Search Console Generative AI Report: How to Read It
The generative AI report in five lines
- 1What it is → a Search Console report counting how often your URLs appeared in Google's generative AI features: AI Overviews and AI Mode in Search, plus AI features in Discover.
- 2Metric → impressions only. It doesn't show which queries triggered them or how many clicks they produced.
- 3Not new traffic → these impressions were already inside your main Performance report totals. The new report separates them out.
If you've opened Google Search Console recently, you've probably seen a new banner: "Get more details on your site's performance in generative AI features on Google Search." It links to the Search Generative AI performance report, which Google rolled out to every website in late August 2026. It answers a question site owners have been asking since AI Overviews launched: is Google using my pages in its AI answers? It answers that question only partly, though, and it's easy to misread. This guide covers what the report measures, what it leaves out, and a practical way to use it alongside the Performance report you already know, using this site's own numbers as the example.
What the Generative AI Report Measures
Google's announcement defines the core metric plainly: impressions are how often URLs from your site appeared in generative AI features in Search and Discover. There are two views:
- Announced June 3, 2026, initially for a subset of websites.
- Rolled out to all websites worldwide as of August 31, 2026, according to an update note on Google's announcement.
- These impressions were already counted in your overall Performance report. The new report breaks them out; it doesn't add to your totals.
| report | covers | breakdowns |
|---|---|---|
| Search | AI Overviews and AI Mode | Pages, countries, dates, devices |
| Discover | Generative AI features in Discover | Pages, countries, dates |
What It Doesn't Tell You
The gaps matter more than the features, because they decide what conclusions you can safely draw:
- No query data. You can see that a page appeared in AI answers, but not for which searches.
- No click data in the report itself, so you can't calculate a click-through rate for AI features from it.
- No position inside the answer. An impression could be a prominent citation or one link among many.
- One AI Mode answer can draw on several URLs at once, so an impression is not the same as ranking first.
- Third-party analyses report that data starts in mid-May 2026, with no earlier history to compare against.
A Real Example: Reading This Site's Numbers
Here's why the separate report is useful. StackNova's own Performance report for the 28 days to October 11, 2026 shows 6,240 impressions, 14 clicks, an average CTR of 0.2% and an average position of 8.6. Over six months the totals are about 19,900 impressions and 81 clicks.
An average position of 8.6 normally means page one. A CTR of 0.2% at that position is far below what a page-one result usually earns. Some of that gap comes from how Google counts: one search can generate impressions for several of the same site's URLs, and positions are averaged. A separate explanation is worth checking too: pages surfacing in AI Overviews or AI Mode, where the answer satisfies the searcher before they click.
The generative AI report is how you check that second explanation rather than guess. If the pages with high impressions and near-zero clicks in the Performance report are also the pages with the most AI-feature impressions, AI answers are absorbing the click. If they aren't, the problem is more likely the title, the snippet or search intent.
How to Use the Report: A 20-Minute Workflow
You don't need any tools beyond Search Console and a spreadsheet.
- Step 1: In the Performance report, set the last 3 months, open the Pages tab and export it. You get clicks, impressions, CTR and position per URL.
- Step 2: Open the generative AI report (Search view), set the same date range and export the Pages table.
- Step 3: Join the two exports on URL in a spreadsheet. Add a column for AI impressions as a share of total impressions.
- Step 4: Filter for pages with a high AI share and a CTR under 0.5%. These are pages Google uses in answers but readers don't visit.
- Step 5: For each one, check what the AI answer says for its main query (search it yourself in an incognito window). If the answer fully covers your page, your page needs something the summary can't give: original data, a tool, a template or worked examples.
What to Change on Pages AI Answers Already Cover
If an AI answer can summarise your whole article in four lines, the article has a reason-to-click problem, and more keywords won't fix it. Changes that give readers a reason to click through:
Lead with something a summary can't carry
A calculator, a downloadable checklist, a code sample that runs, a comparison table with your own measurements. AI answers paraphrase text well and reproduce interactive or first-hand material poorly.
Stop repeating the answer five times
Many pages restate the answer in a summary box, an intro, a takeaways list and a conclusion. That makes the page easier to summarise and gives readers who got the gist in the first few seconds no reason to stay. Answer once, briefly, then go deeper.
Match the title to what the page uniquely offers
"What is X" titles compete directly with the AI answer. "X: tested on a real project, with the config file" promises something the answer doesn't have.
Keep facts current
Pages on fast-moving topics lose trust when AI answers show newer facts than your page. Add a visible "last reviewed" date and update it when the facts change.
Frequently Asked Questions
Strategic Summary
Final Thoughts
The generative AI report is the first first-party data on whether Google uses your pages in AI answers. It's useful as long as you read it for what it is: a count of appearances, without queries or clicks. Joined with your Performance report, it shows which pages AI answers already cover and therefore which ones need something a summary can't replace. For the structural side of earning citations, see AI SEO strategy for teams. For a test of one popular tactic that turned out to do little, see whether llms.txt does anything. --- Last reviewed: October 20, 2026. Report details are from Google's Search Central announcement and its August 31 update; the example figures are from StackNova's own Search Console. No affiliation with Google.
Run the 20-minute workflow on your own site and tell me what share of your impressions come from AI features. I'll compile anonymised results in a follow-up.
Next Up
Continue your research
Sources & Research
Google Search Central Blog: Introducing Search Generative AI performance reports in Search Console (June 3, 2026, updated August 31, 2026)
https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
Evolv: Google's new Search Console AI reporting: what it shows and what it doesn't
https://evolvagency.io/learn/generative-search/google-search-console-generative-ai-performance-reports
Google Search Central: Performance report documentation
https://support.google.com/webmasters/answer/7576553
