80% never
get past the
first three
seconds.
Which makes the hook most of the ad. Four out of five people you pay to reach will only ever experience three seconds of your work. Six ideas and you'll never write a bad one again.
A hook has
two jobs
Everyone only knows the second one.
- Attract the right audience
- Get this wrong and job 2 actively hurts you.
- Keep them watching
- Stop the scroll. Get them into the ad.
Meta watches who holds through the first half of your ad and decides those must be your people. So it goes and finds more of them. If your hook pulled a crowd that has nothing to do with your product, your ad is now being optimised toward an audience that will never buy.
The ad that proves it
A video opens as a genuinely interesting explainer about how submarine crews escape a failing sub. Then it hard-cuts, mid-sentence, into a commercial snow and ice management company.
Nathan got served it. He lives in Australia, where it does not snow. He got it because he's the kind of person who watches educational reels. The algorithm did exactly what the hook told it to do. Millions of views. Almost no business.
Why it survives on organic and kills you on paid
Same creative. Same ten leads. One of them is free money and one of them ends a company.
There are exceptions — gaming, maybe, where the meme genuinely is the target market's language. But the test is never "is it a meme." The test is whether the hook attracts people who will buy the thing.
A hook is
three things
Most people only ever think about one of them.
Visual
Audio
Copy
Real example: an ad performing fine got a text overlay added on top reading "this transformed my life." Hook rate doubled. Creative performance doubled. The visual didn't change. The script didn't change. One line of copy did it.
Most people think about the visual, auto-generate captions from the audio, and never treat copy as its own lever. That's a third of the hook thrown away before you start.
Try it — one hook, three layers
Same ad. Flip each layer from flat to dialled and watch what happens.
Learn one framework.
Then drop it.
Templates get you started. Then they make you look like everyone else.
If you can't yet tell a good hook from a bad one, grab templates. They'll get one foot in front of the other. Then leave them, because they go generic fast and every competitor is using the same ones. The goal is writing hooks from scratch, because that's where the whitespace is.
It's also the opening of every YouTube video ever made, which is why it moves so cleanly between organic and paid.
A fifty-year-old woman cannot open with "if you're a twenty-year-old guy." There's no trust in the casting, and no framework saves you from that.
⚠ The callout thing people get confused about
You'll hear that persona callouts are overrated. That's true of the bare label. It's not true of a callout welded to stakes.
- "Agency owners who need X, listen up."
- Asks you to raise your hand and gives you nothing back. The "listen up" is the tell — it demands attention instead of earning it. You feel targeted. You scroll.
- "If you're using protein powder, you're slowly destroying your gut."
- You're not being told who you are. You're being told what you do — and the problem lands before you can scroll. You feel seen. You watch.
The callout is a filter, not a hook. It buys you the right to deliver one. Quick test: delete it and read what's left. If there's still a problem with stakes in the line, you're fine. If deleting it leaves nothing, the callout was your hook — and that's the version people scroll past.
Grade it in
this order
The order is the whole lesson.
The first three are the job. The last two sit lower not because they don't matter, but because you have three seconds. Fitting proof into three seconds is hard and it isn't a priority. Don't distort a good hook trying.
Run a completely normal hook and put an influencer or a genuinely famous person in front of it. You get credibility instantly, without spending a single one of your three seconds on it.
And stop grading as soon as the first two fail. A novel, specific, credible hook that's unclear and irrelevant is still a dead hook.
Two small things
that do a lot
These are the highest value-per-word lessons here.
Put a number on it
- "We taste tested the bestselling jerkys and found the same problems in almost all of them."
- Generic. Holds nobody. Reads as: uh huh, you're about to sell me something.
- "We taste tested the bestselling jerkys and found the same two problems in almost all of them."
- Holds.
The number does two jobs at once. It bounds the time commitment, so the viewer can price the watch before committing to it — only two, I can sit through that. And it turns a vague gap into a finite one, which pulls far harder. "There are problems with jerky" opens nothing. "There are exactly two things to look for" opens something shaped that your brain wants closed.
It also smuggles in proof. I bought all of them and found exactly two things implies work was done.
Generalise it: any time your hook has a plural noun with no number in front of it, ask whether one belongs there.
The visual has to carry what the copy opened
This is where people take a good written hook, film it flat, watch it fail, and blame the hook.
- Creator says the whole line to camera.
- One shot. One face. Nothing moves.
- "We taste tested—"
- CUT at one second → supermarket, someone sweeping jerky off the shelf, arms full, through the checkout
- CUT back → "—found two problems in almost all of them"
- Creator pulls out a whiteboard. 1. and 2. are written on it, blacked out. Pen in hand.
The visual isn't decorating the line. It's carrying the thing the line opened. Same words, completely different ad.
And this is where hook volume actually comes from
The bridge is
where ads die
And it's an expert's mistake, not a beginner's.
Once you know what a good hook, a good body and a good CTA look like, the thing you still don't know is how to join them without a cliff in the retention curve. That's the ceiling that stops good creative people becoming great ones.
"If you think agencies can't be trusted, you're not wrong. Here's how to tell if yours is the exception. Book a call below and talk to me and my team—" Perfect hook. Then it just falls into the sale, and answers the gap it opened with an ad instead of an insight.
Bleed it. Don't stepstone it.
Product awareness is how aware the viewer is that they're being sold to. It should climb slowly across the whole ad and arrive at the CTA feeling inevitable.
- Hard jump at the hook — product named in sentence one
- Another jump at the bridge
- Another at the CTA
- The viewer feels every step. They leave at one of them.
- Zero through the hook. They don't know they're being sold to.
- It just creeps.
- The CTA arrives and feels obvious.
- They never notice they were moved.
A good electrolyte ad opens "drinking three litres of water a day doesn't mean you're hydrated," teaches for a while, mentions the brand in passing inside a sentence about sodium and magnesium, and immediately goes back to teaching. The viewer thinks: okay, this is an ad — but I'm still learning something, so I'll stay. It's the dwelling that spikes the curve, not the mention.
Two more things worth stealing
The sponge
The second bridge
If you forget everything else
- 01Attract the right audience first. Keep them watching second.
- 02Hooking the wrong people teaches Meta to find more wrong people.
- 03Memes work on organic because reach is free. In paid you buy every wasted impression.
- 04A hook is visual + audio + copy. Fixing one layer can double it.
- 05Persona → Problem → Interrupt → Promise. The interrupt is the part you'll skip.
- 06Clarity, relevance, novelty. The rest are bonuses.
- 07Don't script credibility. Cast it.
- 08Put a number on it. "Two problems" beats "problems."
- 09The callout is a filter, not a hook. Behaviour plus stakes, never a label.
- 10Bridge into education. Bleed product awareness. Never stepstone.
Hooks — Six Ideas. Internal training. Third in the set, after Creative Strategy and Static Ads.
