Bing Webmaster Tools AI Performance Report: How to Analyze and Optimize Your Copilot Citations - AI Search Visibility Blog | Insights and Data | OtterlyAI

Microsoft just launched its AI Performance report in Bing Webmaster Tools – giving website owners the first official look at how Copilot uses their content. We analyzed our own data (as well as other demo data) from day one with OtterlyAI and Bing Webmaster Tools. Here’s what we found, what it means, and how to combine it with OtterlyAI to build a complete Copilot optimization strategy.

What Is the Bing Webmaster Tools AI Performance Report?

On February 11, 2026, Microsoft officially released the AI Performance report inside Bing Webmaster Tools. This is a significant moment for anyone working on AI search visibility: for the first time, a major AI search provider is offering direct performance data to website owners.

The AI Performance report shows how Microsoft Copilot interacts with your content. It introduces two key metrics that didn’t exist before in any public-facing tool:

• Grounding Queries are the search phrases Copilot generates internally when it needs to retrieve web content to answer a user’s question. These are not the user’s original prompts. They’re reformulated queries that Copilot creates behind the scenes to find the most relevant sources. Think of them as the “real search queries” that drive AI-powered answers.

• Citations in this context refer to “website citations”. How many times Copilot used a specific page from your website to inform its response.

The Headline Numbers: What We Found in OtterlyAI’s Data

We verified our own domain on Bing Webmaster Tools as soon as the AI Performance report went live. Here’s what three months of data (November 2025 through February 2026) revealed:

647 unique grounding queries triggered Copilot to use content from otterly.ai, generating 30,398 total grounding events across 173 pages on our site.

To put that in perspective: Copilot didn’t just glance at our content occasionally. It used our pages over 30,000 times to formulate answers to real user questions about AI brand monitoring, GEO, search optimization, and related topics.

Prompt Research based on Grounding Queries

The AI Performance report reveals something no other tool currently shows: the actual queries Copilot uses to find your content. Here’s what we learned from analyzing all grounding queries.

Having analyzed 600+ grounding queries (used in the last few months), I found the following patterns. When analyzing our own performance, I can see that branded grounded queries (obviously) perform better, resulting in nearly a 2x citation density compared to non-branded grounding queries.

Funnel Stage Breakdown

• Consideration (MOFU) dominates query volume: 47% of all grounding queries are driven by tool-discovery patterns. This is where the market is actively shopping.
• Decision (BOFU) is a desert: Only a few queries but lots of citations, all about conversions. I personally think it makes sense. From a volume view, there’s much more research being done at a category / discovery level, than at BOFU stage. But important for us: We are getting cited A LOT when people make purchasing decisions and are up for a product test.
• Branded queries span all stages: from pure navigation (“brand name”) to vertical-specific evaluation (“(my product) for (industry)”) to competitive comparison (“my brand vs competitor brand”).

How We are Adjusting our prompt data set

Based on those grounding queries, we adjusted our prompt dataset that we automatically monitor with OtterlyAI. And I recommend everyone do the same thing: regularly refresh your prompt list.

I’ll basically do two things:

1. Prompt Upload: I upload the grounding queries from Bing’s Webmaster Tools directly into OtterlyAI.
2. Prompt Research: I use the grounding queries to research and find related prompts.

Here are my takeaways that shape the prompts we care about:

1. Vertical-specific queries are a gold mine: Non-branded and Branded queries tagged to verticals (retail, financial services, pharma, advertising, B2B SaaS) drove most website citations for us. This means that Copilot is actively matching us (as a website) to industry-specific buyer intent. This is a clear signal to build dedicated vertical landing pages (which we already had on our list, but now becomes more important).

2. Feature demand reveals buyer priorities: Data security/compliance, alerts, share of voice, reporting, and audit are the features people ask about most. These should be prominent in our product messaging and structured for AI citation.

3. The competitive set is clear: It’s interesting to see also other brands getting mentioned within the grounding queries. For us, we saw for example HubSpot, Mangools, and other – more direct competitors – in there. This is great for comparison content (which we don’t have right now), but also for potential technology partnerships. Copilot is grounding competitive evaluations.

Analyzing our Website Citations

The headline: 5 pages carry 74.6% of all citations. That’s both impressive and dangerous.

Page Type Distribution

Based on our website’s performance, I identified the following patterns.

