Short-form video removes the room to warm up. On a written page a reader will give you a paragraph to find the point. In a feed, the first two seconds are the whole negotiation, and if nothing in them promises something specific, the scroll continues and your video is over before it started.
This is why the same content can perform completely differently depending on the order it is delivered in. The information is identical; the sequencing decides whether anyone stays for it.
The structure below has three parts and fits almost any topic into thirty to sixty seconds. It is deliberately rigid, because a rigid structure is what lets you spend your attention on the hook instead of on the arrangement.
Why the first two seconds decide everything
Viewers are not evaluating whether your video is good. They are evaluating whether it is for them. Those are different judgements, and the second one happens much faster.
This is why a hook that describes the audience outperforms a hook that describes the topic. "How to fix a leaking tap" is a topic. "If your tap drips at night and you can hear it from the bedroom, this takes four minutes to fix" tells a specific person that this video is about them.
The practical test: read your opening line and ask whether a stranger could accurately decide whether they are the intended viewer. If they would have to watch longer to find out, the hook has not done its job.
The three-part skeleton
Every short script you write can be built from three parts, in this order:
- The hook — names the viewer and the payoff, in one or two sentences.
- The payoff — delivers the thing the hook promised, with no detour.
- The exit — tells the viewer what to do next, or closes the loop cleanly.
The mistake nearly everyone makes is inserting a fourth part: an introduction. "Hey everyone, welcome back to the channel, today we're going to talk about…" This is a written-article habit transplanted into a format that cannot support it. In short-form, the introduction is the hook, or it does not exist.
Writing a hook that is specific
There are a handful of hook shapes that work repeatedly, and none of them depend on hype:
- The situation. Describe a moment the viewer recognises. "You've written three drafts and it still sounds like a robot wrote it."
- The correction. Name a belief you are about to overturn. "Most people fix this in the wrong order, and that's why it never sticks."
- The stake. State what is at risk. "Get this wrong and you have to redo the whole edit."
- The constraint. Commit to something narrow. "Here is the shortest version of this that still works."
What all four share is that they make a claim the rest of the video can pay off. A hook that cannot be paid off — "you won't believe what happens next" — creates a promise the video does not keep, and the drop-off shows up at the two-second mark.
Write five hooks for the same video. Keep the one with the most concrete noun in it. That is usually the one that survived editing pressure, because vague hooks look acceptable in isolation and only fail when compared.
The payoff section
The payoff has one job: deliver what the hook promised as early as possible. If the viewer was told they would learn a four-minute fix, the fix starts at second three, not at second twenty.
Two rules keep this section tight. First, cut every sentence that establishes context you do not need. The viewer arrived knowing what they searched for. Second, cut every sentence that is interesting but not necessary — tangents feel like value while you are writing them and read as delay when you watch them.
Ordering inside the payoff matters too. Lead with the step that produces a visible result, then the steps that refine it. A viewer who sees something change in the first ten seconds is much more likely to stay for the details.
Timing the beats
A sixty-second script divides cleanly. Note that this is a writing budget, not a rule the viewer perceives.
| Beat | Time | Job |
|---|---|---|
| Hook | 0–3s | Name the viewer and the payoff |
| Setup | 3–8s | State the one condition that matters |
| Steps | 8–45s | Deliver the payoff in two or three moves |
| Result | 45–53s | Show or state what changed |
| Exit | 53–60s | Next action or clean close |
A thirty-second version collapses Setup into the Hook and reduces Steps to two. The proportions stay roughly the same, which is why the same script can be cut down without rewriting it from nothing.
Calls to action that fit
A call to action that asks for a big commitment from someone who just met you performs badly, and it also weakens the video by association. Match the ask to the relationship the video has actually established.
For a first-time viewer, the only ask that reliably works is another small piece of value: a related video, a saved post, a follow if the topic is genuinely recurring. For someone who has watched before, a larger ask is reasonable.
The other option is to skip the ask entirely and close the loop instead — finish by answering the question the hook raised, and stop. A clean ending with no request is not a wasted opportunity; it is a reason to trust the next one.
One topic per video
The most common reason a script runs long and loses people in the middle is that it is carrying two ideas. Both feel necessary at the outline stage, and combining them always produces something worse than either would have been alone.
When you find two ideas in one script, split them and give each its own hook. Two videos with distinct hooks will each find their audience. One video trying to serve both audiences will be recommended to neither, because the platform cannot tell who it is for.
Checking the script before you shoot
Read the script out loud at normal speaking pace and time it. Scripts reliably read shorter than they speak, and the overrun always comes out of the payoff, which is the part that mattered.
Then apply three checks. Does the first sentence let a stranger know whether they are the viewer? Does the payoff arrive before the halfway point? Could the second half be deleted without losing the point — and if so, should it be?
A script that passes those three is ready to shoot. The performance will still decide the result, but the performance cannot rescue a structure that gives the viewer no reason to reach the second sentence.