Google Search Console AI Performance Report: Guide

Google Search Console's Generative AI performance report is now available worldwide. It shows how often a site's URLs appear in AI Overviews and AI Mode, with breakdowns by page, country, device, and date. Use it as a first-party visibility baseline, then connect page-level trends to clicks and conversions elsewhere.

What changed in Google Search Console on August 31, 2026?

Google completed the worldwide rollout of its Generative AI performance reports on August 31, 2026. The reports give eligible Search Console properties dedicated views of impressions from generative AI features in Search and Discover. Google originally announced the reports in June and added the global-rollout note to both its Search Central announcement and publisher update.

For Search, the dedicated report currently covers AI Overviews and AI Mode. Google says experiments in Search Labs are excluded. Discover has a separate Generative AI performance report, so Search and Discover should not be combined casually when reporting trends.

This is a meaningful measurement change because publishers no longer have to infer all Google AI visibility from ordinary search performance or third-party prompt samples. However, the report is not a complete AI analytics suite. It measures exposure inside supported Google experiences, not universal visibility across ChatGPT, Copilot, Claude, Gemini, and Perplexity.

What does the Generative AI performance report measure?

The report's core metric is impressions: the number of times links to a site were shown to a user in a supported generative AI feature on Google Search. It then lets you segment that exposure by page, country, date, and device.

Dimension	Question it answers	Important detail
Impressions	How often did the site appear in supported Google AI features?	Exposure is not a click or visit
Pages	Which canonical URLs appeared?	Data is assigned mainly to the canonical URL after redirects
Countries	Where did the search originate?	It identifies demand geography, not user identity
Devices	Was the impression on desktop, tablet, or mobile?	Available for the Search report
Dates	When did visibility rise or fall?	Dates use Pacific Time and recent data may be preliminary

Google's report documentation explains an aggregation detail that can otherwise distort analysis. In the chart, data is aggregated by property: if two results from the same site appear in one generative AI search feature, they count as one site impression. When a URL filter is applied, the chart is aggregated by URL. The page table, meanwhile, is aggregated by page.

That difference means chart totals and the sum of page rows can disagree without either being wrong. Search Console's usual row limits and reporting constraints also apply, and exported ~ or unavailable values become zeros in the downloaded file.

What does the report not show?

The dedicated report currently documents impressions, pages, countries, devices, and dates. It does not list clicks, CTR, average position, original prompts, answer position, or conversion value as dedicated Generative AI metrics.

This creates four important limits:

1. An impression does not prove that someone noticed the link.
2. A high-impression page does not automatically have a high click-through rate.
3. The report does not reveal why Google's system selected one passage or source.
4. It cannot measure visibility in non-Google answer engines.

Google says Generative AI data remains included in the overall Search performance report, while the new view isolates AI impressions. That does not make every standard Search click attributable to AI Mode or an AI Overview. Avoid calculating an "AI CTR" by dividing ordinary Google Search clicks by dedicated AI impressions unless Google exposes matching click data for the same scope.

The safest language for a dashboard is: Google AI impressions show supported-feature exposure; analytics and product events show downstream engagement.

How do you find and configure the report?

Open the verified Search Console property, then look for the Generative AI performance report under Search results. If the report is missing, Google lists two likely causes: the property has not received enough qualifying impressions, or the site has been excluded from Search generative AI features.

Before interpreting a zero, check this sequence:

• Confirm that the canonical domain property is verified.
• Make sure important pages are indexable and eligible to show a search snippet.
• Check the Search Console control for inclusion in generative AI features.
• Remove unintended noindex, snippet restrictions, robots blocks, or CDN bot challenges.
• Use canonical URLs consistently in internal links and the XML sitemap.
• Wait for complete data rather than treating a preliminary day as final.

Google's official AI optimization guide says eligibility still depends on ordinary Search foundations: a page must be indexed, eligible for a snippet, crawlable, and included in Search generative AI features. The guide also says that indexing and serving are never guaranteed.

Run the free AI visibility checker alongside Search Console when you need to diagnose crawl access, Answer First structure, schema, evidence, and trust signals. The checker assesses page readiness; the Search Console report observes actual exposure within supported Google AI surfaces. They answer different questions.

How should you analyze AI impressions?

Start with complete weekly windows. Export the current seven days and the previous seven days, then compare the same page, country, and device segments. Do not compare a partial Monday with a complete Sunday or mix property-level chart totals with page-level table totals.

Use this five-step workflow:

1. Establish the baseline. Record total AI impressions and the number of pages receiving impressions.
2. Find concentration. Identify the pages responsible for most visibility instead of focusing only on the sitewide total.
3. Segment demand. Compare countries and devices to see whether the audience matches your market and page experience.
4. Annotate changes. Mark publication, substantial edits, indexing events, migrations, and crawler-control changes.
5. Connect outcomes. Compare the exposed pages with organic landing sessions, engaged visits, checker completions, registrations, and revenue events.

