On June 3, 2026, Google announced a Search Console report site owners had been waiting for: a dedicated view showing how often a site's URLs appeared in Google's generative AI features. The Search report covers visibility in experiences such as AI Overviews and AI Mode, while a separate Discover report covers generative features in Google Discover.
The announcement matters, but the report is easy to overstate. At launch it measures visibility, not complete business performance: it shows impressions, pages, countries, devices and change over time. The dedicated report does not separately provide AI clicks, CTR, queries or conversions. This guide keeps documented facts separate from practical interpretation.
In one sentence: the AI report answers “Did my content appear in Google's generative answers?” It does not yet fully answer “How many leads and how much revenue did that visibility produce?”
What changed in Search Console in June 2026?
Previously, traffic from AI Overviews and AI Mode was included in the standard Search Console Performance report under the “Web” search type, mixed with other organic data. That made it impossible to isolate how much visibility came from generative AI surfaces. The Generative AI performance report adds a dedicated view for AI impressions while those signals remain part of overall Search performance.
Google initially rolled the feature out to a subset of sites to test it and gather feedback. If you cannot see the report in July 2026, Google documents three possible reasons: your property is not yet included in the rollout, the site has not received enough AI impressions, or you excluded the site from Search generative AI features.
Exactly what data does the AI report show?
According to Search Console Help, the core metric in the Search Generative AI report is impressions. An impression is recorded when a link to your site is shown to a user in a supported generative AI feature. You can break the data down across several dimensions:
- Pages: see which canonical URLs appeared in AI features and which earned the highest or lowest impression counts.
- Countries: understand where the search that triggered your visibility originated.
- Devices: segment desktop, mobile and tablet visibility in the Search report.
- Dates: inspect trends at hourly, daily, weekly or monthly granularity.
- Date range: compare a content update or campaign with a matching period before the change.
Page data is generally assigned to the canonical URL. This matters on multilingual sites and sites with duplicate variants: incorrect canonical or hreflang implementation can make data appear under a different URL than the one you expect.
What does the dedicated AI report not show?
Four commercially important metrics are absent from the dedicated report at launch: clicks, click-through rate, search queries and average position. This does not mean Google ignores every click from an AI experience. Clicks on external links in AI Overviews and AI Mode can still be included in aggregate “Web” performance data; the dedicated AI view simply does not isolate them as a separate click metric.
- You cannot see the exact query or prompt variation that produced an appearance.
- There is no separate AI click or AI CTR metric.
- You cannot directly connect an AI impression to a form submission or purchase.
- The report does not measure citations in ChatGPT, Perplexity, Claude or any non-Google system.
- An impression alone does not prove that a user read or trusted the cited link.
“10,000 AI impressions” does not mean 10,000 visitors. This is an upper-funnel visibility metric; sessions, leads and revenue still need to be evaluated in your analytics and business systems.
Search and Discover are separate reports with different logic
Google documents separate reports for generative features in Search and Discover. Search data is tied to an active search and can be segmented by device. In Discover, a link must actually be scrolled into view to receive an impression; the same result is counted only once per session if a user scrolls away and back.
Do not automatically combine both reports into one “AI ranking” number. They represent different user contexts: Search captures active information seeking, while Discover captures recommended content consumption. For a service business, Search landing pages usually carry more immediate commercial value; for a publisher, Discover may be equally important.
How to interpret the data: five useful KPIs
Google does not prescribe a business KPI framework for the report. The following five measures are a practical analysis model, not official ranking factors.
- AI impression trend: compare at least 28 days with the previous matching period instead of overreacting to daily volatility.
- Number of AI-visible pages: learn whether visibility depends on one article or spans several service and expertise pages.
- Commercial-page share: separate blog, service, portfolio and contact URLs, then monitor which group is growing.
- Market and device fit: confirm that visibility appears in countries you serve and on devices where the experience performs well.
- Downstream outcome: review organic landing pages, engagement and conversions in Analytics, treating the relationship as correlation until direct AI attribution exists.
