Most advice about hooks is a list of openers. Start with a question. Start with a number. Start mid-sentence. The lists are not wrong, exactly, but they describe what good hooks look like rather than what a hook is for, which makes them impossible to apply to a topic the list did not anticipate.
A hook has one job: make the next second feel necessary. Not interesting, not clever, not surprising. Necessary. Everything else follows from that, including most of the openers on the lists.
Why the distinction matters
“Grab attention” is advice you cannot act on, because attention is already grabbed — the video is playing, full screen, with sound. The viewer is not deciding whether to look. They are deciding whether to keep looking, and they are making that decision continuously, not once.
That reframing has a practical consequence. If the job is to grab attention, you optimise the first frame. If the job is to make the next second necessary, you optimise the gap — the distance between what the viewer now knows and what they want to know. A hook works when it opens a gap the viewer would rather close than scroll away from.
A hook is not the first thing you say. It is the reason the second thing is worth waiting for.
What the viewer is actually deciding
In the opening moments a viewer resolves three things in order, and a hook can fail at any one of them:
- Is this for me? Category recognition. If the viewer cannot place the video in a world they care about, nothing later rescues it.
- Is something going to happen? Promise. There has to be a payoff implied, even a small one.
- Is it worth the wait? Proportion. The implied payoff has to look bigger than the implied wait.
Most weak hooks pass the first test and fail the second. They establish the topic and then keep establishing it. The viewer knows what the video is about and has no reason to believe anything is coming.
Why this connects to distribution, not just taste
This is not only an aesthetic argument. Instagram has been explicit that how quickly and how much people engage with a post feeds its ranking, and watch time has been reported as the heaviest signal across surfaces since Mosseri's remarks in early 2025. A hook that fails is not just unwatched — it suppresses the distribution of everything behind it, including the part that was good.
That is the asymmetry worth internalising. The body of a video can only be seen by people the hook already kept. Improving the body improves the experience of a shrinking group. Improving the hook improves the size of the group.
The test
Write the hook, then ask a question that has nothing to do with whether it sounds good: after this sentence, what does the viewer want to know? If you cannot name the question, the hook has not opened a gap and the video starts from a standstill.
If you can name it, the second test is whether the video answers that exact question rather than a nearby one. A hook that opens a gap the video does not close is worse than a flat opening, because it trains the audience that your promises are decorative.
Where the lists come from
Read the standard openers again with the gap in mind and they stop being arbitrary. A question opens a gap explicitly. A number implies a list whose end is not yet known. Starting mid-sentence implies a beginning the viewer missed. A contradiction opens a gap between what the viewer believes and what was just said.
They are all the same move. Once you can see the move, you can invent openers the lists never mention, which is the point — a hook copied wholesale from a list arrives in a feed alongside everyone else who read it.
For what to do once the gap is open, see the four ways hooks fail. For how we decide whether a hook actually worked rather than merely felt good, see how we measure a breakout.