Search Console AI Report: What It Shows and How to Read It

What is the Search Console AI report?

The Search Console AI report is a performance report that shows how often your pages appear in Google's generative AI features. It measures impressions in AI Overviews, AI Mode and the generative AI features in Discover. It gives you a dedicated view, separate from classic search results.

Google announced the reports on the Search Central blog under the title Search Generative AI performance reports. So you no longer have to guess your AI visibility. Instead, you can read Google's own data.

Our team takes this announcement seriously. Clients ask us the same question again and again: "Do we show up in AI answers?" Until now, we answered with third-party tools and rough estimates. From now on, Search Console is our first source.

Below, we explain what the Search Console AI report measures, what it still lacks and how you can read it without fooling yourself.

When and how did Google announce the AI reports?

Google published the announcement on the Search Central blog in June 2026. It started by testing the reports on a limited group of sites, first in the United Kingdom. Then it opened them step by step. According to Search Engine Journal, the rollout reached all sites worldwide by 31 August 2026.

So if you manage a property today, you should see the report. If you do not, check your property type and your permission level first. Also remember that the report is new, so give it some time to collect data.

The timing matters. Users now ask more questions inside AI surfaces, while site owners watched those surfaces from a blind spot. The new report closes part of that gap, at least at the impression level.

The official blog archive lists a second related item from the same period: a new resource on optimizing for generative AI in Google Search. In September 2026, Google also announced separate reporting for multimodal search. We covered that report in our guide to the Search Console multimodal search report. Together, the two news items show the direction: search surfaces multiply, and reporting follows.

Which surfaces does the Search Console AI report measure?

The report covers impressions from three surfaces. Search Engine Journal names them as AI Overviews, AI Mode and the generative AI features in Discover. Google's announcement also describes separate reports for Search and for Discover.

• AI Overviews: AI summaries that appear above the regular results.
• AI Mode: A conversational AI search experience where users continue their query.
• Generative AI in Discover: AI-powered areas inside the Discover feed.

This split matters because each surface serves a different intent. AI Overviews mostly answer information queries. Discover, in contrast, is a discovery feed where users do not search at all.

Each surface also follows its own habits. For example, Discover suggests content when the user is not looking for anything, so visuals, headlines and freshness weigh more there. Therefore, read the surfaces one by one instead of trusting a single total.

Which dimensions and metrics does the report include?

The report is impression based. According to Search Engine Journal, you see data by page, country and date. The Search reports also add a device breakdown. So the Search Console AI report lets you answer a simple question: which page appeared in an AI feature, in which country and on which day?

This structure lets you think in content types. For example, your guides may appear often in AI Overviews while your product pages never appear. That picture shows the gap between content types for the first time.

The country dimension helps multilingual sites in particular. Your English pages may earn impressions in other countries than your German pages. In practice, that hint tells you which language deserves more investment.

The device split helps too. If you appear more on mobile, you can prioritize mobile page experience. However, always check the sample size. On small sites, a handful of impressions can be pure chance.

Which data is still missing from the report?

The answer to this question is the key to reading the report correctly. Search Engine Journal reports that the report had no click data at publication. It also has no query breakdown. So you see that your page appeared, but you cannot see whether the user clicked.

This gap shapes how you should use the report. An impression is a visibility signal, not traffic. Therefore, never treat the report as a success metric on its own. Read it together with GA4 and the classic Search Console performance data.

Note that these details describe the period of the announcement. Google may expand the report over time. Before you decide anything, check the current scope in the Search Console help pages.

Another limit is time depth. A new report may not offer a long history. So start your own archive today: each month, save the page list and a screenshot in a simple table.

Our rule of thumb is simple. The report tells you where you appeared. It does not tell you what you earned. You close the second question with other sources.

How does the AI report differ from the classic Search Console performance report?

Both reports live in the same tool, but they answer different questions. The table below summarizes the difference. The classic columns are general knowledge, while the AI report scope comes from the announcement and the news sources above.

In short, the new report does not replace the old one. It complements it. You use the classic report for ranking and traffic, and the new one for AI visibility.

What does an AI impression really mean?

An impression means your page was offered to a user inside an AI feature. It does not guarantee that the user looked at the link or clicked it. So read an impression as "potential attention", not as "earned visit".

AI answers often give the solution directly. As a result, users may leave without visiting the source. We already know this behavior from zero-click searches. For details, see our article on zero-click searches.

Still, an impression is not worthless. When your brand name appears as a source in an AI answer, the user may remember you in the next search. You cannot measure this effect directly. However, you can watch branded searches for an indirect signal.

An AI impression can also work as a trust hint. If Google chooses your page for a summary, your content probably fits that question. Even so, this is an inference, and the report does not state it.

How do you read the numbers with a worked example?

The figures below are a pure example calculation, not data from a real client. The goal is to show which comparison to make when you read the report.

• A guide page gets 20,000 classic search impressions in a month.
• The same page gets 8,000 impressions inside AI features.
• Classic search brings 600 clicks, while the AI report gives no click number.

