How to write a hook that stops the scroll

Every piece of advice about hooks says the same three things: the first three seconds matter, be attention-grabbing, use a pattern interrupt. All of it is true and none of it helps, because it describes the effect rather than the mechanism.

Here is the mechanism. A viewer scrolling is in a low-effort state, making a continuous, barely-conscious decision to keep going. Something stops that decision only when it creates a small piece of unfinished business: a question they now have, a claim they want to check, a thing they can see but not yet explain. Attention is not grabbed. It is loaned, against the promise that the loop will close.

That single reframe explains why most hooks fail, and it makes them fixable.

The four ways a hook fails

It closes the loop instead of opening one

This is by far the most common failure, and it is invisible if you are thinking in terms of “strong opening lines”, because the sentence often sounds confident and well-written.

It is not specific enough to be interesting

Generality is the enemy. A hook aimed at everybody addresses nobody, and the paradox of short video is that the narrower the opening, the wider the reach, because a narrow line makes the people it does apply to stop dead.

It writes a cheque the video does not cash

A hook is a loan. Overpromise and the viewer stays for four seconds, realises the payoff is not coming, and leaves, which is worse for you than if they had never stopped: you have now trained the platform that people who see your video abandon it. Clickbait does not just fail to work, it costs.

It exists only in the audio

A significant share of first views happen with the sound off. If your hook is entirely spoken, those viewers see a person moving their mouth and nothing else. Put the hook on the screen as text as well, in as few words as you can manage, and treat that as a hard requirement rather than an accessibility nicety.

Write the hook before the video

The order most people work in is: have an idea, film it, then try to find a hook in the footage. This is backwards and it is why so many hooks feel bolted on.

Write the hook first, as one sentence, before anything else exists. If you cannot make that sentence interesting, the idea is not ready. That is not a failure, it is the cheapest possible discovery: you have saved yourself a filming session and an edit. Change the angle until the first line works, and only then make the video.

A useful discipline is to write five hooks for every idea and keep the fifth. The first two are the obvious ones, the third is a variation on them, and the fourth and fifth are where anything surprising happens.

Ten shapes worth knowing

Templates, not sentences: a finished hook about skincare is no use to somebody filming about woodwork, but the shape underneath it transfers to anything. The full set, with a worked example each and a note on where each one backfires, lives in the hook library. In summary:

• The curiosity gap. Name something specific, withhold one piece of it.
• The direct callout. Address one narrow group and let everyone else go.
• The contrarian open. Contradict something the viewer believes.
• Result first. Show the finished state before asking for any attention.
• The problem open. Say the thing they have felt and never articulated.
• The counted promise. A number tells them how much time is being asked for.
• Starting mid-story. Skip the setup entirely and orient them afterwards.
• The silent demonstration. No spoken hook. Something unresolved happening in frame one.
• The real question. Only when the question is genuinely hard to answer.
• The stake. Attach a cost to not watching, and only when the cost is real.

Length, and the myth of three seconds

The received wisdom is that you have three seconds. In practice you have considerably less than that to stop the scroll, and considerably more than that to justify it. Those are two different jobs and conflating them produces frantic openings that exhaust the viewer.

1. Second zero to one: stop the scroll. Text on screen, motion, a face already speaking, something visually unresolved.
2. Second one to four: open the loop properly. This is where the actual sentence lands.
3. Second four to eight: give something back. A partial payoff, a first item, evidence for the claim. This is where most videos lose people, because they are still setting up.

Testing hooks without guessing

The cheapest test is the one described above: read it aloud with the sound off in your head, and ask what is unresolved. The second cheapest is to write the hook, leave it for a day and read it cold. Almost every weak hook is obviously weak twenty-four hours later.

Beyond that, you can only learn from your own footage. Keep the first lines of your last twenty videos in a single list next to their retention at three seconds. The pattern usually appears inside a month, and it is usually not the one you expected.

Viral Pop does the same thing from the other direction: it scores the hook on any video you paste, tells you specifically what the opening is doing, and generates alternative hooks for an idea before you have filmed it. But the list of your own twenty first lines costs nothing and is nearly as good.