Short-video hooks come in exactly five types, and there is no sixth. What decides the outcome is not how punchy the writing is, but whether you picked the right type — the wrong type cannot be edited into working later.
Five hooks, and the risk in each
| Type | One-line template | Fits | Risk |
|---|---|---|---|
| Claim first | “X is not actually Y”, “stop using X” | Content with a genuinely contrarian point | A weak claim reads as clickbait |
| Promised benefit | “Learn X in three minutes”, “this saves you X” | Tutorials, methods, tools | If the promise is not paid off, followers leave |
| Visual shock | Frame one shows the result, the finished object, the before/after | Crafts, makeovers, beauty, food | Nothing works if the image is not strong |
| Open question | “Why does X always Y?” | Explainers, debunks, trivia | Asks the viewer to invest comprehension first |
| Identity | “If you work in imports you know”, “anyone on 5k a month has lived this” | Niche audiences, emotional resonance | Cut the group too narrowly and reach caps out |
Why the third one fails most often
Visual shock looks like the easiest of the five, which is exactly why it fails: it does not depend on writing, it depends on having a result worth showing. If the finished object is ordinary, no cut, no music and no subtitle will rescue it.
It also fails quietly. A talking-head opening with a weak hook loses viewers, and you can see it in the retention curve. A visual opening with a weak image loses them at the same speed, but creators tend to blame the algorithm instead of the frame.
The fix is to shoot the result first and judge the frame on its own: cover the caption, mute the audio, and look at a single still for two seconds. If that still does not create a question, the hook is not there yet.
Three checks for any hook
- Is it understandable within three seconds? Play the opening to someone who has not seen the rest. If they cannot say what it is about, it failed.
- Is there an information gap? If the viewer thinks “I already knew that”, they scroll. A hook has to contain something they did not know.
- Is the first screen unobstructed? Subtitles sitting at the very bottom get covered by the interface, and the button column on the right of a vertical video cannot hold key information.
Common mistakes
- Stacking three hooks in one opening. A question, a promise and a twist at once means the viewer receives none of them.
- A hook that does not match the content. Opening with “three tricks” and delivering one costs both completion and account weight.
- Hooking in text only. If the picture is a static talking head, the viewer sees a person who is not moving; strong copy cannot hold that alone.
A hook solves “stay”. What solves “watch to the end” is structure — see the 30-second structure template.
Frequently asked questions
Can the five hook types be combined?
Yes, but keep one primary hook per video. “Visual shock plus claim first” is common: frame one shows the result, the first line states the contrarian claim. Three or more dilute each other.
Does the hook have to be spoken?
Not necessarily. Visual hooks can work on the image alone, with subtitles as support. If the content depends on an opinion, say it in the first line instead of making viewers wait.
How long is “three seconds” really?
It is the psychological moment a viewer decides to scroll, not a strict stopwatch. The first three seconds are the safe zone; every second after that raises the chance of leaving.
How do I know my hook works?
Two metrics: three-second view rate (how many survive the opening) and completion rate. A low three-second rate is a hook problem; a low completion rate is a structure problem.
Notice: posts are for studying public content and are not professional advice. Copyright stays with the original creators; commercial use requires permission. Results vary by category, account and platform — trust your own numbers.