Hook formulas that work (and why)
The first line of a short video decides whether anyone sees the second. These are the patterns that show up again and again in videos people finish, with honest notes on when each one backfires.
Updated October 15, 2026 · 9 min read · by the Postfield team
- A hook is the first line, said or shown, in roughly the first two seconds. Keep it to about 4 to 12 words.
- Good hooks open a loop: a question, a mistake, a surprising result, a specific number.
- Every formula below has a failure mode. Most fail by over-promising or sounding like an ad.
- Salesy phrases ("check this out", "use code") belong at the end, not in the hook.
- Formulas are starting points. Test several versions with real posts and keep what works for your audience.
What a hook actually does
A hook has one job: give the viewer a reason to stay for the next few seconds. It doesn't need to explain the product. It needs to make the viewer curious, seen or slightly surprised.
Three things show up in most strong hooks:
- It is short. On-screen text has to be readable at a glance. Around 4 to 12 words is a useful target.
- It is specific. "3 mistakes" beats "some mistakes." "$7 cold brew" beats "expensive coffee."
- It talks to a person. "You" or "I", like a friend, not "our innovative solution."
14 formulas, with examples
Each formula has an example, why it works, and when it fails. Examples are illustrative, not real campaign results.
1. POV
"POV: you finally stopped paying $7 for cold brew". It drops the viewer into a scene they recognize. Fails when the scene is too generic or doesn't match the audience's life.
2. I tried it
"I tried sleeping with mouth tape for 14 days". A personal experiment promises an outcome. Fails when the result is boring or the poster never actually tried it (which is also an FTC problem for paid posts).
3. Stop doing this
"Stop rinsing your rice like this". A mild warning creates urgency. Fails when the "mistake" is not really a mistake, which reads as clickbait.
4. The numbered list
"3 things I wish I knew before my first marathon". A number tells the viewer how long they're committing to. Fails when the list is padded or the number is huge ("27 tips").
5. Nobody tells you
"Nobody tells you this about renting your first apartment". It promises insider knowledge. Fails when the payoff is common knowledge.
6. The question
"Why does your skin feel tight after washing?". A real question the viewer has asked themselves. Fails when it is a yes/no question they can answer in their head and scroll.
7. Hot take
"Unpopular opinion: meal prep is overrated". Mild controversy invites comments. Fails when it is fake controversy or punches down.
8. Before and after
"Day 1 vs day 30 of learning guitar". Visible change is satisfying. Fails when the change is small, or in regulated categories where before and after claims need substantiation.
9. Here's how
"Here's how I pack a week of lunches in 20 minutes". Clear utility. Fails when the how-to is slow to start.
10. The specific number
"I spent $412 on groceries this month for 4 people". Precise numbers feel true and invite comparison. Fails when the number has no context.
11. The wrong way
"You're probably cleaning your air fryer wrong". Gentle challenge to the viewer. Fails when it sounds condescending.
12. The reveal
"The real reason your plants keep dying". Promises a cause, not just a tip. Fails when there is no single real reason.
13. Things I'd do differently
"If I started my dog's training over, I'd do this first". Hindsight feels earned and honest. Fails when the poster has no real experience to look back on.
14. The callout
"If you get headaches at 3pm, watch this". Self-selection: the right viewers stop. Fails when it slides into health claims you can't back up.
What kills a hook
| Problem | Example | Fix |
|---|---|---|
| Sounds like an ad | "Check out this amazing product!" | Lead with the viewer's problem or a result |
| Too long | A 20-word sentence | Cut to the one idea that matters |
| Too vague | "This changed everything" | Say what changed, with a number |
| Hashtags in the hook | "#skincare #glow #fyp" | Hashtags go in the caption (with #ad first on paid posts) |
| ALL CAPS | "YOU NEED THIS NOW" | Normal case reads as a person |
| Discount up front | "Use code SAVE20" | Save the offer for the end |
Hooks on paid posts
A good hook and a clear disclosure are not in conflict. On paid posts the caption starts with #ad, the platform label is on, and the video says it is sponsored. The hook still does its job: make people curious about the content.
