, choose your property, and go to Performance. Alongside the familiar Search results report there is now a Generative AI features entry marked Beta. It shows total impressions and clicks with a daily chart, and breakdowns by page, country and device.

An impression here means one of your pages was cited or surfaced inside an AI-generated answer - either an AI Overview at the top of a results page or a response in AI Mode. A click means someone followed that citation through to your site. Query-level data is not shown yet, which is the report's biggest current limitation: you can see which pages appear in AI answers, but not which questions triggered them.

Why This Report Matters

Until now, AI search visibility was guesswork. Businesses could see referral traffic from ChatGPT or Perplexity in their analytics, but Google's own AI surfaces - by far the largest by audience - were a black box. You could rank well and still have no idea whether AI Overviews were citing you or quietly answering with competitors.

This report ends the guesswork for Google's surfaces. It also makes AI visibility a number you can be accountable for: a baseline to record, a trend to watch, and a way to test whether content changes actually earn AI citations. If AI search matters to your strategy - and for UK service businesses it increasingly does - this is the report to check monthly alongside your regular website SEO fundamentals.

What the API Supports (Tested, Not Guessed)

Several guides circulating online show code that filters the Search Analytics API by a GENERATIVE_AI search appearance to pull this data programmatically. We tested this directly against a live property with full API credentials, and the result is unambiguous: the API rejects the expression - "not a valid searchAppearance for type WEB". We then asked the API to list every search appearance type it recognises for the property; no AI-related value exists yet.

The plain reading: the interface report is ahead of the API, which is normal for Search Console Betas. The data will very likely reach the API later, as most reports eventually do. Until then, the numbers live only in the interface, and the practical approach is a monthly manual record - or an integration built to detect the moment Google switches the API on, which is how we have wired our own reporting.

How to Actually Grow AI Impressions

The report measures outcomes; it does not create them. Watching the chart changes nothing. What earns AI citations follows a consistent pattern:

• Rank first, get cited second. AI Overviews overwhelmingly cite pages that already rank well for the underlying query. A page sitting on page five of results will not be quoted by the AI answer above page one. Ordinary ranking work - relevance, authority, internal linking - remains the foundation.
• Answer questions directly. Pages that open with a clear, self-contained answer to a specific question are far easier for AI systems to lift than pages that circle the topic. One question per page, answered in the first paragraph, then explained in depth.
• Keep business facts machine-readable. Consistent structured data, a single business entity across every page, and honest FAQ blocks give AI systems verifiable facts to work with. AI engines recommend what they can verify.
• Let the crawlers in. AI fetchers are strict about access. Aggressive bot-blocking, broken HEAD responses, or security rules that ban rapid readers will silently remove you from AI answers while your normal rankings look fine. Crawl health is now AI visibility health - a theme we cover in depth in our AI search and LLMO guide for UK businesses.

Reading the Numbers Sensibly

Two warnings from experience. First, small sites will see small numbers - hundreds of impressions, not thousands - and that is normal at the start. The metric that matters is the trend across months, not the absolute figure in week one. Second, do not compare your numbers against big-site case studies. A local business appearing in AI answers for its town and trade can win real customers from a few hundred well-targeted impressions, because the searcher intent behind an AI answer citation is often strong.

Record a baseline now, change one thing at a time - an answer-first rewrite, a new question-focused page, a structured data fix - and check the report on a fixed monthly date. That loop, not any single trick, is what moves AI visibility. It is the same measurement discipline that underpins good SEO work generally, applied to a new surface.