Skip to content
Why we do / don’t

Why is the Copilot engine labelled a Bing index measurement?

We label Copilot a Bing index measurement because that is what it is. Copilot draws its web grounding from the Bing index, so we measure that index directly and say so on the label. The honest alternative was to proxy Bing silently and present it as a Copilot reading, which would hide real sampling error from you.

Every measurement we publish carries a provenance label describing how it was collected, not just what it is called.

The correlation between an index position and a Copilot citation is directional. It is a useful signal, and it is not the same thing as observing the assistant answer. Labelling it precisely is what lets you weigh it correctly against the engines we do observe on the consumer plane.

The alternative most tools take is to present a proxy under the assistant’s name. That reads cleaner and hides the sampling error inside a number you cannot interrogate. We would rather show you the seam.

What this answer is based on

  • Engine registry — per-engine provenance labels
  • Provenance labelling policy

Last updated August 3, 2026

← All answers