How to Write Facebook Ad Hooks

You cut a new video ad, drop it in, and three days later the numbers are grim. Same offer that worked last quarter, same landing page, but almost nobody is watching past the first frame. Before you blame the product or the audience, look at one column: how many people made it three seconds in. That number is your hook doing its job, or not doing it, and it is the cheapest thing in the whole ad to fix.

Short answer: A Facebook ad hook is the first three seconds of the ad, and hook rate (3-second views divided by impressions) is the one metric that grades it in isolation. Write it for one specific buyer's pain in their own words, then read the hook rate against your account's baseline. Below baseline means the opener died, not the offer.

The takeaways

• Hook rate isolates the opener: 3-second views over impressions grades only the first three seconds, so a low hook rate pins the problem on the opener while the offer and landing page stay out of it.
• There is no official Meta benchmark: the ranges people cite (roughly 25 to 35% is healthy per Exposure's guide) are directional, so your own account history for the same placement beats any published table.
• You can iterate the hook alone: swap the first three seconds inside the same ad set logic instead of rebuilding the creative, and you learn ten times faster for the same spend.

What actually is a hook, and why does it decide the ad?

A hook is the first three seconds of a video ad: the frame, the first line, the motion, the sound that either earns a pause or gets scrolled past. It decides the ad because everything downstream is gated behind it. A brilliant offer, a tight landing page, a sharp CTA, none of it gets seen if the viewer is gone at second two. That is why the hook is the highest-leverage line in the whole creative, and also why it is the one you should test first. On static ads the same logic sits in the top line of copy and the thumbnail: the part that has to survive a thumb moving at speed. Everything after the hook is optimising a conversion that only happens if the hook already won the attention.

What is a good hook rate, and how do you read a dead one?

Hook rate is 3-second video plays divided by impressions. There is no official Meta benchmark, because hook rate is a custom metric, so the ranges you see are practitioner rules of thumb: Exposure's hook rate guide calls 25 to 35% a healthy range and treats anything under 20% as an opener that needs rebuilding. Treat those as directional and build your own account baseline for the same placement and objective, because Reels, Stories and Feed behave differently. Reading a dead hook is simple once you have that baseline: a hook rate well under your own recent average, on the same placement, is a dead opener. The tell is a sharp drop-off in the first three seconds while your older ads on the same audience hold steady. When the hook rate is fine but people leave later, the problem moved downstream and the hook is not your fix.

Which hook actually works for your specific buyer?

The formulas are commodity. Question, bold claim, specific number, pattern interrupt: every listicle has the same list, and your competitors are working off it too, which is why so many feeds now open the identical way. The format barely matters. The words inside it are the lever. A question hook only lands if it names the pain the buyer actually feels, in the language they actually use, and that language does not come from your own head. It comes from how customers describe the problem in reviews, forums and comments. Adscalr pulls this vocabulary from five voice-of-customer sources and captures the exact phrase markers people use, which is the raw material a hook is built from. You can do the manual version: read fifty reviews of your product and your competitors', write down the exact phrases that repeat, and open the ad with one of them instead of a formula filled with your own marketing words.

How do you iterate a hook without rebuilding the whole creative?

You change only the first three seconds and leave the rest of the ad intact. Re-record or re-edit the opener, keep the body and CTA, and you have a new test for a fraction of the production cost of a fresh creative. This is what makes the hook the cheapest thing to fix: the expensive parts of the ad, the demo, the offer, the proof, all stay. Run two or three openers against the same body, judge them on hook rate first, and only promote the winner once its rate clears your baseline and holds. A word on discipline: three days of data on a fresh opener proves little, small samples swing hard, so give each variant enough impressions that the rate stops jumping before you crown one. Iterating the opener alone is also how you keep a proven ad alive when it starts to fatigue, without risking the winner you already have.

Does the raw founder-to-camera opener beat the polished one?

Often, but not as a rule. A phone-shot, founder-talking-straight-to-camera opener frequently outperforms the color-graded studio version, because it reads as a person instead of an ad, and in a feed of native content the ad-shaped thing gets skipped. That is a real pattern worth testing early. It is not a law, though: raw wins for some offers and audiences and loses for others, especially where the product's credibility depends on looking premium. The honest move is to test the raw cut against the polished one on hook rate and let the number decide, rather than adopting "raw always wins" as folklore. There is a whole failure mode here worth reading, the ads that work precisely because they don't look like ads and where that stops being true.

A hook is not a slogan you brainstorm once and lock. It is the most measurable, most iterable line in the ad, and the discipline is reading the hook rate honestly and swapping openers cheaply until one clears your own bar. That loop, real audience language into the opener and the number deciding what stays, is the ad-creation work: in Adscalr a copywriting critic scores every concept against direct-response principles and quotes the exact line that fails, so a weak opener gets caught before it ever spends.