ubricHelp

Estimate vs measured citations

What your citability score predicts, what it can't tell you, and how to calibrate it against real AI citations.

Your Rubric score is an estimate of how citable a page is, built from on-page, structural and technical signals. It predicts citability; it does not count the citations you have already earned. To compare the prediction with reality, calibrate it against your Bing Webmaster Tools AI Performance export, which reports where AI surfaces actually cited your pages.

In short
  • →The score is an estimate from on-page signals. It predicts how citable a page is; it does not count citations you have earned.
  • →A measured citation is the opposite: a record, after the fact, of where an AI surface actually quoted you.
  • →Calibrate the estimate against your Bing Webmaster Tools AI Performance export to see how prediction and reality line up.

What does the score predict?

The score predicts citability: how likely an AI engine is to quote a page when it answers a question on that topic. Rubric reads what an engine reads and asks, in effect, "if a model were choosing a source for this question, how ready is this page to be the one it names?" It is forward-looking. You get a score before you have earned a single citation, which is exactly what makes it useful for deciding what to fix. An estimate you can act on today beats a citation count that only tells you what already happened.

What can the score not tell you?

The score cannot tell you three things, and being clear about them keeps you from over-reading it.

  • It does not count earned citations. The number is a prediction from signals, not a tally of times an engine has quoted you.
  • It does not measure off-page authority. Brand demand, backlinks and reputation all affect whether you are cited, and none of them are on the page. Even the Compare teardown is on-page only; it does not measure a competitor's off-page authority.
  • It is not a guarantee. Rubric is guidance, not a promise. A high score improves your odds; it does not commit an engine to citing you.

How is an estimate different from a measured citation?

An estimate and a measured citation answer different questions and point in opposite directions in time. The table sets them side by side.

Rubric citability scoreMeasured citations
What it isAn estimate from on-page signalsA count of where an AI surface cited your URL
DirectionForward-looking, before you are citedBackward-looking, after the fact
Question it answersIs this page ready to be quoted?Which pages were actually quoted?
Blind spotOff-page authority and brand demandWhy a page was or was not cited
Where it comes fromYour Rubric reportBing Webmaster Tools AI Performance, for example

The two are complementary. The estimate tells you what to change; the measurement tells you whether the change worked.

How do I calibrate the estimate against real citations?

Calibration means lining your Rubric scores up against a source that records real AI citations, and Bing Webmaster Tools is the practical one to use.

01
Export your Bing Webmaster Tools AI Performance data. It reports, per URL, where Copilot and Bing's AI surfaces cited your pages.
02
Match it to your Rubric page scores. Put the measured citations next to the estimate for the same URLs.
03
Read the gap. A high score with citations means the estimate is working. A high score with no citations usually points to off-page factors like authority or demand. A low score that is still earning citations flags a page to protect and strengthen.
Note
Bing Webmaster Tools is the most practical measured window because most AI engines do not yet expose which pages they cited. It covers Copilot and Bing's AI surfaces, so treat it as a strong sample of reality rather than the whole of it. The dedicated calibrating with Bing Webmaster Tools article walks through the full workflow.

Why would a citable page still not be cited?

A page can score well above 70 and still not be cited, because the score measures on-page readiness and not the whole picture. Off-page authority, brand demand and simple competition all decide which ready page an engine picks, and engines do not return the same source every time. The honest way to read the score is as improving odds, not a promise: it tells you a page is ready to be quoted, then your measured data tells you whether it was. When a strong page is not earning citations, the fix is usually off the page, not on it.

Common questions

Did this answer your question?