Bing Webmaster Tools Now Shows Which AI Queries Surface Your Brand — Here’s What the Data Is Missing

For years, AI search was a black box. You knew your content was going into the systems somewhere. You had no way to verify it was coming back out. That changed on February 9, 2026, when Microsoft launched the AI Performance dashboard inside Bing Webmaster Tools — the first time publishers could see real citation data: which AI queries surface their pages, how often, and which URLs are being pulled into AI-generated answers. It is genuinely the most useful new diagnostic tool B2B marketers have gotten this year.

And it has a blind spot that will mislead every team that stops at the dashboard and declares victory.

What the Bing AI Performance Report Actually Shows

The AI Performance report — available in public preview through Bing Webmaster Tools — gives marketers three things they have never had before. First, total AI citation volume: how many times your pages are referenced inside Copilot and Bing AI-generated answers. Second, the grounding queries behind those citations. Third, on March 23, 2026, Microsoft added Grounding Query–Page Mapping, letting you click any grounding query to see which pages it cited — or click any page to see which queries triggered it.

This is real intelligence. If your “how it works” page is generating citations while your pricing page generates none, that tells you something actionable about what AI retrieves and why. If your citation volume has grown 40% over two months, you can correlate that against content changes. For the first time, GEO measurement has a native data source from one of the major AI platforms.

One analysis found a site with minimal organic search traffic that had generated more than 33,000 AI citations. Citation volume in Bing WMT is real — and tracking it over time gives you a genuine signal that your content is being retrieved.

What Is a Grounding Query — and Why Marketers Are Misreading Them

That distinction matters more than most people realize. When a user asks Copilot “what’s the best B2B analytics platform for a 50-person team,” the system doesn’t retrieve your page by sending that exact sentence into an index. It generates one or more grounding queries — internal search phrases designed to pull structured, relevant content — and your page gets cited if it ranks for one of those internal queries.

Here’s the problem: a significant portion of the grounding queries appearing in Bing WMT are not buyer searches at all. Hive Digital’s analysis of their own top GEO-optimized blog post found that many grounding queries for that page were automated tool-driven probes — queries generated by GEO monitoring platforms, not actual humans researching a purchase decision. Marketers see high citation volume attached to impressive-looking queries and report upward: “We have strong AI visibility.” What they actually have is strong retrieval from automated systems scanning the index.

Citation volume alone cannot tell you whether a real buyer ever saw your brand in an AI answer.

What Bing Cannot Tell You About Your Brand

Bing WMT tells you when AI retrieves your content. It tells you nothing about what AI says once it gets there. A brand can generate 5,000 citations in a month and have its product described incorrectly in every single one. Bing WMT would report that as a success.

The gap is not theoretical. Across 29 major LLMs — including GPT-5, Gemini, and Claude — hallucination rates range from 15% to 52%. One analysis found that 87% of a brand’s AI citations referenced a single page — a state employment law explainer with nothing to do with their product positioning. High citation volume, completely misaligned narrative.

An Ahrefs researcher made this concrete in December 2025. They built a fake brand called “Xarumei” and seeded false information across the web. Within weeks, Perplexity, Grok, Copilot, Gemini, and AI Mode were confidently repeating fabricated founders, cities, and product lines as verified fact — Gemini trusted a Medium article over the official FAQ. Real brands with strong Bing WMT citation numbers can be just as badly misrepresented. There is no signal in the dashboard to indicate a problem.

The dashboard also only covers Bing and Copilot. ChatGPT, Gemini, Perplexity, and Claude are completely dark. If 46% of Americans now use AI for information-seeking, the majority of those queries aren’t happening in Bing’s ecosystem.

How to Use the Bing Webmaster Tools AI Performance Report the Right Way

Bing Webmaster Tools gives you retrieval intelligence. Use it for exactly that: understand which content topics AI is pulling, which pages are invisible, and how citation trends move when you publish new content. Combine the page-level citation mapping with broader GEO measurement practices to build a baseline of what’s being retrieved.

What retrieval data cannot tell you is whether the narrative inside those retrieved answers is accurate, positive, or even current. If you want to check your brand in AI answers beyond retrieval counts — to see what the system actually says — that requires a different layer entirely. A brand with strong Bing WMT numbers and a broken AI narrative is in a worse position than a brand with modest citation volume and a crisp, accurate story. The citation count gets you into the conversation. The narrative determines what happens once you’re there.

The right workflow: run Bing WMT for retrieval signals on a monthly cadence. Then use Shensuo to see what story AI tells buyers across ChatGPT, Gemini, Perplexity, and Copilot — not just whether you showed up, but what it actually said.

Bing Webmaster Tools is a genuine step forward. It just isn’t the whole picture.

Steven Breslin is a contributor to Shensuo’s AInsider blog covering AI brand visibility, GEO measurement, and the operational realities of marketing in an AI-first search environment. Reach him at steve@shensuo.ai.