Technical explanation
Each execution is stored as a run with its engine, locale, timestamp and full response, which is what makes variability, trend and confidence analysis possible.
Example
Running "best data recovery service in London" weekly across three engines produces a comparable time series.
How it differs
A single run is an anecdote. A tracked prompt with repeated runs is data.
Related terms
Where Varazm measures this:
AI Visibility MonitoringMethodology