Ad hooks live or die in about three seconds. Hold rate at that mark predicts performance better than almost any other early metric. If people do not stay, nothing else in the ad matters — the offer, the proof and the call to action all sit behind a door nobody opened.
Here is what opens it.
Ad hooks that work
The mid-action open
Start in the middle of doing the thing. No logo, no introduction, no hi guys. A hand already applying the product, a screen already showing the result. Introductions are where scrolls happen.
The number that sounds wrong
"We turned off 80% of the ads and sales went up." A claim that creates a question is a claim people stay for. It must be true, which is the constraint that makes it work.
The named objection
"I thought this was another overpriced serum." Saying the doubt out loud signals that you are not about to pretend it does not exist.
The specific person
"If you run a clinic in Delhi and your consults cost more than ₹400 —" Narrow beats broad. The people it excludes were not going to buy anyway, and the people it names lean in.
The unfinished sentence
Starting mid-thought creates a small gap the brain wants closed. "...and that's the part nobody tells you about COD orders." Used sparingly, it is very effective. Used constantly, it becomes a tic.
The visual contradiction
Something on screen that does not match what you expect — a product used in an unusual place, an unexpected object. Stops the scroll physically rather than verbally, which matters because most people are watching on mute.
You are not selling in the first three seconds. You are buying the next ten.
Hooks to retire
- "Hi guys, welcome back to my channel." Feed content has no channel and no back.
- "Are you struggling with..." Overused, and in most markets it is also a personal-attributes policy risk.
- The logo sting. Three seconds of branding at the top of a feed ad is three seconds of nothing.
- "In this video I will..." Just do the thing.
- Slow establishing shots. Beautiful, and nobody is there by the time the point arrives.
Practical rules
- Captions burned in. Assume mute. The hook has to work as text.
- The hook is a separate test. Same body, four different openings — that is a clean test and it isolates the variable that matters most.
- Write ten, use three. Hooks are cheap to generate and expensive to be precious about.
- Pull language from reviews and comments. The best hooks are usually things a customer already said. Mining that material is the part that cannot be automated.
How to judge them
Look at 3-second and 15-second hold rates separately from conversions. A high 3-second and a collapse by 15 means the hook wrote a cheque the ad could not cash — which is a different problem from a weak opening, and it needs a different fix.
How many new concepts a month your spend actually needs is arithmetic — the creative volume calculator works it out.
Questions people ask
How long do I have to hook someone?
Roughly three seconds on Meta and under two on Reels, where the thumb is already moving. On YouTube in-stream you have five before the skip button, which is a different problem: you are competing with a button, not with the next post.
What hold rate is good?
On Meta, a three-second hold rate above 30 percent of impressions is healthy for prospecting, and above 20 percent is workable. Below 15 percent the opening is the problem, and no amount of budget or audience work will fix it.
Should the hook state the offer?
Not usually. An offer stated first reads as an ad and gets filtered before it is processed. Earn two more seconds with a situation the viewer recognises, then name the offer while they are still watching.
How many ad hooks should I test?
Cut three to five openings onto the same body footage and let them run against the same audience. Hold rate separates them within a few thousand impressions, which is far cheaper than judging whole ads against each other.
Want this tested properly on your account? Tell us what you sell.