Search Console's AI Report: What It Measures, What It Hides, and What to Add

What changed

Until recently, AI visibility was a black box. Whether a site appeared as a source in an AI Overview could only be checked through manual spot checks or third-party tools. Search Console counted those appearances but did not break them out.

Google announced the dedicated performance reports on June 3, 2026. An update to the official announcement confirms the worldwide rollout of these insights on August 31, 2026. Search and Discover have dedicated views: these describe visibility, not an end-to-end conversion journey (Google Search Central).

What the Generative AI report covers

The report sits as its own section under "Performance". It shows:

• Impressions in AI Overviews and AI Mode, plus generative AI features in Discover
• Pages: which URLs appeared in AI features
• Countries
• Devices (Search only, not Discover)
• Dates at hourly, daily, weekly or monthly granularity

One point matters for interpretation: this is not additional data. According to Google, it is already included in the overall performance report and simply shown as a separate segment (PPC Land). Adding AI impressions to total impressions double counts them.

Counting depends on the view: chart totals are aggregated by property, while page breakdowns are aggregated by URL. Table and chart totals can therefore differ. Use the same aggregation for month-to-month comparisons (Search Console Help).

If the report is missing from a property, Google says there may not be enough data for that site yet (Google Search Central).

New since September: text-based vs. multimodal

On September 24, 2026, Google split the "Web" search type in Search Console. "Text" covers classic text searches. "Multimodal" covers searches that start from an image: Google Lens, Circle to Search on Android, uploaded images and Chrome image search (Search Console Help; PPC Land). The filter also applies to the Generative AI report.

This matters for online shops and manufacturers. Someone who photographs a product and searches for it can end up in an AI answer too, and according to Google these interactions were not counted in any report before. The filter shows no queries, though. You see which pages appeared and how often, not what image was searched.

Bing AI Performance: citations across AI experiences

Microsoft moved first. Since February 2026, Bing Webmaster Tools has offered an "AI Performance" report as a public preview (Microsoft Bing). It shows:

• Citations per day and the number of pages cited
• Page statistics: which URLs were cited and how often
• Grounding queries: internal phrases used to retrieve sources across supported AI experiences

Grounding queries are internal phrases used to retrieve sources, not the original user prompts. Microsoft explicitly describes them as a sample of overall citation activity. They can inform topic analysis but cannot reconstruct an entire conversation. The report covers several supported Microsoft AI experiences and selected partner integrations, not just Copilot (Microsoft).

What is still missing

The new reports are progress. On their own, they are not enough to judge AI visibility, because three things are missing.

Clicks. The dedicated AI visibility reports described here do not provide complete click or conversion attribution. Google clicks remain part of the overall performance report, without a complete attribution to individual AI appearances. High AI impressions can go along with few visits. Whether an appearance leads to leads or sales only shows up in your own analytics.

Prompts. Google shows no queries for AI features. You cannot tell which questions triggered an appearance, or whether your site shows up for the questions that matter commercially or for side topics.

Competitors and context. Bing Citation Share now provides an aggregate comparison for a grounding query. It does not fully reveal which providers appeared in an individual answer or how they were described or recommended. Documented answer samples remain necessary.

On top of that, ChatGPT, the standalone Gemini app and Perplexity provide no comparable data to site owners so far.

The new control: switching off rarely makes sense

Under Settings > Search generative AI, property owners can manage inclusion of links and content in AI Overviews, AI Mode and generative Discover features. This includes answer grounding, but does not control model training or the rest of conventional Search (Search Console Help).

For publishers whose business model is the content itself, this can be a real trade-off. For B2B companies, manufacturers and online shops, opting out is usually counterproductive: if you are not in AI answers, you cannot be recommended there. Review the setting deliberately and document it so it is not switched on by accident.

Turning it into a monthly report

The official data works best as one of four layers, each answering a different question.

