Google has finally separated generative AI visibility from the rest of the data in Search Console.
On June 3, 2026, Google announced dedicated Generative AI performance reports for Search and Discover. The announcement fell on my birthday, which felt briefly thoughtful until I realized the gift was all impressions and no clicks.
The Search report shows impressions from AI Overviews and AI Mode, along with the URLs that appeared and breakdowns by country, device, and date. Google is initially rolling the reports out to a subset of website owners while it tests them and gathers feedback.
This closes an important reporting gap. Until now, website owners could not isolate which pages were appearing specifically within Google’s generative search experiences.
The new report gives us that separation, but not a complete picture of AI search performance.
At the moment, this is primarily an impression report. It tells us that links to our content appeared somewhere within Google’s generative features. That is useful, but it is not the same thing as knowing what Google did with the content, whether anyone noticed the source, or whether the appearance accomplished anything beyond making a graph move.
Google has given us a new diagnostic lens, not a new scoreboard.
What The Generative AI Report Includes
The Search report currently includes impressions from AI Overviews and AI Mode. Experimental Search Labs features are excluded, while generative features in Discover are covered in a separate report.
Website owners can segment the data by:
- Page.
- Country.
- Device.
- Date.
The report does not currently include:
- Queries.
- Clicks.
- Click-through rate.
- Average position.
- Citation placement.
- The passage used to support the answer.
- Conversion or revenue data.
Some of this may be added later. Google says it is still working with website owners to determine which additional metrics would be useful.
Google is also testing a separate Search Console control that allows site owners to include or exclude their content from AI Overviews, AI Mode, and generative AI features in Discover. Excluding a site prevents its links and content from appearing in or grounding those features without removing the site from traditional search results. The default setting is inclusion.
This is the one actual lever Google is providing alongside the new reporting. Before using it, however, site owners should understand that opting out means forfeiting both AI impressions and any traffic those features might send. The report can help establish a baseline before anyone reaches dramatically for the off switch.
For now, the report answers a relatively narrow, albeit useful, question:
Which parts of my website are appearing within Google’s generative search features, and how often?
What Counts As An AI Impression
Google defines an impression as an instance in which a link to a website is shown to a user within a generative AI feature.
The counting method changes according to the level of aggregation.
In the chart, data is aggregated at the property level. If two URLs from the same website appear in one generative response, they count as one impression in the chart total.
In the page table, the two URLs may each receive an impression.
Before anyone opens Excel and starts announcing that Search Console cannot add, remember that the difference comes down to aggregation.
Imagine that an AI Overview links to both a company’s product page and its supporting research report. At the property level, that may count as one impression. At the page level, each URL may receive one.
This means adding together page-level impressions will not always reproduce the property-level total. The math is not broken; the dimensions are just different.
AI Impressions Are Not Traditional Search Impressions
Both metrics are called impressions, but the user experiences are not equivalent.
In conventional search, an impression usually means a discrete listing was presented for consideration. In AI Overviews and AI Mode, the principal product is the synthesized answer. The supporting link may be prominent, one of many citations, hidden until expansion, or introduced after a follow-up question.
Google’s counting rules also vary by feature. In AI Overviews, a link must be scrolled or expanded into view before it counts as an impression. In AI Mode, a follow-up question is treated as a new query, and links shown in the subsequent response can generate additional impressions.
Please do not pour generative impressions and ordinary organic impressions into the same reporting bucket and call the resulting bull-jive “total search visibility.”
They are not interchangeable units, no matter how convenient the chart may be.
For the same reason, do not calculate a blended click-through rate. Yes, you can produce the number. The existence of such a number does not automatically make it mean anything.
What The Report Can Help You Diagnose
The real value of the report is not the total impression count. It is the ability to compare which pages Google uses in generative search with how those pages perform elsewhere.
Several patterns are worth investigating.
High Organic Visibility, Low AI Visibility
A page may rank well and receive substantial organic impressions while appearing rarely in the generative report.
That does not automatically indicate a problem. Not every query triggers an AI response, and not every page type is useful for synthesis.
Still, the discrepancy justifies a review:
- Does the page answer identifiable questions directly?
- Is the important information clearly stated?
- Are the headings descriptive?
- Can individual passages stand on their own?
- Is essential information hidden in tabs, images, videos, or JavaScript?
- Is the content relevant to queries likely to generate an AI response?
A page can be perfectly capable of ranking and still be a terrible source for synthesis.
Ranking gets the page into consideration. It does not guarantee that Google can lift a clean, useful answer from it. That’s on you, mon frère. That is where technical SEO still matters.
Modest Organic Visibility, High AI Visibility
The opposite pattern may be even more revealing.
A page with modest conventional visibility may receive a disproportionate number of generative impressions because it contains a useful definition, statistic, comparison, or explanation.
These are the pages I would study first.
Not because they reveal a secret AI ranking factor, and certainly not because one successful URL constitutes a strategy, but because they may show which parts of the site Google finds easiest to use.
Look for recurring characteristics:
- Direct answers near the beginning of sections.