In this table, AI impressions equal 40 percent of classic impressions. That ratio says the page appears noticeably in AI features. However, without click data, you cannot draw a conclusion about traffic.

Instead, open the GA4 sessions from AI sources for that page. If sessions do not grow, you have impressions but weak referrals. Then you strengthen the title, the first paragraph and the call to action.

To adapt the calculation, open a simple table with three columns: classic impressions, AI impressions and the ratio. If the ratio stays stable, your visibility is steady. If it swings fast, look for the cause.

How do you open the report and run a first review?

The interface may change over time, so do not memorize menu names. The logic stays the same, though. Our team follows the steps below at first glance.

1. Select the right property and confirm that you have full permission.
2. Look for the AI report in the section where the performance reports live.
3. Set the date range to at least the last three months.
4. Switch to the page view and note the ten pages with the most impressions.
5. Check the country and device views and mark any clear differences.

Then find the same pages in the classic performance report. When you place both lists side by side, you see which content stands out in AI features. If you are new to the tool, read our guide on what Google Search Console is first.

Which questions should you ask the data first?

Looking at a report without a goal easily misleads you. So define your questions first. We usually start with these five:

• Which of my pages appear most in AI features?
• Do these pages share a content format?
• Which pages should appear but do not?
• Do impressions spread across countries the way I expect?
• Is there a clear gap between mobile and desktop?

Each question leads to an action. For example, if a page you expect to see has no impressions, rewrite it so it answers the question directly. Our guide on how to write content for AI Overviews helps here.

Writing the questions down also makes meetings easier. The team looks for the same answers, so the discussion does not drift. Collect the answers within a week and store them in one shared document.

How do you turn page level data into a content strategy?

Split your page list into three groups: pages that stand out in AI features, pages that never appear and pages in the middle. Each group needs a different move.

For standout pages, your goal is to keep the momentum. Keep them current, refresh their sources and support them with internal links. If a page appears in AI features, Google may see it as a reliable source for summaries. That is an assumption, because the report does not give the reason.

For pages that never appear, look for the cause first. Does the content really answer the question, or does the page have an indexing or technical problem? On the technical side, our article on AI crawlers and robots.txt offers a checklist.

Pages in the middle are the easiest wins. Write the heading as a question, put a direct answer of 40 to 60 words in the first paragraph and support it with a table. Then reopen the report after a few weeks.

Use your content inventory as well. List each page's type, publish date and last update. Then you can watch how AI impressions relate to content age on your own site.

What should you do when AI impressions rise but clicks fall?

This is the worry the report will raise most often. Impressions go up, clicks go down. First, do not panic, because the two numbers may not come from the same source. Then rule out the likely causes one by one.

• The AI answer may satisfy the user's question completely.
• Your page title may look weak next to the answer.
• Low brand awareness may keep users from clicking the source.
• Seasonality or a channel shift may affect clicks.

The fix is not a single move. Deeper information, original data and a clear brand voice carry users beyond the summary. For example, pages with a worked example, a template or a calculator give value that a summary cannot give.

You can also test titles. Rewrite the title so it states the value more clearly, and look at the classic click rate after a few weeks. Change only one thing at a time.

To measure the click side, follow the method in our article on AI traffic in GA4.

How do you combine the Search Console AI report with GA4 and other tools?

One report never shows the whole picture, and the Search Console AI report is no exception. So we suggest a three layer reading order for the Search Console AI report. Layer one is the AI report, which tells you about visibility. Next comes the classic performance report, which tells you about clicks. Finally, GA4 tells you about behavior and conversions.

Use the page address as the shared key. Then, for each page, write the AI impressions, the classic clicks and the GA4 conversions side by side. Keep the table to five or six rows, because the goal is direction, not detail.

The three layers also show which layer is weak. If visibility is high but clicks are low, the title and the value proposition are the problem. If clicks are high but conversions are low, the page content or your offer needs work.

Third party visibility tools can enrich the picture. For the technical side, our llms.txt generator helps. Note that such tools sample results, while Search Console shows Google's own records.

For our full measurement approach, read our guide on how to measure AI visibility.

Why should you think carefully about opting out of AI surfaces?

According to Search Engine Journal, site owners can exclude their content from these three AI surfaces with a property level setting. Google states that this choice is not used as a ranking signal elsewhere in Search. The setting also does not affect AI training, which separate controls govern.

Read the official announcement to verify this yourself. Our own view is balanced: for most businesses, turning the setting on makes little sense. Appearing in AI features has become a part of brand visibility.

Exceptions exist, though. A publisher with paid content, or a research site that protects original data, may consider it. Before such a decision, weigh your impression level in the report and your business model together.

In short, this setting is a strategy decision, not a simple switch. Also make it reversible: note the date and the reason. Then, if impressions drop months later, you find the cause fast.

Which mistakes should you avoid when you read the report?

A new report brings new misreadings. Here are the mistakes we see most often in the field.

• Treating impressions as traffic: An impression is not a click.
• Deciding on a short period: Data from the first weeks may not have settled.
• Expecting query analysis: The report had no query breakdown at publication.
• Claiming causation: A page may appear for reasons other than your last edit.
• Focusing on one page: Patterns come from page groups, not from single pages.

