ubricHelp

Calibrating with Bing Webmaster Tools

Compare the Rubric estimate with measured AI citations from Bing, per URL, to see where they agree.

To check the Rubric estimate against real citations, export your Bing Webmaster Tools AI Performance report and compare it with your Rubric scores page by page. Rubric estimates how citable a page is from on-page signals; Bing's AI Performance data measures citations a page actually earned. Comparing the two per URL shows where the estimate and the measured result agree.

In short
  • →Rubric estimates citability from on-page signals. It does not count citations you have already earned.
  • →Bing Webmaster Tools AI Performance is measured data: it reports pages that were actually cited by Bing and Copilot.
  • →Compare the two per URL to see where a high score matches real citations and where a gap needs investigating.

Why calibrate the estimate against measured data

Rubric's score is an estimate, not a citation count. It reads what an engine reads and predicts how ready a page is to be quoted, but it cannot see the citations a page has already earned. Measured data can. Calibrating means putting the estimate next to the measurement so you can trust the number: where a high Rubric score lines up with real citations, the estimate is holding, and where they diverge, you learn something useful about that page.

Bing Webmaster Tools is the practical source of measured AI citations, because its AI Performance report shows which of your URLs were used to answer AI queries.

What Bing Webmaster Tools AI Performance gives you

Bing Webmaster Tools AI Performance is a report of how your pages performed in AI-generated answers on Bing and Microsoft Copilot. It is measured data: it reflects what actually happened, URL by URL. That makes it the honest counterpart to the Rubric estimate. Rubric tells you how citable a page should be; Bing tells you whether it was cited.

The two answer different questions, which is exactly why comparing them is worth doing.

How to compare Bing data with your Rubric scores

Line up the measured data against the estimate on a per-URL basis.

01
Export the AI Performance report from Bing Webmaster Tools. This is your measured citation data.
02
Export pages.csv from your Rubric report. This gives every crawled page with its score.
03
Match the two on the page URL. Join the Bing rows to the Rubric rows so each URL has both its score and its measured citations.
04
Read each URL's Rubric score against whether Bing shows it cited. Look for agreement and for gaps.

How to read the gap

Four patterns tell you what to do next.

Rubric scoreCited in Bing AI dataWhat it suggests
At or above 70YesThe estimate is holding. Keep the page fresh and defend the position.
At or above 70NoOn-page is ready. The gap is likely off-page authority, freshness or topic demand.
Below 70YesYou are cited despite weak on-page signals. Shore the page up to hold the position.
Below 70NoExpected. Work the Action Plan to lift the page above the line.

The most instructive row is a score at or above 70 with no citations. Rubric measures on-page citability only, so this pattern points away from the page and towards off-page work: authority, freshness or whether anyone is asking the question at all.

Note
Bing Webmaster Tools AI Performance reflects Bing and Microsoft Copilot. Rubric weights six engines: ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot and Claude. Treat the Bing data as one measured signal, not the full picture across every engine.
Before you read too much into a fix
Measured citation data reflects what has already happened. After you improve and re-crawl a page, give engines time to re-crawl and cite it before you expect the measured column to move.

What this can and cannot tell you

Calibration tells you how well the estimate is tracking reality on the pages you can measure, which is the honest way to build confidence in the score. It cannot turn Rubric into a citation counter, and Bing's data covers only its own engine and Copilot. Used together, though, the estimate and the measurement give you both a forward-looking guide and a backward-looking check.

The related articles go further: estimate vs measured citations covers the difference between the two in principle, why 70 is the line explains the threshold you are calibrating around, and what is AI citability sets out the bigger picture on being quoted by AI.

Common questions

Did this answer your question?