Ask most marketing teams how their website is performing and they'll pull up Google Analytics or a rankings tracker. Ask them how their content is performing inside Microsoft Copilot, and you'll usually get a shrug. Until a few months ago, that shrug was the only honest answer available, because no major AI engine gave publishers a real window into it.
Microsoft just closed that gap for its own ecosystem. Bing Webmaster Tools now includes a dashboard called AI Performance, showing exactly which grounding queries and cited pages are pulling your content into Copilot's answers, how often, and whether that's changing over time. For teams that already talk about AI visibility in strategy meetings but have mostly been measuring it by instinct, this is the first real data set to check that instinct against.
This matters most for the teams stuck in the middle: too big to ignore AI-driven demand, too lean to run a dedicated GEO department. If that sounds like your team, here's what the AI Performance report actually shows inside your webmaster dashboard, how to turn those numbers into decisions, and why Bing's view, however useful, is still only one piece of the AI visibility monitoring you actually need.
Why Bing's AI Performance Report Is a Big Deal Right Now
For most of the last two decades, "search visibility" meant one thing: where you ranked on a results page. Bing Webmaster Tools and Google Search Console built entire reporting suites around that idea, tracking impressions, clicks, and position for a list of blue links.
That model is cracking. Microsoft Copilot now answers questions directly inside Bing, Windows, and Microsoft 365, instead of just handing back ten links to click through. When someone asks Copilot a question, it doesn't send them off to find the answer themselves. It writes the answer, pulling from your content as one of several sources, and the person may never click through to your site at all. By one McKinsey estimate cited on Microsoft's own advertising blog, roughly half of consumers already use AI-powered search, a shift projected to influence something in the neighborhood of $750 billion in commerce by 2028.
Until February 2026, publishers had no way to see any of this happening on their own sites. You could suspect your content was being pulled into Copilot's answers. You had no real way to confirm it, measure it, or improve it. Then Microsoft rolled out AI Performance inside Bing Webmaster Tools, first as a limited public preview, then expanded in June 2026 with four new reporting layers. For the first time, a major search engine gave publishers a direct, first-party window into how often their content gets cited in AI-generated answers, and which pages and queries are actually driving that visibility.
Google didn't ship a comparable feature, a Search Generative AI performance report inside Search Console, until June 3, 2026, and even then it launched as a UK-only test showing impressions only, with no citation-level detail and no click data. Bing got there first, and right now Bing's version goes considerably deeper. That gap matters. If your team has been waiting for "official" AI visibility data before taking this seriously, that data now exists, and it already has several months of maturity behind it.
What features does the AI performance report offer in webmaster dashboards?
AI Performance lives inside Bing Webmaster Tools, right alongside the crawl stats and search performance reports most SEO teams already check. But it measures something fundamentally different: not rankings or clicks, but whether and how often your pages get pulled in as a source when Microsoft Copilot, Bing's AI-generated summaries, or a handful of partner surfaces build an answer.
At launch, the dashboard shipped with a core set of metrics:
- Total Citations: the number of times your site was used as a source in AI-generated answers over the selected date range.
- Average Cited Pages: the daily average of unique URLs from your site that showed up as sources.
- Grounding Queries: the query phrases the AI system used internally to retrieve your content. These aren't always what a person actually typed. Copilot often rewrites a user's question into one or more retrieval-style queries behind the scenes, and Bing surfaces those rewritten phrases rather than the raw prompt.
- Page-level citation activity: citation counts broken down by individual URL, so you can see exactly which pages are earning the most references.
- A trend timeline showing how citation volume moves over the selected period.
About a month after launch, Microsoft connected the queries and cited pages views to each other directly. Click a grounding query and you see every page cited for it. Click a page and you see every query that triggered a citation to it. That mapping is arguably the most useful part of the whole dashboard, because it turns a flat list of citations into something a content team can actually act on.
