Google Search Console now reports how often links to your pages appear in AI Overviews and AI Mode, but only as impressions, with no clicks, no queries and no split between the two surfaces. Google announced the Search generative AI performance reports on its Search Central blog on June 3, 2026, and its help documentation says they reached all websites as of August 31, 2026. This post explains what the report shows, how counting works, what it can't tell you, and the analyses worth running each month.
What Search Console reports today
AI feature data now lives in two places in Search Console, and they measure different things.
| Where | What it shows | What it doesn't |
|---|---|---|
| Search generative AI performance report (Performance section) | Impressions of links to your site in AI Overviews and AI Mode, by page, country, device and date | Clicks, CTR, position, queries, AIO vs AI Mode split |
| Discover generative AI report | Impressions in generative AI features in Discover | Search data |
| Standard Performance report, Web search type | Clicks, impressions, CTR and position for all Search results, with AI Overviews and AI Mode included | Any filter to isolate AI features |
Google's announcement says generative AI impression data also remains in the overall Performance report. That means the dedicated report is a view into a subset of your totals, not extra visibility on top of them.
How counting works
The definitions decide what the numbers mean, so learn them before reporting anything. Based on Google's documentation and its earlier statements about AI features:
- Impression: counted when a link to your site is shown to a user in a generative AI feature. Being paraphrased without a visible link isn't an impression.
- Same URL in two places: if a page appears in an AI Overview and as a regular result on the same results page, Search Console aggregates them rather than counting both.
- Chart vs table: like the rest of Search Console, the chart counts by property and the Pages table counts by page, so two of your pages in one AI Overview count once in the chart and once each in the table.
- Position (standard report): Google has documented that an AI Overview occupies one position, and every link in it gets that position. This affects average position in the Web report.
- Clicks (standard report): Google has said clicking a link in an AI Overview or AI Mode counts as a click, and that follow-up questions in AI Mode are treated as new queries.
These rules make the numbers best for trends and page comparisons, not for precise counts of how many people saw your brand.
Analyses worth running
The report is most useful at the page level, so build your analysis around pages. Five worth doing:
1. Which pages appear in AI features at all
Export the Pages table for the last three months. Compare it with your list of priority pages (pricing, comparisons, top guides). Priority pages with few or no AI impressions are candidates for content work; see getting your content into AI Overviews.
2. AI impressions vs Web clicks per page
Join the generative AI Pages export to the standard Performance Pages export. Pages with high AI impressions and few Web clicks may be answering the question fully inside the AI response. That isn't necessarily bad; it's visibility. But it tells you where to strengthen reasons to click, such as tools, templates, data or pricing detail. More in zero-click answers.
3. Before and after content changes
Annotate the dates you change a page, then watch that page's AI impressions over the following weeks. Compare with similar pages you didn't change, so a sitewide trend isn't mistaken for your edit's effect.
4. Country and device splits
AI features roll out unevenly by market and device. Country and device views show where you're exposed to AI features and where classic results still dominate.
5. Landing-page outcomes in GA4
Match the pages with high AI impressions to their GA4 landing-page performance: engagement, sign-ups and conversions. Clicks from AI Overviews and AI Mode generally arrive in analytics as ordinary Google organic traffic, so this is a page-level association, not attribution. See tracking AI referral traffic in GA4.
Workarounds for the missing query data
The biggest gap is the absence of queries, which you can partly reconstruct. Three approaches:
- Use the page's Web queries as a proxy. The standard Performance report shows which queries a page ranks for; the AI impressions likely come from a related set.
- Group pages by topic. Report AI impressions by topic cluster (pricing, comparisons, how-to) instead of page by page.
- Track prompts directly. Run your key questions in AI Mode and check AI Overviews for your target queries on a schedule to see which pages appear for which questions. See how to build a prompt set.
Example: a monthly Search Console AI summary
This example is illustrative; the numbers are made up.
| Metric | September | October | Note |
|---|---|---|---|
| Generative AI impressions (Search) | 41,000 | 47,500 | Up after comparison page refresh |
| Pages with any AI impressions | 212 | 236 | 24 new pages appearing |
| Priority pages with AI impressions | 31 of 40 | 34 of 40 | Pricing FAQ now appearing |
| Web clicks to those 40 pages | 18,900 | 19,400 | Roughly flat |
| Sign-ups from those landing pages (GA4) | 310 | 335 | Page-level association only |
The commentary would note that AI visibility rose on the refreshed comparison pages, clicks held steady, and six priority pages still have no AI impressions, which become next month's content work.
The Search generative AI control
Search Console also added a setting that lets you exclude a property from Google's generative AI features. Google's help documentation describes it as a Search generative AI control covering AI Overviews, AI Mode and AI Overviews in Discover; the default is to include. Google says excluded sites won't receive traffic or impressions from those features, that the control doesn't affect AI training (Google-Extended is the control for that), and that it isn't used as a ranking signal outside those features.
For most brands that want to be recommended, excluding is self-defeating: you give up visibility on a growing surface. Publishers with licensing concerns may weigh it differently. If you're considering it, compare it with the narrower training decision in should you block AI training crawlers?.
What Search Console still can't tell you
Search Console covers Google only, and only impressions for AI features. It can't show:
- Whether your brand is mentioned without a link
- How you're described, or whether facts are correct
- Which competitors appear alongside you
- Your visibility in ChatGPT, Perplexity, Claude, Copilot or the Gemini app
That's why it's one panel in a broader dashboard rather than the whole picture. See the AEO reporting dashboard.
The bottom line
Search Console's generative AI report finally gives you first-party data on AI Overviews and AI Mode, but it's impressions-only, page-level and combined across both surfaces. Use it to see which pages appear in AI features, compare that with Web clicks and GA4 outcomes, measure the effect of content changes against untouched pages, and fill the gaps (queries, mentions, accuracy, other assistants) with prompt tracking. Don't add its numbers to your overall totals, and think hard before using the exclusion control.
For Copilot and Bing, the Bing AI Performance report adds free citation data.
