What does Bing's AI Performance report show about Copilot citations?
Microsoft now reports how often Copilot and Bing's AI answers cite your pages. Here is what each metric measures and what it leaves out.
By Kuraib AliPublished Updated 9 min read
Bing's AI Performance report is a section of Bing Webmaster Tools that counts how often your pages are cited as sources in AI answers. It covers Copilot, the AI summaries in Bing search and some partner products that Microsoft does not name. Microsoft released it as a public preview on 10 February 2026. It reports citation counts, the pages cited and the "grounding queries" behind them. It does not report clicks, rankings or traffic.
Where the report came from
Microsoft announced AI Performance in a Bing Webmaster blog post dated 10 February 2026. Its authors are the product managers Krishna Madhavan, Meenaz Merchant, Fabrice Canel and Saral Nigam. The post calls the report "an early step toward Generative Engine Optimization (GEO) tooling in Bing Webmaster Tools."
On 16 June 2026 Microsoft added four features in preview: Intents, Topics, Citation Share and Compare. The June post says they are "now beginning to roll out in preview within Bing Webmaster Tools globally." Both releases launched as previews, so labels and definitions may still change.
What the AI Performance report measures
The February announcement lists five parts of the dashboard. The table below gives Microsoft's description of each one and the limit Microsoft attaches to it.
| Metric | What Microsoft says it shows | The limit Microsoft states |
|---|---|---|
| Total citations | How many times your content was shown as a source in AI-generated answers in the date range | It does not indicate placement or presentation within an answer |
| Average cited pages | How many different pages from your site are cited on an average day | It says nothing about a page's rank, authority or job inside an answer |
| Grounding queries | The phrases the AI searched with when it found the pages it cited | The data is a sample of overall citation activity |
| Page-level citation activity | Citation counts for each URL on your site | It counts how often a page is cited. It does not show importance, ranking or placement |
| Visibility trends over time | A timeline of citation activity across supported AI experiences | No limit stated |
A Microsoft Advertising post from 23 March 2026 added a link between two of these views, which maps grounding queries to pages. Selecting a grounding query shows the pages cited for it, and selecting a page shows its grounding queries. The same post warns that "one grounding query may be mapped to multiple pages and vice versa."
What a grounding query is
In my view, a grounding query is the clearest new idea in the report. Microsoft describes it as "the key phrases the AI used when retrieving content that was referenced in AI-generated answers." My reading is that it is the search the AI ran on the user's behalf, which can differ from the prompt the user typed.
In my view, this is the most useful part of the report for content planning. A grounding query tells you which retrieval phrase led to your page. My inference is that one question can lead to several retrieval phrases. So I treat the list as a picture of topics.
What the June 2026 update added
The June post adds four views on top of the original dashboard.
- Intents. Grounding queries are grouped into categories that Microsoft lists as "Informational, Commercial, Navigational, Learn and Solve, Research, Creation, Local, and more."
- Topics. Related grounding queries are grouped into themes. Microsoft's example says "solar panels", "solar energy efficiency" and "residential solar installation" may all map into one topic, Solar Energy.
- Citation Share. For one grounding query, this is your site's citations as a percentage of every citation shown for that query, from any site.
- Compare. This overlays an earlier period on the current one, such as the last 30 days against the 30 days before.
Microsoft attaches clear limits to these features. Intents and Topics are "powered by evolving AI/ML classification systems." The post adds that during the preview "some labels may still be broad", especially for specialised domains.
In my view, Citation Share comes with the strongest caveat. Microsoft calls it an observational metric, adding that it is "not a ranking system or a competitive scoreboard." It "does not expose competitor domains, represent traffic share, or assign quality scores to content." So a 40% Citation Share means your pages took 40% of the citations shown for that phrase. It says nothing about clicks.
What the report does not show
The gaps matter as much as the metrics. These limits come from Microsoft's own descriptions.
- No clicks or traffic. None of the metrics in the February or June posts is a click count. Citation Share, Microsoft says, does not "represent traffic share."
- No split by surface. Microsoft's posts describe one combined view across Copilot, Bing's AI summaries and partner products. None of the posts mentions a per-surface filter, and none names the partners.
- A sample of grounding queries. Microsoft says the grounding query data "represents a sample of overall citation activity."
- No placement data. Total citations do not show where in an answer your link appeared.
- No other AI platforms. Microsoft does not say which partners are included. So do not read the report as a measure of Google AI Overviews, Google AI Mode or ChatGPT.
The click gap means you need two sources of data. The AI Performance report shows whether Microsoft's systems cite you. Your analytics show whether those citations send visitors. The guide to tracking AI referral traffic in GA4 covers the second half.
