What is Video ad?
A video ad carries a message over time rather than all at once, which makes it the only format that can demonstrate, sequence or build. It also has to earn its opening before any of that is delivered.
The format that can sequence
A video ad delivers its message over time. That is the only structural difference from a static, and everything else about the format follows from it.
Sequence buys four things no single frame can provide. It can demonstrate — show a thing working rather than assert that it works. It can build — establish a situation, then complicate it, then resolve it. It can reveal — withhold something and deliver it, which is what makes a viewer stay. And it can pace — control the order in which information arrives, which a static surrenders entirely.
The price is that none of it is delivered unless the opening earns the time. A static is taken in whole in the instant it is on screen; a video that loses its audience in the first second has delivered nothing at all, however good the rest is.
The opening decides how much of the ad exists
On any scrolled surface, most of the audience an ad is served to leaves in the first second or two. Whatever follows is delivered to the remainder.
That makes the first moments the highest-leverage part of the asset by a wide margin, and it produces a set of practical rules that hold across surfaces.
Open on motion. A still first second gives up the mechanism the format offers on a scrolled feed.
Put the subject in the first frame. It is also the thumbnail, and it is what a fast scroll sees.
Start mid-action. A shot already in progress reads as content; a shot easing in reads as an advertisement beginning.
Put the point before the branding. Establishing who is speaking before saying anything worth hearing spends the decisive moment on information nobody has a reason to want yet.
Write for silence. Captions carry the message; audio decorates it.
Length follows the idea
The most common structural fault in video advertising is a video longer than the thing it has to say.
The retention curve names it. A curve that decays gently and evenly all the way down is almost always length rather than content: nothing is going wrong at any particular moment, and there is simply more video than idea. A shorter cut of the same footage usually outperforms the original on every measure.
A curve that drops sharply at a specific moment is different — it names a scene, a claim, a price or a tonal shift, and that moment is editable.
Neither diagnosis comes from a duration guideline. The useful practice is to cut to the idea and read the curve, rather than to pick a length and fill it.
Silence, aspect ratio and the surfaces
Every feed surface imposes the same three constraints and accounts underestimate all of them.
Sound is off. Most views begin silently and many end before anybody would have turned it on.
The frame is vertical, or square, or both. A landscape asset cropped to vertical loses its subject and its composition. Producing for the shapes the campaign will buy is a real production cost and the alternative is running broken ads on most of the inventory.
Interface elements cover parts of the frame. Text placed without accounting for safe areas is swallowed by the platform's own furniture.
Automatic cropping and automatic caption generation both exist and both are fallbacks: they prevent the worst outcome and do not produce a good one.
What the format is not for
Simple offers. A price and a deadline land faster in one frame.
Dense information. Specifications, comparisons and lists are read, not watched, and a viewer cannot pause a feed video to take notes.
Buying attention for its own sake. A campaign optimised toward views will find the cheapest available watching, which is autoplayed and unchosen. Views are a delivery metric and the outcome is somewhere else.
Producing it as a system
The accounts that sustain video treat one shoot as a source of many assets rather than as the production of one ad.
Shoot for cut-downs. A single session that yields a long version, several short versions, a set of alternative openings and a handful of stills is worth several sessions that each yield one finished piece.
Vary the opening, hold the body. The cheapest meaningful video test is the same footage with four different first seconds, and it isolates the part of the asset that decides how much of the audience exists.
Keep a record of angles. What argument each video makes, how it performed, against which audience. A library organised by asset is hard to reason about; one organised by argument is a record of what has been learned.
Revive rested assets. Recognition decays and audiences turn over, so a strong video returned after a break frequently performs close to its original level at no production cost.
Reading it
Three numbers in sequence, and each answers a question the others cannot.
The share of impressions that produced a stop says whether the opening worked. The share of those that stayed says whether the body repaid the attention. The cost per outcome says whether any of it sold anything.
Strong on the first and weak on the second is a promise the video did not keep. Weak on the first and strong on the second is the most actionable pattern there is — a new opening against working footage is a cheap edit with a large effect. Strong on both and weak on the outcome is a creative that entertains and does not argue.