- Strong heading structure.
- Original research or first-party information.
- Clear definitions.
- Useful tables or lists.
- Focused topical scope.
- Language that matches how users ask questions.
One successful URL is a clue. It is not a playbook.
Visibility Concentrated In A Few Pages
If most generative impressions come from a handful of URLs, group those pages by topic, content type, author, template, and intent.
Perhaps one template exposes content cleanly in HTML while another hides it behind scripts, tabs, or decorative bull-jive.
Perhaps the visible pages give direct answers while the invisible pages require readers to excavate the point from six paragraphs of positioning language.
Perhaps the site has published 300 articles on a topic, but only three say anything original.
The report will not diagnose the issue for you, but it can tell you where to start looking.
Changes After Content Revisions
The report can also help you monitor whether meaningful revisions correspond with sustained changes in generative visibility.
Relevant revisions might include:
- Adding a clear summary or definition.
- Updating outdated information.
- Consolidating overlapping pages.
- Improving headings.
- Making important content visible in HTML.
- Adding original evidence or expert commentary.
- Strengthening internal links.
Please resist the urge to change one heading, see the graph rise three days later, and announce that you have reverse-engineered AI Mode.
Demand, seasonality, competing sources, feature prevalence, and Google’s systems may all change during the same period.
A sustained increase following a substantial revision is useful evidence. It is still a signal, not divine revelation.
How To Work With The Data
You do not need a new 12-tool AI visibility stack to begin analyzing this report. You do need exports, consistent date ranges, sensible page classifications, and enough restraint not to confuse correlation with revelation.
A practical workflow:
1. Export AI-Visible URLs
Use a meaningful date range. Avoid drawing conclusions from a few days of rollout data.
2. Add Conventional Search Data
For the same URLs and dates, export organic impressions, clicks, CTR, average position, and top queries where available. The traditional Performance report provides query and click data that the generative report currently lacks.
The purpose is not to treat the impression types as equivalent. It is to identify pages whose generative visibility differs meaningfully from their conventional performance.
3. Categorize The Pages
Add attributes such as:
- Page type.
- Topic.
- Intent.
- Template.
- Author.
- Publication date.
- Last substantial revision.
- Funnel stage.
A list of URLs and impression counts is not analysis. It is inventory.
4. Investigate The Outliers
Look for pages that overperform or underperform in generative search.
Ask why one template, topic, or author appears more often than another. Then review the visible HTML, structure, answer placement, internal links, authorship, and source quality.
This is the point where someone actually has to look at the website.
I know. Shocking.
5. Add Analytics Separately
Use analytics to assess identifiable AI referral traffic and conversions.
Do not assume that generative impressions correspond directly with sessions. Search Console and Analytics measure different things and use different attribution methods. They have never reconciled perfectly, so there is no reason to expect AI search to suddenly deliver accounting-grade precision.
6. Track Trends, Not Daily Drama
Google warns that the newest Search Console data may be preliminary and subject to change.
Multi-week or monthly comparisons will usually be more useful than daily fluctuations.
If your dashboard sends an emergency alert because AI impressions dropped on Tuesday, your dashboard may be the problem.
What Not To Put In Your Executive Dashboard
The moment executives see a large new number labeled “AI,” someone will want it enlarged, colored green, and placed at the top of the dashboard.
Fine. Put it on the dashboard, if you must. Just do not let the font size imply a level of business meaning the metric cannot support.
Avoid reporting:
- A blended total of generative and traditional impressions.
- A blended CTR.
- “AI market share” based only on your own impression total.
- Revenue attributed to impressions without supporting evidence.
- A visibility increase presented as proof that one optimization worked.
- Page-level totals mislabeled as property-level exposure.
A responsible dashboard can include:
- Generative impressions over time.
- The number of pages receiving impressions.
- Topics and page types represented.
- The relationship between AI-visible pages and organic performance.
- Meaningful revisions made during the period.
- Identifiable AI referral traffic and conversions, reported separately.
- Questions and hypotheses requiring further investigation.
We already have enough dashboards celebrating numbers that went up without anyone being able to explain why they mattered. There is no reason to build another one.
A New Lens, Not A New Scoreboard
The arrival of dedicated AI search data in Search Console is important. Website owners can finally isolate part of their visibility within AI Overviews and AI Mode instead of trying to infer it from blended reports or third-party prompt samples.
The responsible response is neither to dismiss the data because it is incomplete nor to treat generative impressions as a definitive measure of AI search success.
Use the report for what it is: a way to identify where Google is willing to surface links to your content within generative experiences.
Its value comes from comparison.
Compare AI-visible pages with pages that rank conventionally. Compare topics, templates, structures, and content types. Compare visibility before and after meaningful revisions. Compare exposure with analytics and business outcomes where those connections can be supported.
Search Console’s generative report is not a final score. It is a diagnostic lens.
The useful question is not simply, “Did the number go up?”
It is: What can this new visibility teach us about how Google is using our content?
More Resources:
Featured Image: ImageFlow/Shutterstock
