Audience retention in short-form video advertising falls fastest at the very start, and the shape of that curve is remarkably consistent across formats and platforms. It makes the opening the most expensive real estate in the edit.
The decision is made before comprehension
A viewer in a feed is deciding whether to continue, not evaluating an argument. That decision happens on visual impression and motion before any sentence has finished.
Advertisements that open with a logo animation, an establishing shot or a slow build present nothing to decide on, and the viewer defaults to leaving.
The consequence is that a considerable share of a video budget can be spent delivering an opening that never reached the part carrying the message.
Retention loss compounds through the edit
Everything after the opening is watched only by whoever survived it, so the cost per viewer of the second half is far higher than the cost of the first.
Improving the opening therefore raises the value of material already produced, without touching the rest of the film at all.
This is why testing openings against a fixed body is usually more productive than testing whole executions against each other, where the effects cannot be separated.
Sound is not available at the start
Autoplay in most feeds begins muted, and viewers who unmute do so after deciding the video is worth attention rather than before.
An opening that depends on a spoken line is therefore silent for the audience it most needs to hold, and the line is wasted.
Visible text, legible motion and a recognisable subject perform the work that dialogue performs in environments where sound is assumed.
Branding early conflicts with holding attention
Placing the brand at the very start guarantees exposure for viewers who leave immediately, but a visible advertisement is also a reason to leave.
Delaying the brand improves retention and risks the campaign generating attention that the viewer never attributes to the advertiser.
The workable compromise is integration rather than sequencing, with the product visibly present in the action so that identity accumulates while the viewer is still watching.
Retention curves are the useful diagnostic
Platforms report where viewers stop, and reading that curve locates the problem far more precisely than an aggregate completion figure.
A cliff in the first seconds is an opening problem, while a steady decline through the middle usually indicates the film is longer than its content justifies.
A late drop before a call to action suggests the payoff arrives after the audience has concluded there is nothing further coming, which is an editing fault rather than a media one.