Separate a technical regression from an audience difference

A technical regression should be tested with a controlled reference before audience behaviour is used to explain the difference.

A practical procedure

1. Replay the same media and question set on the previous and changed setup.

2. Compare playback continuity, timing and errors without a live audience.

3. Then review audience differences separately, keeping technical and behavioural observations in distinct columns.

Worked example: a fictional session

A fictional team reproduces delayed audio with the same local sample after an update, so it investigates the setup before blaming a quieter audience.

Check the limits

A normal controlled test does not prove that every live workload is technically identical.