Most viewers of a feed video decide whether to continue within the first couple of seconds. Everything after that point is only seen by whoever survived the opening.
The scroll is the default action
A feed presents an endless supply of alternatives, each one costing nothing to reach. Continuing to watch is the choice that requires justification.
The viewer is not evaluating the ad against other ads. They are evaluating it against the next item, which is free and arrives instantly.
That comparison happens before comprehension, on whatever the first frames present.
Traditional structure puts the payoff last
Broadcast advertising was built for a captive audience, so building tension and resolving it at the end was a sound structure.
In a feed the ending is seen by a small remainder of those who saw the beginning, which inverts where value should be placed.
Ads that open with the product in use, or the strongest visual in the sequence, retain more viewers than ads that build toward it.
What survives the opening
Movement in the first frame, a face, an unusual composition, and immediate visual clarity all extend viewing measurably.
Openings that begin with a logo or an establishing shot perform poorly, because they present nothing the viewer can act on.
The reliable test is whether a still of the first frame communicates anything at all, since that still is what many viewers effectively see.
Retention curves are more useful than view counts
A view count aggregates everyone, including those who left immediately. Two ads with identical view counts can have entirely different audiences.
The retention curve shows exactly where people leave, and the shape of the first seconds separates ads that were skipped from ads that were watched and abandoned.
A sharp initial drop is an opening problem. A gradual decline through the middle is a pacing problem, and the two need different fixes.
Why the same ad needs several openings
The body of a video is expensive to produce and the opening is cheap to vary, which makes opening tests the most efficient creative work available.
Cutting three or four alternative first seconds onto the same body often changes retention more than a full reshoot would.
It also extends the usable life of the asset, since a worn-out ad frequently recovers with a new opening rather than needing replacement.