Bing AI Performance report: what Copilot citation data shows
What the Bing Webmaster Tools AI Performance report shows about citations in Copilot and Bing AI answers, what each metric means, its limits and how to use it.
By Crawlito · 6 min read
In short
- The AI Performance report in Bing Webmaster Tools, launched as a public preview on 10 February 2026, counts how often your pages are cited in Microsoft Copilot, AI summaries in Bing and select partner integrations.
- Its metrics are Total Citations, Average Cited Pages, grounding queries, page-level citation activity and trends over time.
- Grounding queries are a sample of the phrases the AI used to retrieve content, not the questions users typed.
- Microsoft says the metrics do not indicate ranking, authority, placement or page importance, and none of them measure clicks.
- The report covers Microsoft surfaces only, so it says nothing about ChatGPT, Perplexity, Claude or Google AI Overviews.
On this page
The AI Performance report in Bing Webmaster Tools shows how often your site's pages are cited as sources in Microsoft's AI answers: Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations. It counts citations, lists which pages were cited, and shows a sample of the "grounding queries" the AI used to find your content. It does not show clicks, rankings or anything about non-Microsoft assistants.
Microsoft introduced the report as a public preview on 10 February 2026 and added further preview features in June. This guide explains each metric in Microsoft's own terms, what it can and cannot tell you, and a practical way to use it alongside your other data.
What the AI Performance report covers
The report is part of Bing Webmaster Tools and is available for sites you have verified there. If you already use Google Search Console, Bing Webmaster Tools can import verified sites from it, which is the quickest way to get started.
Microsoft describes the scope as citations in:
- Microsoft Copilot
- AI-generated summaries in Bing
- Select partner integrations
Microsoft does not list which partners are included, and the data is aggregated across these surfaces. There is no filter to see Copilot on its own. The report does not include ChatGPT, Perplexity, Claude, Gemini or Google AI Overviews, even where those products use web search of their own.
The metrics, as Microsoft defines them
| Metric | Microsoft's definition | What it does not tell you |
|---|---|---|
| Total Citations | The total number of citations displayed as sources in AI-generated answers during the selected period | Placement or presentation within an answer |
| Average Cited Pages | The average number of unique pages from your site shown as sources per day | Ranking, authority or a page's role in an answer |
| Grounding queries | "The key phrases the AI used when retrieving content that was referenced in AI-generated answers" | What users actually typed; it is a sample |
| Page-level citation activity | Citation counts for specific URLs over the selected range | Page importance, ranking or placement |
| Visibility trends | How citation activity changes over time across supported AI experiences | Why it changed |
The caveats in the right-hand column come from Microsoft's own announcement. It is unusually explicit that these are observations of citation activity, not quality scores.
Grounding queries are not user prompts
This is the metric most often misread. When Copilot answers a question, it searches for supporting content. The grounding query is the phrase used for that retrieval step. A user might ask a long conversational question; the grounding query might be a short phrase extracted from it. Microsoft says the list is a sample of overall citation activity and that it will keep refining the metric.
Read grounding queries as "the topics Microsoft's systems associated with my pages when they cited them", not as keyword research.
The June 2026 preview features
On 16 June 2026 Microsoft announced four more preview features, rolling out globally:
- Intents groups grounding queries into broad categories such as Informational, Commercial, Navigational, Learn and Solve, Research, Creation and Local.
- Topics clusters related grounding queries into themes, so you can see citation activity by subject area rather than by phrase.
- Citation Share shows, for a given grounding query, the percentage of all citations shown that went to your site. Microsoft calls it "an observational metric" and says it does not expose competitor domains, does not represent traffic share and is not a quality score.
- Compare overlays a previous period, such as the prior 30 days, on the current one.
Microsoft notes that Intents and Topics rely on classification systems that are still evolving, so labels may be broad, especially for niche subjects. It also says Compare shows correlation over time, not cause: citation patterns shift with user behaviour, model changes, freshness and partner refresh cycles.
What the report cannot tell you
Knowing the limits stops you drawing conclusions the data does not support.