One day of growth is an observation, not proof of causality. Recent data can be preliminary, and Google does not disclose the query or selection logic in this report. A stronger decision uses several complete weeks and controls for seasonality, publishing cadence, and major site changes.

Recent randomized evidence also shows why exposure and traffic must remain separate: the AI search publisher traffic study found materially different click behavior under forced AI Mode use, without turning an impression count into an unsupported AI CTR.

For the broader framework, the GEO hub explains how technical eligibility, entity clarity, evidence, and live answer-engine monitoring fit together.

How do you turn AI impressions into clicks?

Prioritize pages with sustained AI impressions and a commercially useful next step. Those URLs have already crossed an exposure threshold, so improving the reason to visit is often more efficient than rewriting an unseen page.

Review each exposed page in context:

• Does the title promise a deeper answer than the short AI summary can provide?
• Is there original data, a template, calculator, comparison, or downloadable asset worth visiting?
• Does the introduction confirm the user's intent immediately?
• Are examples, limitations, methodology, and sources visible rather than buried?
• Is the primary action specific to the page instead of a generic "learn more" button?
• Does the mobile experience make the next step obvious without covering the content?

For example, a page about measuring AI visibility can offer a live domain check, a transparent scoring methodology, and an exportable report. The AI answer may satisfy the definition, while the page earns the visit by providing the tool and evidence needed to act.

Do not remove the concise passage that may be earning the impression. Preserve the useful answer and strengthen what surrounds it: original proof, current examples, internal links, and a relevant action. Google's guidance emphasizes unique, non-commodity, people-first content rather than special "AI-only" rewrites.

The guides to Google AI Mode optimization and Google AI Overviews optimization cover the underlying content and technical foundations in more detail.

How should you combine Google, Bing, and analytics data?

Use each source for the layer it can actually observe.

Source	Best evidence	Do not claim
Google Generative AI report	AI Overview and AI Mode impressions by page, country, device, and date	Universal AI visibility or isolated AI clicks
Bing AI Performance	Citations, cited pages, and sampled grounding queries across supported Microsoft AI experiences	Google or ChatGPT coverage
Plausible or GA4	Human sessions, landing pages, engagement, and conversion events	Every untagged organic visit came from an AI answer
Server-side product events	Checker completions, verified leads, registrations, sites, and audits	Marketing attribution without a reliable source signal

The Bing AI Performance guide explains why citations are not clicks. The same measurement discipline applies here: exposure, engagement, and conversion are separate stages.

Create a weekly page-level join outside Search Console. For each important URL, store Google AI impressions, Bing citations, organic landing sessions, and conversions. Use canonical URLs as the join key, document time zones, and retain raw exports. This will not create perfect attribution, but it will reveal which exposed pages produce business outcomes.

What is a practical 30-day measurement plan?

During week one, export the baseline and identify five pages with meaningful AI impressions. Confirm their indexability, canonical tags, mobile rendering, and analytics events. Do not change every page at once.

During week two, improve one conversion-relevant element on two or three exposed pages: a stronger title promise, a useful template, clearer methodology, or a contextual call to action. Keep the answer and topic stable so the test remains interpretable.

During week three, add relevant internal links from established pages and verify that the updated URLs remain present in the sitemap. Record the exact deployment date.

During week four, compare complete post-change and pre-change windows. Review AI impressions, ordinary Search clicks, landing sessions, engagement, and conversions separately. Keep improvements that help users even when the AI-impression change is inconclusive.

For referral measurement outside Google, use the guide to tracking ChatGPT traffic in GA4 and Plausible. Tagged AI referrals are useful evidence, but they should not be merged with Google's new first-party impression metric.

FAQ

Is the Google Search Console AI Performance report available worldwide?

Yes. Google states that the Generative AI performance insights were rolled out to all websites worldwide on August 31, 2026. A property may still lack a visible report if it has insufficient qualifying impressions or is excluded from generative AI features.

Does the report show AI Overview clicks?

The current dedicated-report documentation lists impressions and dimensions for pages, countries, dates, and devices. It does not list dedicated clicks or CTR. Do not divide general Search clicks by AI impressions and present the result as an official AI Overview CTR.

Does it reveal the queries that triggered an AI result?

No query dimension is documented for the dedicated Generative AI performance report. Use page themes and first-party conversion data for decisions, but do not infer an exact user prompt from the page alone.

Why do the chart and page-table totals differ?

The chart is normally aggregated by property, while the Pages table is aggregated by URL. If several URLs from one property appear in the same AI feature, the property chart can count one impression while multiple page rows receive impressions.

Should I create special content only for Google AI features?

Google says no special AI markup or rewrite format is required. It recommends established SEO foundations and unique, helpful, reliable content. Improve clarity and usefulness for people rather than generating thin pages for every possible fan-out query.

Sources

• Introducing Search Generative AI performance reports in Search Console - Google Search Central, updated August 31, 2026
• Generative AI performance report (Search) - Google Search Console Help
• New opportunities, control and insights for website owners - Google, updated August 31, 2026
• Google's guide to optimizing for generative AI features - Google Search Central