How to Read Your Bing AI Performance Report (And What the Zeros Mean)
- Patrick Duggan
- 3 minutes ago
- 3 min read
The Bing AI Performance report, found inside the free Bing Webmaster Tools, shows how often Microsoft Copilot and Bing's AI-generated answers cite your website's content. It launched in public preview on February 11, 2026, and expanded on June 16 with intent labels, topic clusters, a Citation Share metric, and period-over-period comparison. It is the first official, vendor-run scoreboard for whether the AI layer that is replacing search can see you at all. Here is how to read it without panicking.
The metrics, one at a time
Citations is the headline number: how many times, in the selected period, an AI answer referenced your content. This is the score that matters most, and for almost everyone it is shockingly low. Cited Pages tells you which of your URLs got referenced — often a much smaller set than you would guess, which tells you where your machine-legible strength actually lives.
Grounding Queries are the reformulated questions Copilot wrote to itself before reading the web on a user's behalf. They are your demand signal: the intents the model was trying to satisfy when it went looking. Intents and Topics, added in June, cluster those queries so you can see the themes you are being pulled into. Citation Share is your slice of the cited results for a given topic versus everyone else competing to be the source — the closest thing yet to an AI-era market-share number.
Why a column of zeros is actually the good news
Pull your report and you will probably see mostly zeros. We did. Microsoft's own instrument says Copilot has cited our main blog eleven times total and our flagship product site eight times, with multi-week unbroken streaks of zero on both. We sell AI-visibility auditing for a living, and those are our real numbers.
Here is why that is not the disaster it looks like. The generative layer in mid-2026 is citing a vanishingly small slice of the entire web. The days of zeros in your export are the days of zeros in nearly everyone's export, including companies a thousand times your size. That means AI visibility is not a field where you are hopelessly behind. It is a near-empty field — which makes it the single most winnable surface on the internet right now. A column of zeros does not mean you lost. It means the game barely started and almost no one is playing yet.
What to do after you read it
First, look for the gap: grounding queries firing on topics you cover, paired with few or zero citations for those topics. That gap is the model asking a question you can answer while failing to choose you. The fix is rarely more content — it is legibility. Add Schema.org structured data so the model can ground itself in who you are. Write answer-shaped pages: clean definitions and direct statements of fact match Copilot's reformulated queries far better than marketing prose. Keep your robots.txt open to the AI crawlers and ship an llms.txt.
Second, measure honestly and repeatedly. The number that predicts whether a company will fix its AI visibility is not its citation count. It is whether the company is willing to look at its citation count at all. We publish ours specifically to prove we read the test before we hand it to anyone else.
So open Bing Webmaster Tools, find the AI Performance tab, and read your citation count out loud. If it is a wall of zeros, you are not behind — you are early. Then go fix the layer the machines actually read. Every figure we cite here is capped at 95 in our own scoring on principle, because we guarantee five percent of any confident number is nonsense — but the scoreboard finally exists, and that part is real.
How do AI models see YOUR brand?
AIPM has audited 250+ domains. 15 seconds. Free while still in beta.