Completion rate is one of the most quoted video metrics and one of the least informative. It responds to the conditions of delivery more strongly than to anything about the ad.
Unskippable placements complete by construction
An ad the viewer cannot skip will show a very high completion rate regardless of quality, because leaving requires abandoning the content behind it.
Comparing that number against a skippable placement compares two different things and always favours the format with fewer exits.
Buying toward high completion therefore pushes budget into forced formats rather than into effective creative.
Length is the other dominant variable
Shorter ads complete more often for the simple reason that there is less of them to abandon.
A six-second unit will out-complete a thirty-second unit even when the longer ad holds attention better in absolute terms.
Comparing completion across durations rewards brevity rather than performance, which is a preference rather than a finding.
A completed view is not a watched view
Video plays while the viewer scrolls, while a tab is in the background, and while the television is on in an empty room.
Each of those can register as a completion, since completion measures the file reaching its end rather than a person reaching it.
Muted playback counts identically to audible playback, so an ad that depends on its voiceover completes at the same rate as one that does not.
The metric cannot separate a viewer who watched deliberately from a device that continued playing after the person left the room, and it treats both as success.
What the number is genuinely useful for
Held constant on format and duration, completion becomes a reasonable comparison between two edits of the same length in the same placement.
It is also useful in reverse, since an unusually low completion in an unskippable placement indicates something actively repellent about the ad.
Beyond those cases, the number describes the buy rather than the work.
The measures that survive scrutiny
Retention curves show where attention was lost, which is actionable in a way that a single completion figure never is.
Audible and visible time, where a platform reports it, separates ads that were experienced from ads that merely played.
And for anything with a commercial objective, downstream behaviour remains the only measure that cannot be improved by changing the placement rather than the ad.