How to measure generated video performance.
Choose metrics that distinguish discovery, consumption, engagement and commercial results.
Write the hypothesis first.
State what the product should improve: more starts, longer sessions, deeper viewing or more visits to the full story. Select a primary metric and a comparison method before looking at the result.
Use the right denominator.
Click-through rate needs a consistent impression definition. Watch time per session needs a consistent session definition. Completion depends on duration. Report the underlying counts so a change in audience size does not look like a change in user behavior.
Control the comparison.
Where possible, split comparable traffic between the existing and new experience. Keep placement, device, geography and content mix comparable. Document the test period and distinguish correlation from a measured product effect.
Read the whole outcome.
More starts can coexist with shorter watch time. More ad inventory can coexist with lower completion. Review discovery, viewing and revenue together, and use content-level analysis to understand which experiences actually help the audience.