On Meta, targeting has become broad and largely automated. What you still control is the creative, and above all the hook: the first seconds of the video that decide whether someone keeps watching. That makes creative testing the main lever for performance. But how many hooks should you test, and how often do you need new ones? This article gives a practical framework. If you'd rather have the creatives made for you, see AI UGC ads.
In short: test several distinct hooks per offer at once rather than one at a time; we work in batches of 10. Keep the offer and the body of the video constant so you learn what the hook does. Replace creatives when the signals of fatigue show up, not on a fixed calendar.
Why the hook matters most
In a vertical feed, people decide in the first moments whether to stop. A strong offer behind a weak hook never gets seen. The same product, explained in the same way, can perform very differently depending on the opening line, the first visual and the first problem named. That is why creative tests should change the hook first.
How many hooks to test per offer
There is no universal number, but a few principles hold:
- Test several at once. Testing one hook per week is slow and confounded by changes in the auction. A batch lets you compare hooks under similar conditions.
- Make them genuinely different. Ten variations of the same sentence are one hook. Use different angles.
- Keep the rest constant. Same offer, same body, same call to action. Otherwise you don't know what made the difference.
We produce batches of 10 hooks per offer because it gives enough variety to see a pattern without spreading the budget too thin. Smaller budgets can test fewer hooks at a time and rotate them.
Ten hook angles to start with
- The problem: name the pain your buyer feels right now.
- The result: show the end state first, then explain how to get there.
- The objection: answer the reason people don't buy.
- The comparison: old way vs new way.
- The mistake: "Stop doing this when you…"
- The everyday situation: a moment your audience recognizes.
- The question: a question the viewer has asked themselves.
- The number: a specific, true detail about the product or routine.
- The routine: "My three-step routine for…"
- The identity: speak directly to a specific group, such as a profession or age.
Avoid hooks that depend on claims you can't support, such as medical outcomes or fabricated results. They get rejected in review and damage trust.
How to structure the test
| Element | Recommendation |
|---|---|
| Offer | One offer per test |
| Variable | The hook (first seconds); body and CTA unchanged |
| Format | Vertical, with subtitles, for Reels, Stories and TikTok |
| Budget | Enough per creative to reach a meaningful number of impressions before judging |
| Metrics | Hook rate (share who watch past the opening), hold rate, CTR, cost per result |
| Decision | Keep the top performers, rewrite the losers into new angles |
Signs a creative is worn out
- Frequency keeps rising while reach grows slowly.
- CTR and hook rate fall week over week with no change in targeting.
- Cost per result climbs on an ad set that was stable.
- Comments repeat ("seen this a hundred times").
When two or more of these appear, it's time to rotate in new hooks. A refresh schedule based on signals wastes less than a fixed calendar.
Example: one offer, ten hooks
An illustrative batch for a hypothetical kitchen cleaner. The body of the video (a short routine and the product) and the call to action stay the same; only the opening changes.
| # | Angle | Opening line |
|---|---|---|
| 1 | Problem | "Grease on the stove that never fully comes off?" |
| 2 | Result first | "This is my stove after two minutes. Here's how." |
| 3 | Objection | "I thought all cleaners were the same." |
| 4 | Comparison | "Scrubbing vs letting it sit for 60 seconds." |
| 5 | Mistake | "Stop spraying and wiping straight away." |
| 6 | Situation | "Guests in 20 minutes and the kitchen looks like this." |
| 7 | Question | "Why does the stove get sticky again so fast?" |
| 8 | Routine | "My 3-step Sunday kitchen reset." |
| 9 | Specific detail | A true detail about the formula or the size of the bottle |
| 10 | Identity | "For people who cook every single day." |
Reading the results
Don't pick a winner on the first day. Give each hook enough impressions to compare, then look at the metrics in order:
- Hook rate: did people stop and watch past the opening?
- Hold rate: did they keep watching into the body?
- CTR: did the ad make them want to act?
- Cost per result: did it produce leads or sales at an acceptable cost?
A hook with a high hook rate and a poor cost per result tells you the opening works but the rest doesn't match it. That is useful too: keep the opening and rework the body. Over several batches, the winning angles show you how your audience thinks about the product, which is worth carrying into landing pages and organic content.
A creative testing calendar
| When | What to do |
|---|---|
| Week 1 | Launch a batch of distinct hooks for one offer |
| Week 2 | Read hook rate, hold rate and cost per result; focus budget on the winners |
| Week 3 | Brief the next batch: variations of the winning angles plus a few new ones |
| Week 4 | Launch the next batch next to the current winners; retire creatives showing fatigue |
| Monthly | Review which angles won across batches and update the hook library |
Hooks for different audiences: cold and warm
The same hook doesn't work for everyone. People who have never heard of you need a reason to stop; people who visited your site or watched your videos need a reason to act now.
- Cold audiences: lead with the problem, a surprising detail or a recognizable situation. Don't assume they know the brand.
- Warm audiences: lead with the objection, the result or a specific offer. They already know what you sell.
- Existing customers: lead with a new use, a routine or a complementary product.
Test hooks separately for cold and warm audiences; a hook that loses with cold traffic can win in retargeting.
Build a hook library
Keep a simple table of every hook you've tested: the angle, the opening line, the offer, the dates and the key metrics. After a few months it becomes your most valuable creative asset. You'll see which angles work for which offers, which ones fatigue fastest and which ones work across platforms. New team members, agencies and AI creators can start from proven angles instead of guessing. A library also stops the same losing idea from being tested twice under a different name.
In-house or agency?
An in-house team knows the product best; an agency usually produces faster and brings patterns from other accounts. A practical split: your team owns the offer, the claims and the decisions; production of hooks in batches, with one consistent presenter, can be outsourced. If you work with an AI creator, the same character can also run an organic channel, so the best organic videos feed new ad tests. That loop is described in AI creators vs paid ads.
Where to get enough creatives
The constraint for most brands is production. Each new hook means a new video, and briefing human creators for every test is slow. Three practical sources:
- AI UGC batches: one consistent character, 10 hooks in 5 business days. See AI UGC: what it is and what it costs.
- An organic channel: videos that perform organically become ads with less testing. See AI creators vs paid ads.
- Human creators: best when the product needs real use or reviews.
Frequently asked questions
How many creatives should be in one ad set?
Enough to give the algorithm choices without splitting the budget too thin. Test in batches and remove clear losers rather than adding creatives indefinitely.
How often should I refresh Meta ad creatives?
When the fatigue signals appear: rising frequency, falling CTR and hook rate, rising cost per result. The interval depends on budget and audience size.
Should I test hooks or whole new videos?
Hooks first. Change one thing at a time so you know what worked.
Are AI-generated creatives allowed on Meta?
Yes, within Meta's advertising policies and disclosure rules. We label AI content and avoid medical claims and fake reviews.
Can an ad creative agency speed this up?
Yes, mainly by producing distinct hooks quickly and keeping one consistent presenter across tests.
Get a free niche audit. Send your offer on the AI UGC ads page. We'll suggest the first batch of hooks to test.