Glossary · Measurement
Response Variance
The statistical spread of a metric — sentiment, mention position, inclusion — across repeated runs of the same prompt or across different engines.
How it differs from consistency
Response consistency is typically framed as a simple stability check — same or different, high or low. Response variance quantifies that spread numerically, which makes it possible to compare, for example, whether sentiment swings more across repeated runs for one prompt than another, or across one engine more than another.
Why it matters
A metric with high variance needs more repeated samples before any single reading can be trusted, while a low-variance metric can be read confidently from just one or two runs. Ignoring variance is how teams end up reacting to noise instead of an actual trend.
See more terms in the full glossary, or read the AI Visibility 101 guide for the full picture.