How to Write a Hook for Short-Form Video

The first two seconds decide whether the other fifty get watched. Hook shapes that work, the openings that reliably fail, and how to find a hook inside footage you have already recorded.

Clipo AI September 8, 2026 10 min read

The opening frames of a short-form video with a bold text hook

Every retention graph for a short-form video has the same shape: a cliff in the first two seconds, then a gentle slope. Almost everything people try to fix — pacing, music, editing rhythm, captions — affects the gentle slope. The cliff is where the audience is actually lost, and the cliff is the hook.

This is good news, because it means the highest-leverage edit in any clip is the first sentence, and the first sentence is cheap to change. You can rewrite a hook in thirty seconds. You cannot re-record a video in thirty seconds.

What a hook is actually doing

A hook is not a clever line. It is an answer to a question the viewer is asking without words: is this for me, and is it going somewhere?

That question gets answered in roughly the time it takes to move a thumb. The viewer is not evaluating your video against some quality bar — they are comparing it to the next unknown video, which costs nothing to try. Your hook is competing with a shuffle.

Two things follow from that. First, ambiguity loses: a viewer who cannot tell what the video is about defaults to scrolling. Second, a hook that appeals to everyone appeals to nobody, because "is this for me" needs a yes, and a vague opening cannot produce one.

Five hook shapes that reliably work

There are not many. Most good hooks are one of these five, and knowing the shapes means you can write one on demand rather than waiting for inspiration.

1. The stakes-first claim

State the consequence before the topic. Not "today I want to talk about pricing" but "we raised our prices 40% and lost two customers — both of whom came back."

This works because it front-loads the reason to care. The topic is inferable; the outcome is not, and the outcome is what buys the next five seconds.

2. The contradiction

Name a thing your audience believes and immediately disagree with it. "Posting more often is the worst advice in short-form video" makes a viewer who posts often want to argue, and arguing requires watching.

The rule that keeps this honest: you have to actually deliver the disagreement. A contradiction hook on a video that then agrees with consensus is the fastest way to teach an audience not to trust your openings.

3. The specific number

Numbers do something adjectives cannot — they imply you measured. "We cut editing time from six hours to forty minutes" carries more weight than "we made editing much faster," and it costs the same number of words.

Specific beats large. "Eleven customer interviews" is more credible than "hundreds of conversations," because the small odd number sounds counted rather than claimed.

4. The named audience

Say who the video is for in the first three words. "If you edit your own podcast —" instantly qualifies the right viewer and dismisses the wrong one.

Losing the wrong viewer in second one is a win, not a cost. A viewer who leaves immediately barely dents your retention; a viewer who stays four seconds and then leaves tells the platform your video does not hold attention.

5. The open loop with a real payoff

Set up a question the video answers. "The reason your captions are hurting retention has nothing to do with the font" opens a loop, and the loop closes.

This is the shape most often abused. An open loop that never closes, or closes on something trivial, is a bait-and-switch — and it is punished faster in short-form than anywhere else, because the viewer can see the timeline and knows exactly when you ran out of video.

The openings that reliably fail

Most weak hooks are not bad ideas. They are good ideas with something useless placed in front of them.

Throat clearing — "so, I have been thinking about this for a while, and..." Every word before the substance is a word spent on nothing
Self-introduction — "hey guys, it's me again." The viewer does not know you yet, so this is information they cannot use
Context first — explaining the setup before the point. Context is what makes the point *land*, which means it belongs after the point, not before it
Announcing the format — "in this video I'm going to show you three things." Meta-description of a video is not a reason to watch the video
The generic question — "do you struggle with consistency?" Recognisable as a formula, and recognising the formula is enough to stop watching
A slow visual — a logo sting, a fade in, or two seconds of establishing shot. The audio may be perfect and nobody hears it, because the frame gave them nothing

Finding a hook in footage you already have

Most short-form video is not filmed for short-form. It is cut from a podcast, a webinar, a customer call or a recorded talk — and in that footage, the best hook is almost never at the start of the clip.

It is usually four minutes in, in the sentence where someone stops explaining and states a conclusion. People warm up before they say the sharp thing. The sharp thing is your hook.