A precise 30-day measurement workflow
- Day 1 — Baseline: export the report and record the date range, total impressions and visible URLs.
- Day 2 — Segmentation: label URLs as blog, service, portfolio, product or other.
- Days 3–5 — Quality review: audit winning and losing pages for indexability, canonical setup, internal links, accuracy and freshness.
- Weeks 1–2 — One controlled change: update a clearly defined page group instead of changing everything at once.
- Weeks 3–4 — Comparison: compare equal periods and record seasonality, campaigns and technical changes.
- Day 30 — Business check: review organic sessions and conversions for the affected landing pages alongside visibility.
For smaller sites, weekly or monthly trends will often be more reliable than daily data. If one URL suddenly earns many impressions, first inspect topic freshness and country distribution; do not immediately rewrite every related article.
What should you optimise according to Google's official guidance?
Google's 2026 guidance is explicit: AI Overviews and AI Mode are grounded in core Search ranking and quality systems and retrieve information from the Search index. There is no separate technical magic trick for “AI SEO”. A page can be eligible when it is indexed, eligible for a normal Search snippet, technically accessible and genuinely useful.
- Create non-commodity content informed by first-hand expertise instead of merely summarising other articles.
- Keep important information crawlable and indexable in HTML with clear internal links.
- Use accurate titles, understandable sections and high-quality relevant images or video where they help the reader.
- Provide a fast, stable and usable mobile experience.
- Keep Google Business Profile details current for local companies; maintain Merchant Center and product data for ecommerce.
- Use supported structured data for eligible rich results and keep it consistent with visible content—there is no special AI schema.
Three popular AI SEO myths Google debunks
The official guide directly addresses tactics that grew into an industry between 2024 and 2026. For Google Search, an llms.txt file does not improve generative search visibility; Google ignores it. Artificial “chunking”, rewriting copy specifically for AI systems and special schema.org markup are not required either.
- “Google AI cannot understand my site without llms.txt.” — False; Google Search does not use it as a visibility or ranking advantage.
- “Every paragraph must be split into tiny AI chunks.” — Not a requirement; length and structure should fit the reader's task.
- “There is a secret GEO schema.” — There is not; structured data can still support conventional Search understanding and rich-result eligibility.
The most common measurement mistakes
- Calling impressions visitors and calculating return on investment from that number.
- Confusing the dedicated AI report with the standard Web Performance report.
- Declaring a trend from one week of low-volume data.
- Attributing every increase in organic traffic to AI Overviews.
- Estimating ChatGPT or Perplexity visibility from Google Search Console.
- Growing blog impressions while leaving service pages and the conversion journey weak.
Frequently asked questions
Why can I not see the AI report in Search Console?
The report initially rolled out to a subset of sites. Google Help also says it may be absent when a site has not earned enough AI impressions or has been excluded from Search generative AI features. Check the correct property and its AI inclusion setting; a missing menu item is not automatically a technical fault.
Can I see the query or prompt that triggered my appearance?
Not in the launch version of the dedicated Generative AI report. You can analyse pages, countries, devices and dates, but there is no query dimension.
Does Search Console measure ChatGPT and Perplexity citations?
No. The reports cover Google's own generative features in Search and Discover. Other platforms require their own data, server logs, analytics or carefully evaluated third-party monitoring.
Do I need llms.txt or special AI schema to appear?
Google says no. Google Search ignores llms.txt and there is no special generative AI schema. Conventional SEO, indexability, original content, page experience and correct use of supported structured data remain the foundation.
Summary: what should you do now?
If the report is available, create a dated baseline export, group visible pages by business role and evaluate trends in 28-day windows. Pair visibility with organic landing-page, engagement and conversion data. If you do not have the report yet, avoid chasing “GEO hacks”: improve indexability, originality, internal linking, images and user experience.
Primary sources
Search Console Help: Generative AI performance report — Search
Search Console Help: Generative AI performance report — Discover
Google Search Central: Optimizing your website for generative AI features on Google Search