Layer	Source	Answers the question
1. Presence on Google	Search Console, Generative AI report	How often, and with which pages, does the domain appear in AI Overviews and AI Mode?
2. Presence across supported Microsoft AI experiences	Bing Webmaster Tools, AI Performance	How often are pages cited across supported AI experiences, and which grounding queries are visible?
3. Relevance and competition	Prompt sampling across several AI systems with an AI visibility scanner	Does the brand appear for business-critical questions, and who is named instead?
4. Impact	GA4: sessions from AI referrers, conversions	Does AI visibility lead to visits and enquiries?

A fixed monthly routine helps:

1. Record a baseline. Export Search Console and Bing figures for a defined period and document them as the starting point.
2. Cluster pages. Group URLs by topic or service to see which parts of the site appear in AI features and which do not.
3. Compare with prompt samples. Differences between platform data and samples call for investigation, not a conclusion about irrelevant topics. Check coverage, dates, language and measurement context first.
4. Review grounding queries. Terms from the Bing report reveal the sub-questions an AI system asks. Missing answers are content gaps.
5. Check impact. Compare AI referral sessions and their conversions in GA4.

What counts is the trend over several months, not a single monthly value. AI answers can vary by system, context and measurement time.

Limits of the method

Even all four layers combined do not produce an exact measurement. The official reports cover only Google and supported Microsoft AI experiences. Prompt samples are samples, and AI referrers are not always passed cleanly to GA4. Honest reporting names these limits and works with ranges and trends rather than seemingly precise numbers.

Measurement: What the numbers actually establish

Google impressions, Bing citations and mentions in prompt samples are different measures. Do not add them together or present them as combined reach. Microsoft describes AI Performance as aggregated data across Copilot, AI-generated summaries in Bing and selected partner integrations; it is not a complete report on every AI system (Microsoft).

Since June 2026, Microsoft has expanded the preview with Intents, Topics, Citation Share and Compare. Citation Share is a website’s share of citations for the same grounding query, not brand market share or position within an answer. Check availability in your own account (Microsoft).

For prompt sampling, record language, country, system or model, web search, account and personalization state, dates and repetitions. API responses are not automatically comparable with the consumer interface. A mention rate is the share of sampled answers containing the brand; source citations are counted separately. Keep the denominator and prompt set stable for comparisons.

AI referrers only capture identifiable visits. Missing referrers, consent gaps and cross-device journeys prevent complete attribution. More impressions or citations do not establish more revenue or prove that a particular intervention caused the change.

FAQ

Where is the AI report in Search Console?

Under "Performance" there is a dedicated "Generative AI" section. Since August 31, 2026, it is available to all sites worldwide. If it is missing, check impression volume and the site’s inclusion setting for generative search features.

Does Search Console show clicks from AI Overviews?

The dedicated AI report shows visibility dimensions, especially impressions. Clicks from AI features can be included in the overall performance report; that is not a separate click report for individual AI Overviews.

Are AI impressions on top of regular impressions?

No. They are already part of the overall performance report. The Generative AI report shows a segment of that data.

What does "multimodal" mean in Search Console?

Searches that start from an image, for example via Google Lens, Circle to Search or uploaded images. Since September 24, 2026, they can be analyzed separately from text-based searches.

What are grounding queries in Bing?

Internal search phrases used to retrieve sources for an answer in supported AI experiences. They differ from what users type but show how a request is split into sub-questions.

Are the official reports enough for GEO monitoring?

No. They show presence on Google and supported Microsoft AI experiences but no complete reconstruction of individual answers or revenue attribution. A proper assessment also needs prompt sampling across several AI systems and your own analytics.

Further reading

• Fact-checking AI content
• Choosing a GEO agency
• Measuring AI Visibility: Tool Comparison
• GEO Audit: How to Check Your AI Visibility in 60 Minutes
• Generative Engine Optimization: The Complete GEO Guide

Sources

• Google: Generative AI performance reports
• Google: Generative AI performance report (Search)
• Google: Multimodal reporting
• Google: Search generative AI control
• Microsoft: AI Performance
• Microsoft: Intents, Topics, Citation Share, Compare