The first three seconds of a Facebook ad do roughly 80% of the work. If the hook fails, none of the body copy matters. The product can be brilliant. The offer can be irresistible. Nobody will see it.

So this is mostly a hook playbook. The body copy section is short because, honestly, the body copy is the easy part.

What a hook actually does

A hook earns the next three seconds. That's the whole job. It doesn't need to sell, persuade, or even mention the product. It needs to interrupt a scrolling thumb long enough that the brain decides to keep watching.

The mistake most ad copy makes is trying to sell in the hook. "Get 30% off our new collection" doesn't stop scrolls. It is a scroll.

14 hook structures that work

  1. The contrarian claim: "Everyone says X. They're wrong." Works when you have a genuinely heterodox take. Fails when you're being edgy for its own sake.

  2. The specific number: "I tested 47 face creams. Only 3 actually worked." Specificity earns trust. Round numbers (10, 100) feel made up.

  3. The before / after curiosity gap: "I had 12 of these. Now I have 1." Implies a story worth waiting for.

  4. The named enemy: "Why I stopped using [well-known competitor]." Polarising, high-CTR, requires the comparison to be honest.

  5. The mistake confession: "I wasted £4k on the wrong protein powder. Here's what I switched to." Vulnerability buys attention.

  6. The insider warning: "This is what supplement brands don't want you to know." Use carefully. Cynical when overused.

  7. The pattern interrupt visual: Hook isn't the words, it's the frame. Unexpected object, odd angle, motion in the first half-second.

  8. The direct question to the avatar: "How long does it take you to fall asleep?" Specific to the audience's lived experience.

  9. The "POV" frame: "POV: it's Sunday night and your skin is wrecked from the week." Works for younger audiences. Forced when older.

  10. The credentialed authority: "I'm a dermatologist. Here's what I actually use." Earns instant credibility if the claim is real.

  11. The unboxing tease: "I didn't expect [outcome] when I opened this." Curiosity hook for product-led stories.

  12. The myth-bust: "You don't need 8 glasses of water a day. Here's what you actually need." Educational, builds trust before selling.

  13. The timeline reveal: "Day 1 vs day 30 vs day 90." Time-lapse structures work because they imply real evidence.

  14. The price flex (when honest): "This costs more than the competition. Here's why I still buy it." Risky, only works when the value is genuinely defensible.

What kills hooks

  • Starting with the brand name. Nobody scrolling cares about your brand yet.
  • Starting with the offer. The offer is the close, not the open.
  • Generic openings: "Looking for X?" / "Tired of Y?" Your audience has scrolled past 80 of those today.
  • Cute wordplay that requires reading to understand. Hooks have to land at audio-off, captions-half-read speed.

Body copy: keep it short, structured, scannable

After the hook, you've got maybe 15-30 seconds of attention. Use it for:

  1. The promise (what the viewer gets)
  2. The proof (why they should believe it)
  3. The push (what to do next)

Three beats. Each beat one or two sentences. No filler. No "as we mentioned earlier". Nobody mentioned anything earlier; they're 4 seconds into the video.

Captions and overlay text

Around 85% of Facebook video is watched without sound. Captions aren't optional. Overlay text reinforces the hook and key claims for the audio-off viewer.

Rules that actually matter:

  • Keep on-screen text under 7 words per beat
  • Burn captions into the video, don't rely on auto-captions
  • Match the visual tone (overly polished captions on rough UGC feels off)

The test you should run this week

Take your best-performing current ad. Write 5 new hooks for the same body. Cut 5 variants. Run them in a clean test against the original. The body and offer haven't changed, so any difference is the hook.

You'll learn more about your audience in that one test than in a month of brand research.