How long is a video hook?

Three seconds, and realistically the first one carries most of the weight. That is the window in which a viewer decides whether to keep watching, and it is shorter than most brand videos take to finish showing their logo.

What makes a good video hook?

Doing three things at once: interrupting the scroll with something unexpected, signalling within a beat that the video is for this specific viewer, and setting up a question the rest of the video answers. A hook that does two of the three still fails.

Should a video start with the logo?

No. A viewer with no relationship to the brand has no reason to watch its logo animate, and those two seconds are the most valuable in the video. Put the logo at the end, where it reads as a signature rather than a toll.

How do I test which video hook works best?

Build the body once, produce three to five different three-second openings against it, and run them as separate creatives with identical targeting and budget. Compare three-second hold rate, that is the hook's own metric, before anything downstream affects it.

The first three seconds: how video hooks actually work

You are not competing with other ads. You are competing with the thumb. Everything a video needs to earn happens before most brands have finished showing their logo.

Fresvo2 September 20265 min read

A close-up of an analogue watch face marking seconds

Every performance report on video creative reaches the same conclusion by a different route: the opening seconds decide the rest. Not because attention spans collapsed, but because the alternative to your video is always one flick away and costs nothing.

So the first three seconds are not an introduction. They are an audition, and most brand videos spend them on a logo, an establishing shot, or a slow fade, three things that are all, in a feed, indistinguishable from nothing happening.

What a hook has to do

Three jobs, simultaneously, in about seventy-five frames.

Interrupt. Give the eye something it did not expect. Motion, a face, an odd object, a hard cut into the middle of something.
Address. Make it obvious within a beat that this is for a person like the viewer. Not everyone, them.
Promise. Set up a question the rest of the video answers. Without one there is no reason for a fourth second.

A hook doing two of the three still fails. Interruption without a promise gets a glance and a scroll; a promise without interruption is never seen at all.

Nine patterns that still work

Start mid-action. Open on the second half of a gesture. The brain fills in the first half and is already invested.
State the objection. Begin with the thing the viewer is about to think. This is more expensive than it looks, it takes a lot of honesty.
Show the outcome first. The finished thing, the result, the after. Then rewind to how.
A specific number. Not most companies. Fourteen minutes, or four thousand euros in two days. Precision reads as knowledge.
The unexpected object. Something in frame that does not belong there, resolved in the next beat.
Direct address with a qualifier. If you run a construction firm and have ever waited nine weeks for a render, this is for you. It excludes people, which is what makes it land for the rest.
A visible mistake. Something going wrong is far more watchable than something going well, and it buys enormous credibility when you then fix it.
The comparison cut. Two things side by side in the first frame. The eye does the work.
Text that is not a title. One line of type stating the stake, over motion. Works with the sound off, which is the default state.

A thumb scrolling on a phone screen — The competition is not the other ad. It is the flick that costs the viewer nothing.Photo, Vojtech Bruzek

Four openings that no longer earn their place

The logo sting. A viewer with no relationship to the brand has no reason to watch its logo animate. Put it at the end, where it is a signature rather than a barrier.
The drone establishing shot. Beautiful, and it says nothing. It has also become its own genre signal, a viewer knows a corporate video is starting.
The slow fade from black. Half a second of nothing, in the half second that decides everything.
The abstract problem statement. In today's fast-moving world. It is the audio equivalent of a stock photo.

Sound off is the default, not the exception

Assume no audio. Most feeds start muted, most desktop viewing is muted out of politeness, and a hook that depends on a line of voiceover is a hook that did not happen.

Anything essential in the first three seconds needs a visual or typographic form.
Captions belong from frame one, not from the first word of narration.
Style captions to the brand. Platform defaults are identical across you and every competitor.
If the hook is a line of dialogue, put it on screen as well as in the mix.

How to test hooks without remaking the video

This is the practical part, and the reason hook work has such good economics: the hook is separable from the rest of the film.

Build the body once, the argument, the proof, the call to action.
Produce three to five different openings, three seconds each, against the same body.
Run them as separate creatives with identical targeting and budget.
Compare three-second hold rate first. That is the hook's own metric, uncontaminated by everything downstream.
Keep the winner, then generate three new challengers against it. Hook testing does not have an end state.

Three-second openings are cheap to produce and, increasingly, cheap to generate, which is the most defensible use of AI in ad production available right now.

A film clapperboard held up at the start of a take — One body, five openings. The hook is the only part of a video worth testing at volume.Photo, Jakob Owens

The metric to watch, and the one to ignore

Metric	What it tells you	Use it for
3-second hold rate	Whether the hook interrupted anyone.	Judging hooks. This is the number.
25% completion	Whether the promise was worth staying for.	Judging the body, once the hook works.
Full completion	Mostly length. Shorter videos always win it.	Comparing cuts of the same length only.
View count	How much you spent.	Nothing. It is a budget report wearing a performance costume.

Fix hold rate before touching anything else. A video with a weak hook and a brilliant body is a brilliant body nobody reached, and measuring what actually matters starts from exactly that principle.

Common questions

How long is a video hook?: Three seconds, and realistically the first one carries most of the weight. That is the window in which a viewer decides whether to keep watching, and it is shorter than most brand videos take to finish showing their logo.
What makes a good video hook?: Doing three things at once: interrupting the scroll with something unexpected, signalling within a beat that the video is for this specific viewer, and setting up a question the rest of the video answers. A hook that does two of the three still fails.
Should a video start with the logo?: No. A viewer with no relationship to the brand has no reason to watch its logo animate, and those two seconds are the most valuable in the video. Put the logo at the end, where it reads as a signature rather than a toll.
How do I test which video hook works best?: Build the body once, produce three to five different three-second openings against it, and run them as separate creatives with identical targeting and budget. Compare three-second hold rate, that is the hook's own metric, before anything downstream affects it.