Then in June 2026, the dashboard grew four more layers, all rolling out globally in preview:
- Intents groups your grounding queries into categories like Informational, Commercial, Navigational, Research, and Local, so you can see what kind of user intent is driving your citations, not just the raw count.
- Topics clusters related queries into broader subject areas. Instead of reviewing "solar panels," "solar energy efficiency," and "residential solar installation" as three separate line items, Topics rolls them into a single Solar Energy theme, closer to how editorial teams actually plan content.
- Citation Share shows what percentage of the citations for a given grounding query belong to your site, out of every source cited for that query. Microsoft has been explicit that this is an observational metric, not a competitive leaderboard; it won't show you which competitor picked up the rest of the share.
- Compare overlays two time periods, current versus the last 30 days or a custom range, so you can see whether your citation activity, by intent or topic, is actually moving in the direction you're pushing it.
Together, these give a mid-sized team something that didn't exist eighteen months ago: a first-party, no-cost way to see not just whether Copilot is citing you, but why, on what kind of query, and whether that's changing over time.
How can I analyze AI-driven site performance metrics in webmaster tools and apply the insights to my business.
Having the data is one thing. Turning it into decisions your team can actually execute on is another. Here's a practical way to work through it, built for a team that doesn't have a dedicated GEO department but does have a content calendar and a few hours a week to spend on this.
Start with a baseline, not a verdict. Verify your site in Bing Webmaster Tools if you haven't already; it's free and takes a few minutes. Then pull your Total Citations, Average Cited Pages, and Citation Share for the last 30 days. Your first numbers will probably look thin. That's normal. Microsoft has said the underlying data is sampled and still maturing, and there's typically a two-to-three day lag before new content shows up at all. Treat this first pull as day one, not a report card.
Read grounding queries as intent, not keywords. Because grounding queries are the AI's own rewritten retrieval phrases, they tell you something classic keyword research can't: how an AI system is actually interpreting the questions people bring to it. Export the list, and instead of asking "did I rank for this," ask "does my content actually answer this specific framing of the question."
Use Intents to find the gap in your funnel. If your citations lean heavily Informational and you're barely showing up for Commercial or Navigational queries, that's a signal worth acting on. It usually means your educational content is strong but you're missing the comparison pages, pricing breakdowns, and "best X for Y" content that AI systems reach for when someone's closer to a decision.
Use Topics the way your editorial team already thinks. Most content teams plan around themes and pillars, not individual keywords. Topics applies that same lens to AI citation data, so you can see which subject areas you already own and which ones are still open.
Treat Citation Share as a health check, not a scoreboard. A low share on a query that matters to your business is worth investigating. Is the content too thin, too generic, or simply not structured in a way that's easy for a model to extract and cite? A rising share across successive Compare windows is a good sign your changes are working.
Close the loop with Compare. Whatever you change, whether it's tightening a page's structure, adding a clear FAQ block, or filling a topic gap, give it a few weeks and then overlay the periods. That's the only way to know if the update actually moved the needle inside AI answers, rather than just feeling like it should have.
None of this replaces the content fundamentals your team already knows. Clear headings, direct answers near the top, claims backed by real evidence, and content that stays current all show up repeatedly in Microsoft's own guidance for pages that get cited more often. AI Performance doesn't hand you a new playbook so much as it finally shows you whether the playbook you're already running is actually working.
Where Bing's Data Runs Out
Here's the part worth sitting with for a minute. Everything above is genuinely useful, and it's free. It's also only a slice of the picture.
AI Performance only sees the Microsoft ecosystem: Copilot, Bing's AI-generated summaries, and a handful of partner integrations. It has no visibility into ChatGPT, Perplexity, Google's AI Overviews or AI Mode, Gemini, or Claude. If your buyers are asking questions across all of those surfaces, and most of them are, Bing's dashboard shows you one lane of a much wider highway.