What one site's data looked like
In my view, independent data on the report is still thin. One published example comes from Will Scott of Search Influence, posted on 12 February 2026. It covers the firm's own site, so it shows one site's experience.
Search Influence reported 91 days of data, from 12 November 2025 to 10 February 2026. In that period it recorded 19,717 citations across 86 pages. One page, a detailed comparison of tools, took 69% of all citations. The top four pages took 90%.
Search Influence also noted that citations "don't equal traffic", because the report showed no click-through data. My reading is that a heavy concentration on a few pages is worth checking for on your own site. It hints at which formats Microsoft's systems reach for. Nothing here suggests every site will see the same pattern.
How Microsoft says to use the data
Microsoft's February post suggests three uses for the report. Validate which pages are already cited. Identify content that appears often. Spot pages that are indexed but cited less often, where clarity, structure or completeness could improve.
The same post lists five content habits it links to citation:
- strengthen depth and expertise
- improve structure and clarity, using "Clear headings, tables, and FAQ sections"
- support claims with evidence
- keep content fresh and accurate
- keep text, images and video consistent about the same entities
These habits match the structure advice in Microsoft's longer guide on AI answers.
Microsoft's October 2025 guide, "Optimizing Your Content for Inclusion in AI Search Answers", explains why structure matters to Copilot. It says assistants "break content down, a process called parsing, into smaller, structured pieces that can be evaluated for authority and relevance." The same guide states that "there's no secret sauce that guarantees selection in AI answers."
IndexNow and freshness
Microsoft ties the report to IndexNow, a protocol that notifies participating search engines when a URL is added, updated or removed. The February post says IndexNow "helps ensure that AI systems reference the most current version of a page when generating answers."
In my view, this matters most for pages that change often, such as prices, schedules or policies. If an AI answer quotes an outdated version of a page, the first fix is to make sure Bing knows the page changed.
Crawler access still comes first
Microsoft's announcement includes one line on control: "Bing respects all content owner preferences expressed through robots.txt and other supported control mechanisms." My reading is that a page Bing is told not to crawl is unlikely to be cited. Robots.txt is one of several controls Microsoft mentions. The robots.txt and AI crawlers guide explains which bots control which uses.
A monthly routine for the report
The report is new, labelled preview and sampled in parts. A short monthly check suits that better than daily monitoring.
- Export the cited pages. Note which URLs carry most citations and whether that list changes month to month.
- Read the grounding queries for your top pages. Check that each page is cited for phrases it answers well. A page cited for an unrelated phrase may need a clearer focus.
- Look for indexed pages with few citations. Microsoft's post points to these pages as places to improve clarity, structure or completeness.
- Check Intents and Topics. Compare the themes Microsoft assigns with the topics you mean to own.
- Use Compare after each content update. Overlay the period before the change with the period after it.
- Match citations to visits. Put the report next to your analytics for the same pages and dates.
Treat any movement with care. Microsoft's June post says citation patterns "can shift due to changes in user behaviour, evolving models, freshness signals, partner refresh cycles, and broader changes across the web itself." A drop after an edit may not be caused by the edit.
For how other assistants choose sources, see how ChatGPT decides what to cite. None of this guarantees that a page will be cited. The report shows what Microsoft's systems did, and it leaves the decisions to you.
Related: AI search hub, tracking AI referral traffic in GA4, how ChatGPT decides what to cite, robots.txt and AI crawlers.
Frequently asked questions
Does the Bing AI Performance report show clicks from Copilot?
No. The metrics Microsoft lists are citation counts, cited pages, grounding queries, page-level citations, trends and, since June 2026, Intents, Topics, Citation Share and Compare. None of them is a click or traffic count. Clicks from AI assistants have to be measured in your own analytics.
What is a grounding query in Bing Webmaster Tools?
Microsoft describes grounding queries as the phrases the AI searched with when it found the content it cited. They are retrieval phrases, so they may differ from the words a person typed into Copilot. Microsoft also says the data shown is a sample of overall citation activity.
Which AI products does the report cover?
Microsoft's announcement says it covers Copilot, the AI summaries in Bing search and some partner products. Microsoft does not say which partners are included, so do not read it as a measure of ChatGPT, Gemini or Google AI Overviews.
Is Citation Share a ranking score?
No. Microsoft's June 2026 post calls Citation Share an observational metric. It says the metric does not reveal which other domains were cited. It does not stand for a share of traffic or score content quality. It is the percentage of citations for one grounding query that went to your site.
Can I get more Copilot citations by blocking or allowing crawlers?
Microsoft states that Bing respects content owner preferences set through robots.txt and other supported controls. My reading is that a page Bing cannot crawl is a page it cannot cite. Allowing the crawler makes citation possible, but nothing in Microsoft's documentation promises that a page will be cited.