Where people decide to stay

The shape of the curve

Every platform that shows you a retention graph shows the same shape: a steep drop in the first few seconds, then a gentler slope, then a long tail. The size of that first drop varies enormously between pieces, and it is the single number most under an editor's control.

The mechanism is not mysterious. A viewer arriving at a video is answering one question — what is this, and is it for me? — and they will answer it whether or not you have given them the information. If the opening does not answer it, they answer it themselves, usually with "no".

What an opening has to do

Three things, in whatever order the material allows:

1. Say what this is. Not the title. The subject, in the first sentence or the first image.
2. Say why it is worth the next few minutes. A question, a problem, a surprising claim, a promise of a specific thing.
3. Establish the register. Serious or funny, formal or personal, fast or slow. People decide whether they want to be in a room with you very quickly.

A piece that opens with a logo animation, a slow fade up from black, thirty seconds of establishing drone footage and a greeting has done none of the three by the time most of its audience has left.

The cold open

The most reliable structure: start with the strongest moment in the piece, then go back and begin properly.

In practice, take the best twelve seconds from anywhere in your material, put them at the front, and cut to your actual opening. The moment does not need explaining — it works because it is interesting, and the explanation arrives in the next thirty seconds.

Two cautions. The cold open must be from the actual piece and must actually be paid off; using your best moment as bait and then delivering something else is the mechanism of clickbait, and it trains an audience to distrust you. And it must be a moment, not a summary — a summary at the front removes the reason to watch.

The first frame is a thumbnail

Whatever your first frame is, it will be shown as a still somewhere — in a feed, in a preview, as an autoplay poster. A black frame or a fade-from-black gives the platform nothing to show.

Start on a real image. If you want a fade, fade in over the first shot rather than from black into it.

Titles are not an opening

A full-screen title card at the start costs you three to five seconds of the most expensive attention you will ever have, and it tells the viewer something the platform already told them.

If the piece needs a title, put it over the first shot, or a few seconds in once something has happened. The BBC, YouTube's most-watched creators and most feature films all do this. A title sequence at the front is a convention from an era when people had already chosen to sit down.

The specific numbers, honestly

Platform advice circulates as hard rules — "the first three seconds decide everything" — and the honest position is that the numbers move, differ by platform, and are mostly inferred from creator-side dashboards rather than published. What is robust across all of them:

• The drop-off is steepest at the very start and flattens after roughly the first tenth of the piece.
• Openings that state the subject retain better than openings that delay it.
• Retention is measured against whatever the algorithm is optimising this quarter, which changes.

So treat the curve as feedback on your own work rather than as a law. The useful practice is to look at your own retention graph, find the moment of the sharpest drop, and watch the ten seconds before it. That is where you lost them, and it is usually something identifiable: a long pause, a wandering sentence, a section that had not earned its place.

The rest of the piece has openings too

Every section transition is a smaller version of the same decision. When you move from one subject to the next, the viewer re-asks whether to stay. Give each section its own small hook — a question, a surprising image, a change of pace — rather than a flat announcement of what is coming next.

The signposting version ("now let's look at the three main reasons") is worth avoiding for the same reason as the summary: it tells people what they are going to get, which is precisely the information that makes staying optional.

The honest tension

There is a real conflict between retention craft and honesty, and it is worth naming rather than pretending it is not there. Every technique in this lesson can be used to make a piece more accessible to a viewer who would genuinely have valued it, and every one can be used to hold somebody's attention on something that will not repay it.

The distinction is whether the promise gets paid. A cold open that delivers is craft. A cold open that does not is bait, and the fact that it works is not a justification.

Today

Open your last published piece and watch the first fifteen seconds as though somebody sent you the link. Write down what you knew at second three, second eight and second fifteen. If the answer at second eight is still nothing, you have found the edit.