The first three seconds of a short video are not a summary. They are a decision point. Someone scrolling has to choose between continuing and moving on, and they make that choice before knowing what the video is about — which means the opening's job is not to describe the topic but to make the next three seconds feel worth spending.
Most scripts treat the opening as an introduction, because that is what writing usually does: set up, then deliver. Short video inverts the order. The delivery has to start immediately, and the setup becomes optional material you can drop in later, or leave out.
What three seconds really means
Three seconds is roughly the window before a platform's next content loads and the viewer's thumb is free again. It is not a stylistic guideline; it is a mechanical constraint imposed by how the feed behaves.
Within that window, the viewer needs one thing: a reason to believe the next thirty seconds will be different from the last thirty they scrolled past. They do not need the topic, they do not need context, and they certainly do not need a greeting. They need a promise or a tension, delivered in a way that does not require effort to parse.
This is why the practical test is not "does the opening explain the video" but "would someone who had no intention of watching this stop anyway?" If the answer is no, the opening is furniture.
Why the usual hook fails
Four openings account for most of the failures. Each one fails for a structural reason rather than a writing one.
- The greeting. "Hey everyone, welcome back." It spends the only three seconds you had on something the viewer already knows, and it signals that the useful part is later.
- The topic announcement. "Today I'm going to talk about internal linking." Accurate, and it asks the viewer to decide whether a topic interests them before you have said anything interesting about it.
- The long setup. Two sentences of context before the first claim. Context is valuable after the viewer has a reason to care, and worthless before.
- The vague promise. "You won't believe what happened next." It creates suspicion rather than curiosity, because there is no subject to be curious about.
Notice that all four are trying to be polite or clear. Both are virtues in an article and liabilities in a feed, where the cost of being slow is that nobody finds out you were eventually clear.
Four openings that hold
The openings that work tend to fall into four shapes, and the choice between them is a choice about what kind of tension you can honestly create.
State the conclusion first. Give the answer in the first sentence and let the rest of the video be the reasoning. This works when the answer is genuinely surprising or useful, because the viewer stays to find out whether they agree.
Name the problem in the viewer's words. Not "efficiency challenges in content operations" but "your drafts come back sounding like a robot." Recognition is a strong reason to keep watching, and it is the easiest of the four to write honestly.
Show the result. Open with the finished thing — the design, the number, the before-and-after — and then explain how it was made. This is the most reliable shape for anything visual, because the payoff is visible immediately.
Break an expectation. State something the viewer assumes is true, and contradict it in the next breath. Powerful when the contradiction is real; disastrous when it is a trick and the video does not deliver.
Write the hook last
The opening is the hardest sentence in the script because it has to carry the whole piece's promise in a few words. Write it after the rest, when you know exactly what the video delivers and which detail is the most interesting thing in it.
That most-interesting detail is usually the hook. When you finish a draft, mark the sentence that made you smile or that you would quote if you had to summarise the whole thing. Move it to the front, and rewrite the two sentences after it so they do not simply repeat it.
Writers resist this because the draft then reads out of order. It does read out of order — that is what the format requires, and the piece is not read, it is watched.
The visual half
Roughly half of an opening is not words. What is on screen in the first second does more work than the sentence does, because it is processed before the sentence finishes.
So the visual should carry the same promise as the line. If the hook is "this is what your export looks like before you fix it," the frame should show the broken thing, not the presenter. If the hook is a result, show the result. A talking head delivering a strong line to a static background is throwing away half the available attention.
Movement helps, but only if the movement is the content rather than decoration. A cut, a zoom to a detail, or a hand placing something in frame all read as "something is happening." An animated arrow does not, because viewers have learned to ignore it.
Matching hook to payoff
The opening makes a promise; the video has to keep it. Most scripts fail this test in one specific way: the hook implies a level of specificity the body does not reach. "The one setting that changed everything" promises a named setting. If the body delivers a general principle, the viewer notices, and the cost is not this video — it is the next one, which they will not click.
The fix is to write the hook with the payoff in view and to downgrade the hook if necessary. A modest promise kept produces more returning viewers than a large promise broken, because the platform's audience and a human audience measure the same thing: whether the ending was worth the beginning.
If the payoff is genuinely good, say less about it in the hook, not more. Specificity without exaggeration — "the three fields I fill in before any AI draft" — creates a concrete expectation you can meet.
Testing openings cheaply
Do not shoot two full videos to test a hook. Write four first lines for the same script, put each one at the top of the same short clip, and see which one gets watched past the three-second mark. Same footage, four openings, one hour of work.
Judge on retention at three seconds and again at thirty. A hook that wins at three seconds and loses at thirty is manufacturing curiosity it cannot satisfy, which is worse than a slow start. The pair of numbers is what makes the result actionable; either one alone is misleading.
Keep a note of the hook that won and why, in one line. After ten videos, that note is the most useful document your channel has, because it is specific to your audience rather than to general advice about openings.
When the hook misreads
Sometimes the hook works — the retention is fine — but the comments tell you it was understood as a different video than the one you made. That is not a hook problem, it is an honesty problem: the opening selected for an audience with different expectations than the body serves.
Fix it by making the opening more specific, not by making the body more generic. Vague openings attract whoever the words happen to suggest, and the narrower your opening is, the more likely the people who stay are the ones the rest of the video was written for. A smaller, correct audience beats a larger, disappointed one on every measure that matters after the first three seconds.