Microsoft launched an AI Performance report inside Bing Webmaster Tools in public preview on February 9, 2026, four months before Google shipped anything comparable in Search Console. It measures how often a site's content is cited across Microsoft Copilot, AI generated summaries in Bing, and select partner integrations, and it surfaces data points no other public tool currently exposes.
Why this matters as the first mover, not just an early one
For years, a site owner could suspect their content was being pulled into Copilot's answers with no way to confirm it, measure it, or improve it. This report changed that first, before any competitor. SEO analyst Aleyda Solis publicly called it the first official AI search visibility tool of its kind. When Google's own version finally arrived in June, it launched narrower, impressions only, UK only at first, no citation level detail, no click data, a gap technical SEO commentators noted directly at the time.
What the original report actually measured
The February 2026 launch shipped four original metrics: Total Citations, the number of times content was shown as a source in an AI generated answer, Average Cited Pages, the average number of unique pages cited per day, a citation activity timeline, and page level citation data for individual URLs. Microsoft described the release plainly as an early step toward real Generative Engine Optimization tooling inside Bing Webmaster Tools, not a finished product.
What expanded in June, and what did not
On June 16, 2026, Microsoft added four genuinely new capabilities: Intents, Topics, Citation Share, and Compare, moving the report from simply showing where content got cited toward showing why, grouped by search intent and topic, and how a site's citation activity stacks up over time. One thing did not change: click data is still missing. Microsoft has stated it is not yet available even through the API, and remains on the roadmap rather than in the product. Citation Share itself is explicitly framed by Microsoft's own product team as observational, not a ranking system, a competitive scoreboard, or a quality score, and it does not expose competitor domains.
Grounding queries, the genuinely unique data point
The most useful metric in the report is also the least understood one. Grounding queries are not what the user actually typed. When Copilot needs to retrieve web content to answer a question, it internally reformulates that question into cleaner, more specific retrieval phrases, and those internal phrases, not the original prompt, are what the report surfaces. This is a genuine first, real insight into the search language an AI model generates on its own to go find a source, something Google's Search Console report still has no equivalent for at all.
A real scope limit worth knowing up front
This report only covers Microsoft's own ecosystem, Copilot, Bing's AI summaries, and select partner integrations. It says nothing about ChatGPT, Perplexity, or Google AI Overviews, an important boundary to set before anyone reads too much into the numbers or assumes Bing citation volume represents AI visibility generally.
What real usage looks like
One publisher tracking three months of data after verifying their site reported 647 unique grounding queries triggering Copilot to pull from their content, generating over 30,000 total grounding events across 173 pages, a genuinely substantial volume most site owners have no visibility into without this report.
How to actually access it
Verify site ownership in Bing Webmaster Tools at bing.com/webmasters, using an XML file, a meta tag, or DNS verification, then open AI Performance from the left navigation once verified. No separate signup or paid tier is required to see the report once a site is verified.