Average view duration

It is the per video counterpart to watch time. Where watch time measures accumulated hours, average view duration measures how well one video holds an individual viewer, which makes it comparable across videos of different popularity. The related percentage figure normalises for length, so a two minute video and a twenty minute one can be judged side by side.

For short form the number is unforgiving. On a thirty second clip the difference between eleven and twenty two seconds of average duration is the difference between a post that stalls and one that keeps being shown, because most drop off happens in the first three seconds. Watching the retention graph reveals exactly where the promise stopped being kept.

The nuance is that averages hide shape. Two videos can share the same duration while one loses half its audience immediately and holds the rest, and the other bleeds viewers steadily throughout. Those need different fixes: the first is a hook problem, the second is a pacing problem, and the single average number cannot distinguish them.

Practically, creators raise it by cutting dead air, front loading the payoff, and placing a pattern interrupt just before the point where the graph dips. Automatic clipping helps by proposing cut points around segments that already hold attention in the source, which removes most of the guesswork about where a clip should start.