- No clicks or traffic. A citation is a link shown as a source. The report does not say how many people clicked it. For visits, use your analytics referral data; our guide to AI citations vs referral traffic explains how to report the two without mixing them.
- No ranking. Microsoft states the metrics do not indicate ranking or authority.
- No other assistants. A rise in Bing citations says nothing about ChatGPT or Perplexity.
- No cause. A change in citations can come from your content, from competitors, or from changes in Microsoft's systems. The report shows that something changed, not why.
- Preview status. Definitions, coverage and features may change. Note the date whenever you record a figure.
How to use the report in practice
A simple monthly routine gets most of the value:
- Record the headline numbers. Note Total Citations and Average Cited Pages for the month, with the date range, in the same sheet as your other search metrics.
- Review cited pages. Sort page-level activity and look at the top cited URLs. Are they the pages you would want cited for those topics? Pages you did not expect can reveal content that answers questions well, or old pages that need updating.
- Read grounding queries by page. For your most cited pages, check that the grounding queries match what the page is about. A mismatch can mean the page is cited for a side point; it may be worth a clearer section on that topic.
- Look for absent pages. Important pages that never appear are worth a technical check: can Bingbot fetch them, are they in the Bing index, do they return a 200 with the content in the HTML? Our guide to Bingbot and AI search covers how to confirm access.
- Cross-check crawling. Compare cited pages with what your server logs show Bingbot fetching. Pages Bingbot rarely visits are less likely to have fresh content in Microsoft's index. See measuring AI crawler traffic.
- Use Compare carefully. After a content change, compare periods, but treat a shift as a hint, not proof.
If your site shows no data
A site with no citations in the report may simply not be cited in the period. Before drawing that conclusion, check the basics:
curl -s -o /dev/null -w "%{http_code}\n" \
-A "Mozilla/5.0 (compatible; bingbot/2.0; +http://www.bing.com/bingbot.htm)" \
https://www.example.com/
curl -s https://www.example.com/robots.txt | grep -i -A3 "bingbot"A non-200 status, a Disallow rule for Bingbot, or a firewall that challenges it will keep pages out of the Bing index and therefore out of Copilot answers. Microsoft's announcement notes that Bing respects content owner preferences expressed in robots.txt and other supported controls, so a block you set is a block it follows.
How it fits with your other AI search data
The AI Performance report is useful because it comes from the operator itself, but it is one view of one ecosystem. A fuller picture combines:
| Source | What it shows |
|---|---|
| Bing AI Performance report | Citations in Microsoft AI surfaces |
| Google Search Console | Google Search performance; AI features are counted within the overall Web search type rather than separately |
| Analytics referral data | Visits from assistants that pass a referrer (tracking AI traffic in GA4) |
| Server or CDN logs | Which AI crawlers fetch which pages |
| Manual sampling | How specific assistants answer specific questions on a given day |
None of these alone tells you how an assistant chooses its sources. Together they separate "can they reach and read my pages" from "do they cite them" from "do people click through".
Frequently asked questions
Is the Bing AI Performance report free?
Yes. It is part of Bing Webmaster Tools, which is free, and available for verified sites. It was released as a public preview, so features may change.
Does the report include ChatGPT citations?
No. Microsoft describes it as covering Copilot, AI-generated summaries in Bing and select partner integrations. Citations in ChatGPT, Perplexity or Google products are not included.
What is a grounding query?
Microsoft defines it as a key phrase the AI used when retrieving content that was then referenced in an answer. It is a retrieval phrase and a sample, not the full question a user typed.
Can more citations be read as better rankings?
No. Microsoft says the metrics do not indicate ranking, authority or placement. They count how often pages were shown as sources.
Make sure Microsoft can read the pages first
Citation data only helps once Bingbot can reach and read your pages. Run the free check to see whether a key page is accessible to search and AI crawlers, or book an audit to check access across the site.
Filed under Strategy and audits. Topics: bingbot, measurement, analytics, ai-search, strategy