Writing hooks that stop the scroll: the first three seconds

On a feed, your ad is not competing with other ads. It is competing with the thumb. Before anyone weighs your offer, your product, or your proof, they make a much faster decision (keep watching or keep scrolling) and they make it in about the time it takes to read this sentence. The hook is whatever earns that pause. Everything else in the ad only matters if the hook works first.

This is easy to say and hard to act on, because most of the craft and most of the budget in a video ad goes into the parts almost nobody sees. The script, the payoff, the call to action: all of it is downstream of a decision the viewer already made in the first second. What is changing now is not that hooks matter; they always have. It is that AI makes it cheap to test many of them, which exposes how little most teams actually knew about which of their hooks worked.

What a hook actually buys you

It helps to be precise about what a hook is for. A hook does not sell anything. Its only job is to buy the next three seconds: to convert a passive scroller into someone who has decided, however provisionally, to keep watching. That is the entire transaction. A great hook on a weak ad gets your weak ad watched; a weak hook on a great ad means the great ad never gets a chance.

This reframes a lot of bad instincts. Teams treat the hook as the place to be clever, or on-brand, or comprehensive, cramming the value proposition into the opening so nobody misses it. But cramming is the opposite of hooking. The opening that tries to say everything says nothing fast enough to register. The hook's job is narrow: create enough tension, curiosity, or recognition that stopping feels more rewarding than scrolling.

Why the first three seconds carry the rest

There are two audiences for your first three seconds, and both judge it before the fourth. The first is the person. The second is the delivery algorithm, which reads early retention as a signal of quality and decides how much more it will spend showing your ad to anyone. A weak hook does not just lose the viewers who scroll past; it tells the platform your creative is mediocre, and your distribution shrinks accordingly. The penalty compounds: fewer impressions, higher costs, and a winner you might have had quietly starved of reach.

This is why the first three seconds carry disproportionate weight. They are the only part of the ad that every impression actually experiences. Spend on the back half of a clip most people never reach and you are optimising a room nobody walks into.

The opening is the only three seconds of your ad that one hundred percent of the audience watches. Every other second is something you have to earn. And the hook is how you earn it.

The kinds of hooks that earn attention

There is no single winning hook, and any list of "proven" openers becomes a cliché the moment everyone copies it. But the mechanisms underneath the good ones are stable, even as the surface execution ages. A few that reliably create the pause:

• The pattern interrupt. Something visually or tonally unexpected in the first frame: an odd object, an abrupt motion, a statement that contradicts what the feed has trained the viewer to expect. It works by breaking the scroll's autopilot.
• The stated stakes. Naming the exact problem or outcome the viewer cares about, fast and specifically, so the right person feels addressed and the wrong person scrolls on. Precise beats broad: "if your ads stopped converting last month" outpulls "grow your business."
• The open loop. Posing a question or starting a story whose ending is withheld, so continuing feels like the only way to resolve the tension you just created. The cost of scrolling becomes the cost of not knowing.

The point is not to memorise these. It is to recognise that a hook works through a mechanism, not a magic phrase, and that knowing the mechanism is what lets you generate ten variations of an idea instead of one.

Testing hooks instead of guessing them

For most of performance marketing's history, the hook was the part you guessed. You shot one opening, maybe two if the budget allowed, and you committed. The expense of production made the hook a single bet placed before any data existed, which meant the most important three seconds of the ad were chosen by the people least able to know what would work: the team, in a room, before launch.

Cheap AI variation breaks that constraint. When generating a new opening costs almost nothing, the hook stops being a bet and becomes a test. You take one proven body and run it behind eight different openings (a pattern interrupt, a stakes statement, an open loop, a customer line, a bold claim) and you let early retention tell you which mechanism this audience responds to. The body stays constant, so the only variable is the hook, and the results are clean enough to learn from.

What teams discover when they do this is usually humbling. The opening they were sure about underperforms a throwaway alternative they almost did not make. The hook that felt too simple wins. This is not a failure of taste; it is the normal result of moving a decision from opinion to evidence. The value of cheap variation is not that it generates more hooks. It is that it replaces the guess with a measurement.

The judgment AI does not replace

None of this removes the need for judgment. It relocates it. AI will generate a hundred openings, and most of them will be noise. The skill that still matters is knowing which mechanisms are worth testing, reading early retention without overreacting to a small sample, and recognising when a hook is winning for the wrong reason, pulling stops from people who will never buy and inflating a vanity metric while the actual conversion rate sits flat.

A hook that stops the scroll but attracts the wrong audience is not a good hook; it is an expensive one. The teams that win the first three seconds are not the ones generating the most openings. They are the ones who understand what a hook is for, test the mechanism rather than the phrase, and read the results closely enough to tell a real winner from a loud one.

• Meta, "The value of the first few seconds: mobile video and attention," Meta for Business insights, 2025.
• Nielsen, "Creative attention and the role of the opening seconds in video advertising," 2024.
• WARC, "What we know about effective video ad openings," 2024.