Another common mistake is comparing competitors. The report shows only your own property. You cannot know a rival's AI impressions, so avoid claims like "we appear more than they do".

All these mistakes share one root: treating incomplete data as complete. So add one question to every interpretation: what does this data not say?

How can a small or mid sized business use the report?

You can get value from the report even without a large content team. The key is to work with few pages. Pick one service page, one guide and one product or category page.

Here is an example calculation. Imagine a local service site with 30 pages. In the report, you see that only 4 pages have AI impressions. All 4 are how to guides, and the service pages have none.

You can draw three conclusions. First, guide content is your gateway to AI features. Second, your service pages may not answer a question. Third, internal links from guides to service pages become critical.

The sample is small in this example, so the results are not final. Even so, the pattern is enough to set a direction. Large sites apply the same logic to hundreds of pages, while small sites go page by page.

If you want help with this work, our AI SEO and GEO services start exactly here. You can still take the first step on your own.

How do publishers, online shops and service sites read the report differently?

The same report means different things for different site types. So build a reading frame that fits your business model.

• Publishers: Impressions may be high, so the real question is where clicks disappear. If ad revenue depends on page views, watch GA4 closely.
• Online shops: Check whether category and product pages earn impressions. If guides appear and product pages do not, add usage notes and comparisons to product content.
• Service and local sites: Watch the frequently asked questions next to the service pages. Users usually ask a how or a how much question before they buy.

For example, if a category page never appears in AI features, the content is often too thin. The page holds only a product list and no text that answers a question. In that case, add a short buying guide and a comparison table.

For our deeper approach to shops, see our e-commerce consulting page. The shared principle stays the same: read the report first, then shape the page as the answer to a question.

What does the report say about technical health?

When a page does not appear in AI features, the cause is not always content. Technical blocks can also reduce visibility. So run a short technical checklist for important pages that never appear.

1. Confirm with URL Inspection in Search Console that the page is indexed.
2. Make sure the page has no noindex or restrictive preview directive.
3. Check that your robots.txt does not block the directory.
4. Verify that the main content is readable without JavaScript.
5. Confirm that the canonical address points to the page itself.

These five steps may look boring. In the field, however, they often bring the fastest gain. No matter how well you write, visibility cannot come if Google cannot reach the page.

You can find Google's official guidance on the page AI features and your website. It explains how your content can take part in AI features.

How do you report the numbers to your team and management?

Telling management that AI impressions rose is not enough. That sentence raises doubt and invites the question: how many sales did it bring? So split your presentation into three parts.

Part one explains visibility. Show which pages appear in AI features and how they changed from the previous month. Part two states the limit honestly. The report gives no clicks, so you complete the commercial effect with other data. Part three lists the action: which five pages will you update this month?

This order protects you from hype and from needless pessimism. It also sets expectations early. We suggest writing this sentence on the first slide: this report gives visibility, not sales.

Our team uses the same structure in client reports. As a result, the report produces decisions instead of numbers. The first decisions are usually which content to update, which to merge and which to write new.

Where can you verify this news yourself?

Unverified news turns into wrong information that is hard to correct later. So you can check the facts in this article yourself. The first source is the official announcement on the Google Search Central blog.

Next, read the Search Engine Journal report, which describes the worldwide rollout and the scope of the report. Finally, open the Search Console help pages, where you find the current interface and scope.

In this article, we used only facts that we could confirm in those sources. The impression based structure, the three surfaces, the page, country and date dimensions and the opt-out setting fall into that group. The interpretations and steps, in contrast, are our field approach.

Separating sourced facts from opinion is essential. You do not have to agree with our interpretation. However, you can open the numbers in your own property and test them.

What is the 30 day plan our team suggests after the launch?

This plan is a field based starting point, not a guarantee. The goal is to turn the report into a habit instead of reading it once.

1. First week: Open the report, then write the ten pages with the most AI impressions next to the ten pages with the most traffic.
2. Second week: Choose five important pages that never appear, and rewrite the first paragraph of each as a direct answer.
3. In week three: Add a table, a short list and frequently asked questions to those pages, and strengthen the internal links.
4. At month end: Reopen the report, note the change and choose three more pages for next month.

The plan looks small. Applied regularly, however, it moves you from guessing to measuring. For the technical base behind the plan, see our SEO consulting page.

Does this news change your SEO and GEO strategy?

The short answer: it changes how you measure, not the foundation. Good content, technical soundness and reliable sources still matter in the same order. The difference is that you can now track AI visibility in Google's own report.

If you wonder how SEO and GEO relate, our article on the differences between SEO, GEO and AEO is a good start. As you will see there, the three fields complement each other instead of competing.

In conclusion, we suggest this: open the report this week, note the first ten pages and look again at the end of the month. Even if the numbers are small, taking your first measurement today puts you a step ahead. Most sites have not opened the report yet.

When a new feature appears, our team checks it in an official source and shares only what it can confirm. You can follow the same discipline. Trust your own data, not claims without a source.