The report also comes with real limitations even inside its own lane. There's no click data, so a citation confirms your content was used as a source, not that anyone visited your site afterward. The data is sampled rather than exhaustive. There's no API yet, which limits how much of this you can automate into an existing reporting stack. And because it's still a public preview, Microsoft has said both the metrics and the underlying methodology are likely to keep changing through 2026.
That combination, genuinely useful but structurally partial, is exactly why AI visibility monitoring needs to mean more than checking one dashboard. A dedicated ai search visibility platform applies the same discipline Bing just brought to its own ecosystem (tracking citations, mapping grounding queries to pages, watching how visibility shifts over time) across every surface where your buyers are actually asking questions, and adds the competitive and content-strategy layer a single-engine report was never built to provide.
For a medium-sized team, that's the real wake-up call buried in this launch. Not that Bing shipped a new report, but that AI visibility has become measurable enough, and consequential enough, that a company the size of Microsoft built first-party infrastructure for it. If Copilot's slice alone is worth a dedicated dashboard, the full picture across every AI surface is worth a real strategy, not an occasional glance at one report.
The Bottom Line
Microsoft didn't build this dashboard as a courtesy. It built it because enough of its own search traffic has shifted into Copilot's conversational answers that "were we cited" became a question worth answering with real infrastructure. That's the actual news here, more than any single metric on the screen.
For a mid-sized marketing or SEO team, the practical move is straightforward. Verify your site this week if you haven't already. Pull your first citation baseline and treat it as day one, not a grade. Read your grounding queries the way you'd read a focus group transcript: as evidence of how people are actually asking, not how you assumed they'd ask. Fix the structural gaps that surfaces. And once checking AI Performance becomes as routine as checking Google Analytics, widen the lens past Bing, because Copilot is one voice in a conversation happening across half a dozen AI surfaces at once, and the report in front of you only covers one of them.
Frequently Asked Questions
What is Bing's AI Performance report, and when did it launch?
AI Performance is a dashboard inside Bing Webmaster Tools that shows how often your website's content is cited as a source inside AI-generated answers across Microsoft Copilot, Bing's AI summaries, and select partner surfaces. Microsoft rolled it out in public preview on February 10, 2026, then expanded it in June 2026 with four additional reporting layers: Intents, Topics, Citation Share, and Compare.
What are "grounding queries," and why do they matter more than regular keywords?
Grounding queries are the retrieval phrases an AI system generates internally while building an answer. They're often reworded, expanded, or split apart from whatever the user actually typed. Because they reflect how the AI is really interpreting a question, they give you a more accurate picture of intent than traditional keyword data, and they're the foundation the report's queries and cited pages mapping is built on.
Does AI Performance cover ChatGPT, Perplexity, Gemini, or Google's AI Overviews?
No. AI Performance only tracks citations inside the Microsoft ecosystem: Copilot, Bing's AI-generated summaries, and a small number of partner integrations. It has no visibility into other AI engines. Google launched a comparable but more limited report of its own, the Search Generative AI performance report in Search Console, on June 3, 2026, though it currently shows impressions only, with no citation-level or click data, and started as a UK-only rollout.
How is an AI search visibility platform different from checking Bing's native AI Performance report?
Bing's report is free and genuinely useful, but it only sees its own ecosystem, offers no click data, relies on sampled data, and has no API for automation yet. A dedicated ai search visibility platform applies the same citation-tracking approach across every AI surface where your buyers actually search (ChatGPT, Perplexity, Gemini, Copilot, and more) and adds the competitive benchmarking and content strategy layer a single-engine dashboard isn't built to provide.
How often should a medium-sized team check AI visibility monitoring data?
A weekly glance is usually enough to catch anything unusual, but the real decisions happen monthly. Pull your citation, intent, and topic data once a month, compare it against the prior period using the Compare feature, and use that cadence to decide what content to fix, expand, or build next. Checking daily rarely adds insight, since citation data is sampled and updates with a short reporting lag.