Audience retention
Also called retention · average percentage viewed
How much of a video people actually watch, reported by YouTube Analytics as both a curve across the runtime and an average.
In more detail
The retention curve shows what proportion of viewers are still watching at each moment; the summary figures are average percentage viewed and average view duration. The curve is the more useful of the two, because it localises problems: a cliff thirty seconds in and a slow bleed throughout are both mediocre averages with entirely different causes.
Example
A steep drop in the opening seconds usually means the video did not deliver what the title and thumbnail implied. A dip in the middle that recovers usually means one segment dragged. The same average could describe either, which is why the average alone rarely tells you what to change.
Why it matters
Retention is where AI-assisted production most visibly fails. A script assembled in disconnected chunks tends to repeat itself and lose its thread, and viewers leave at exactly the point where it stops going anywhere. It is a direct read on script quality rather than a vanity metric, and it is the reason coherence across a long script matters more than the polish of any individual paragraph.