• Blog posts dominate at 42.7% of all citations, but the power law is brutal. 3 blog posts account for 72% of blog citations, while the majority of posts have fewer than 10 citations each.
• Programmatic pages are volume without impact. Programmatic pages generating only 4.8% of citations. These are programmatic pages that Copilot mostly ignores.
• Help Center is a missed opportunity. Too few pages. But the format (Q&A, definitions) is exactly what Microsoft Copilot loves to cite.

Enter OtterlyAI: What Bing Webmaster Tools Can’t Tell You

The AI Performance report is a significant step forward. But it has clear limitations that matter for any serious AI search strategy:

• It only covers Microsoft Copilot. There’s no equivalent data for ChatGPT, Perplexity, Google AI Overviews, or Gemini. If you’re optimizing for AI search, Copilot is one piece of a multi-platform puzzle.
• It doesn’t show brand mentions in AI responses. Knowing Copilot grounded on your page doesn’t tell you if Copilot mentioned your brand name in the answer. With OtterlyAI, we’ll exactly see some blind spots between being cited as a source, but not mention in the AI answer.
• It doesn’t track competitors. You see your own grounding data, but you can’t see which competitors are cited alongside you, what share of voice you have in AI responses, or how your visibility changes over time relative to the market.
• It doesn’t cover prompt-level analysis. You see grounding queries, but not the original user prompts that triggered them. OtterlyAI tracks the prompts people type into AI engines, mapped to your brand’s appearance in the response.

This is where OtterlyAI complements the picture. OtterlyAI monitors your brand’s visibility across ChatGPT, Perplexity, Google AI Overviews, Microsoft Copilot, and other AI search engines. Tracking visible citations, brand mentions, competitor co-citations, and share of voice from the user’s perspective.

Together, Bing Webmaster Tools + OtterlyAI give you the complete view: how Copilot uses your content behind the scenes AND how users actually see your brand across all AI search engines.

How to Use Both Tools Together: A Practical Workflow

Here’s how we recommend combining Bing’s AI Performance data with OtterlyAI for a complete Copilot optimization strategy:

Step 1: Identify Your High-Grounding, Low-Citation Pages

Export your grounding data from Bing WMT and compare it against your visible citation data in OtterlyAI. Pages with high grounding events but low visible citations are your biggest optimization opportunities.

Step 2: Map Grounding Queries to Content Gaps

The grounding queries in Bing WMT reveal what Copilot is trying to answer when it uses your content. Cross-reference these with the prompts OtterlyAI tracks to find queries where you’re being grounded but not cited – or where competitors are getting cited instead.

Step 3: Track Cross-Platform Visibility

Your Copilot grounding data might look strong, but what about ChatGPT, Perplexity, and Google AI Overviews? Use OtterlyAI to monitor your brand’s presence across all AI search engines. A page that’s well-grounded in Copilot might be completely absent from Perplexity.

Step 4: Monitor Competitor Co-Citations

Bing WMT shows your data. OtterlyAI shows who appears alongside you. In our data, Semrush appeared as a competitor on 5 of our 13 visibly-cited pages. Understanding your competitive context in AI responses lets you target the right comparison content and positioning.

Step 5: Optimize for Both Grounding and Visible Citations

Based on our analysis, different content types serve different purposes:

For grounding authority (getting Copilot to use your content): Create comprehensive, well-structured informational content. Use clear entity definitions, structured headings, and factual claims with sources. Our blog posts and guides drive the bulk of grounding events.

For visible citation rate (getting Copilot to link to you): Invest in third-party validation — awards, reviews, analyst coverage, and community discussions. Our G2 award post was the only page where visible citations exceeded grounding events. Make it easy for AI to cite you by including clear, quotable statements and structured FAQ content.

What This Means for AI Search Optimization

The launch of Bing’s AI Performance report marks a turning point for GEO and AI search optimization. For the first time, we have first-party data from an AI search provider about how AI systems use web content.

The key takeaway from our analysis: the relationship between AI content usage and visible attribution is far more complex than most people assume. Being “used” by AI 44,469 times while being visibly cited 169 times means 99.6% of your AI influence is invisible. That’s not a problem to solve — it’s a reality to understand and optimize around.

Bing Webmaster Tools tells you how Copilot consumes your content. OtterlyAI tells you how users experience your brand across all AI engines. Together, they give you the data foundation for a complete AI search optimization strategy.