The Hook: The First Two Seconds Decide The Rest

The short version

The hook is whatever stops someone from scrolling past. In a feed, the default is to keep moving, so the hook is not an introduction — it is the entire reason anything else in the ad gets seen.

Most video ads fail here, and the failure is invisible downstream. A video nobody watched still produces a full set of conversion metrics, which look like an offer problem or a landing page problem. They are not.

Hook, angle, and offer

These get conflated and they are different things.

The angle is the argument — which reason to buy you are leading with.
The hook is the entry point — how you get someone to hear the argument.
The offer is what you are actually proposing.

One angle supports many hooks. A hook that stops the scroll but has nothing to do with the angle produces a watched ad that persuades nobody, which is a specific and common waste.

What actually stops a scroll

Motion or visual disruption in the first frame. The literal first frame, not the first second.

A specific claim. Specificity reads as real. "Cut our CPA by 40%" stops more people than "improve your marketing."

A recognizable situation. The viewer seeing their own problem stated.

A question with an uncomfortable answer. Works when it is genuinely uncomfortable, not when it is rhetorical.

Something visually unresolved. A state that demands explanation.

A face and direct address. Reliable, which is also why it is everywhere and fatigues quickly.

What does not

A logo. It reads as "this is an ad" and gives permission to leave.
A slow build. There is no time for one.
Generic aspiration. Attractive people being happy is invisible through repetition.
A question everyone answers no to. "Want to grow your business?" reads as an ad.
Anything requiring sound. Most feed video is watched silently, so a spoken hook with a static frame has no hook.

Testing hooks properly

Hooks are the cheapest thing to test because you can change the first two seconds and leave the rest of the video alone.

Change only the hook. Same body, same offer, several openings. Anything else confounds it.
Read hook rate, not conversion. Conversion is too far downstream and too slow to isolate the opening.
Test more than you think reasonable. Hook performance is hard to predict, and the winner is regularly one nobody expected.
Retire on schedule. Hooks fatigue faster than offers. A hook that worked for a quarter is probably done.

The diagnostic that matters

If conversion is weak, check hook rate before touching the landing page.

A weak hook means only a small self-selected group saw your argument, and their conversion rate is not measuring your offer — it is measuring the tolerance of people who stayed through a bad opening. Fixing the landing page will not move it.

Practical rule

Treat the first two seconds as a separate creative asset with its own testing cycle. Diagnose from the top of the funnel down, because a hook failure disguises itself as every other kind of failure.