Non-skippable video guarantees the viewer sees the whole spot. What it cannot guarantee is that anyone is watching, and the distinction determines when the format is worth buying.
Forced exposure changes viewer behavior
A viewer who cannot skip has two options: watch, or look away and wait. A substantial share choose the second and check a phone until the content resumes.
The impression is served and counted regardless, so the reported delivery is perfect while the actual attention is unmeasured.
Which means completion rate in this format describes the player rather than the audience, and treating it as an engagement metric is a category error.
Skippable formats sort the audience
When skipping is available, viewers who stay have made a small decision to keep watching. That self-selection is information the non-skippable format destroys.
Advertisers using skippable inventory can read the skip point as feedback on the creative, since the moment people leave says something about where it lost them.
Non-skippable buying removes that feedback, so creative problems persist without any signal that they exist.
The trade depends on the objective
For a message that works on recognition, a familiar brand, a simple announcement, forced exposure delivers what is needed even from a half-attentive viewer.
For a message that requires comprehension, an unfamiliar product or a complex offer, an inattentive viewer contributes nothing while costing full rate.
The format is therefore well matched to broad awareness and poorly matched to explanation, which is the opposite of how it is often sold.
Short durations change the calculation
Very short non-skippable spots are over before the viewer can disengage, which is why the six-second format exists and why it behaves differently.
At that length the entire message has to be a single idea, and anything requiring setup will not survive the cut.
Longer non-skippable placements ask more of the viewer's patience than the exposure is usually worth, and irritation is a cost that lands on the brand.
Measuring what actually happened
Because completion is uninformative here, useful measurement has to come from elsewhere: brand tracking, search behavior after exposure, or a controlled holdout.
Those methods are slower and less precise than a delivery report, which is why teams keep reaching for the number that is available rather than the one that answers the question.
Deciding in advance which measure will judge the buy is what prevents a campaign from being declared successful on the strength of a metric that cannot fail.