What you can't do is use a hook to claim experience the poster doesn't have. "I tried this for 30 days" on a paid post has to be true. The FTC's endorsement rules require endorsements to reflect real experience. If you want "I tried" hooks, ship product early enough that the poster actually tries it.
A simple process for writing hooks
- Write the one thing the video delivers in plain words.
- Draft five hooks from five different formulas above.
- Read each out loud. Cut any word you wouldn't say to a friend.
- Check length (4 to 12 words), one specific detail, and no salesy phrase.
- Score them with a tool like our free Hook Score to catch obvious problems. It is a transparent checklist, not a prediction.
- Post the top two or three as real variations and compare.
Worked rewrites: weak hook to strong hook
Most weak hooks are fixable with one or two moves. Here are rewrites with the reasoning, using made-up products.
| Weak | Stronger | What changed |
|---|---|---|
| "Check out our new protein bar!" | "I replaced my 3pm vending machine snack for 2 weeks" | From ad voice to a personal experiment with a number |
| "This app is so helpful" | "I stopped forgetting bills after one setting change" | A specific problem and a concrete fix |
| "Amazing skincare routine for everyone" | "3 steps I cut from my skincare routine" | A list, a number, and a surprising angle (cutting, not adding) |
| "Our restaurant has great tacos" | "POV: you found the taco spot locals don't post about" | A scene the viewer can step into, plus curiosity |
| "Big sale this weekend, use code SAVE20" | "Why I waited 6 months to buy this" | Moves the offer to the end and opens a loop |
| "Tips for dog owners" | "You're probably walking your puppy wrong" | Talks to the viewer and hints at a mistake |
Each rewrite is still a guess. The point is to give your test slots to hooks that clear the basics: short, specific, personal, not salesy.
Matching formulas to your category
Some formulas fit some categories better. This is a starting map based on how the formulas work, not data from a study.
| Category | Formulas to try first | Watch out for |
|---|---|---|
| Beauty and skincare | I tried, Stop doing this, Things I'd do differently | Before and after claims need substantiation |
| Food, beverage and restaurants | POV, Specific number, Hot take | Taste is personal; avoid "best ever" claims you can't back |
| Apps and software | Here's how, The wrong way, The callout | Show the app quickly; long setups lose people |
| Fitness | Before and after, I tried, Numbered list | Results claims and health claims |
| Supplements and wellness | Question, Things I'd do differently | Health claims need extra review; keep hooks about routine, not cures |
| Fintech | Specific number, Nobody tells you | Money claims and promises of returns |
For category-specific plans with example hooks, see the industry pages, for example beverage, beauty or apps.
Formulas are hypotheses
What works for a skincare audience may flop for a fintech app. Treat every formula as a guess and test it. Our guide on testing hooks at volume covers how to run fair comparisons without fooling yourself with one viral outlier.
If you want real people posting your hook variations at volume, all disclosed, that's what Postfield does.
FAQ
How long should a TikTok hook be?
A useful target is 4 to 12 words, short enough to read as on-screen text in a couple of seconds. Shorter hooks often lack a reason to stay; longer ones don't get read.
Should the hook be spoken or on screen?
Both, when you can. Many people watch with sound off, so on-screen text carries the hook for them. Saying it reinforces it for everyone else.
Do hook formulas get stale?
Some do as audiences see them everywhere. That's why testing matters. A formula is a structure; fresh specifics keep it working.
Is the Hook Score an AI prediction?
No. It's a transparent, rule-based score: length, proven patterns, voice, specificity, curiosity, and penalties for salesy phrasing. It helps catch obvious issues. It doesn't predict views.
Sources
Short, useful emails on hooks, disclosure and posting at volume. Five emails, then we stop.
No spam. Unsubscribe in one click. We never sell your email.