The long definition
Video retention describes how viewing continues over time. A retention curve can show where viewers leave, stay, or replay a moment, depending on the platform’s reporting method. It helps connect an edit or narrative beat to observed viewing behaviour.
Retention is not the same as completion rate or average watch time. Compare videos of similar lengths, audiences, and distribution conditions. Public view and like counts cannot reconstruct a private retention curve, so missing data should remain explicitly missing.
Anatomy: Retention
Read the metric in context before deciding which moment to change.
- 01SourceIdentify the platform and the metric definition.
- 02OpeningLook at viewing behaviour around the hook.
- 03BodyLocate exits near transitions or delayed explanations.
- 04ComparisonCompare like-for-like videos before drawing conclusions.
A concrete example
Illustrative example · product tutorial
“Move the demonstration before the long introduction.”If your own analytics show viewers leaving during the introduction, test a shorter opening.
This is a hypothesis informed by an observed exit point. Only a subsequent comparison can show whether the new edit helped.
Do
- Record the metric source.
- Compare similar video lengths.
- Distinguish observations from hypotheses.
Don't
- Infer retention from likes.
- Apply one universal benchmark.
- Treat watch time as proof of sales.
How Reacast uses this concept
Use available viewing data to inform your Reacast brief. If an inspiration has no retention data, describe its structure as a creative hypothesis rather than a measured retention result.