The re-order edit

Take the strongest sentence in the clip and move it to the front, then let the original opening run as the explanation behind it. This one edit improves more clips than any other single change, and it requires no new footage.

It works because conversational speech is structured backwards for this medium: people build to a point, whereas short-form needs the point first and the build after.

Searching instead of scrubbing

The practical obstacle is finding those sentences across an hour of footage. Scrubbing a timeline is a bad way to do it — you are looking for a semantic thing ("the moment they disagreed with the standard advice") and scrubbing only shows you a waveform.

Searching the transcript by keyword, or describing the moment you want in plain language and letting the tool locate it, turns a forty-minute scrub into a thirty-second query. That is the difference between a repurposing habit that survives and one that quietly stops after three weeks.

Testing hooks without guessing

Opinions about hooks are cheap and mostly wrong, including yours and mine. The feedback loop is short enough that you should not need them.

Watch the first three seconds of the retention graph, not the average view duration. Average duration hides the cliff; the graph shows it
Compare 3-second view rate across posts, which isolates the hook from everything downstream of it
Publish the same clip twice with different openings on a platform where that is acceptable, a few weeks apart. The footage is identical, so the difference is attributable
Read your own drop-off timestamps out loud. If viewers leave at 0:04, the problem is the hook. If they leave at 0:20, the hook worked and the middle did not

A workflow you can repeat

Turning this into a habit matters more than any individual hook.

Find the strongest sentence in the source footage — the conclusion, not the setup
Put it first, and let the original context follow it as explanation
Cut everything before it, including the breath and the word "so"
Write the on-screen text separately from the spoken hook, so a muted viewer gets the same promise
Check the first frame in isolation — pause on it and ask whether it alone would stop you
Review the retention graph for the first three seconds before deciding whether the clip or the hook underperformed

Key takeaways

The first two seconds carry almost all of the drop-off — fix those before anything else.
A hook answers 'is this for me, and is it going somewhere', not 'is this clever'.
In repurposed footage the hook is usually buried mid-clip; move it to the front.
Losing the wrong viewer in second one is a win; losing them at second four is not.
Judge hooks on the 3-second view rate, not on how the line reads on paper.

Frequently asked questions

How long should a hook be?

One sentence, and ideally under two seconds spoken. If it takes longer than that to state, it is context rather than a hook — which usually means the real hook is the sentence after it.

Should the hook be spoken, on-screen text, or both?

Both, and they should not be identical. A large share of short-form viewing is muted, so on-screen text has to carry the promise on its own; when the spoken line and the text say slightly different things, the viewer gets two reasons to stay instead of one.

Do I need a new hook for every platform?

Usually yes for the wording, no for the footage. The same clip can open with a curiosity-driven line on TikTok and a specific, credentialed line on LinkedIn — the audiences reward different registers even when the content underneath is identical.

Why do my videos lose viewers at three seconds even with a strong opening line?

Check the frame rather than the script. A static talking head, a slow fade, or a first frame that looks like every other video in the feed will lose people regardless of what is being said. Movement, a legible caption, and a visually distinct first frame all buy time for the line to land.

Is it dishonest to move a sentence from the middle of a clip to the start?

Not if the meaning survives the move. Re-ordering someone's own words so the conclusion precedes the reasoning is standard editing; splicing two unrelated statements into a claim the speaker never made is not. The test is whether the speaker would agree the clip represents what they said.

How many hooks should I write per clip?

Three is enough to escape the obvious one. The first hook you write is almost always a summary of the topic, the second is usually the same idea with more words, and the third is where the actual angle turns up.

Do hooks matter for longer videos too?

Yes, but the mechanism is different. In short-form the hook competes with the next video; in a ten-minute video it competes with the viewer's own doubt about whether the video is worth ten minutes. Long-form hooks can spend a little more time earning that, but they still cannot start with throat clearing.

Keep reading

Find the strongest 30 seconds in an hour of footage

Clipo reads the transcript, surfaces the moments worth clipping, and reframes and captions them — so the hook you need is a search away rather than a scrub away.

Try Clipo Free