Guide
Same body, six openings, and the media is the bill
A hook test is six ads that are identical after the first three seconds. You eliminate on the three-second view rate. You confirm on cost per acquisition, never the other way round. This page covers the method. Then the arithmetic showing ninety percent of the budget is media rather than production. Then the four ways a hook test quietly becomes six files and no lesson. You will finish able to run one this week.
- Share of a hook test's budget that is media, not production
- 90%
- Openings in one test, recorded in one sitting
- 4-6
- Export cost of six thirty-second variants
- 60 CR
Identical after the first three seconds, or the result is unreadable
Same body, same proof, same ask, same captions, same music. Only the first sentence differs.
That constraint is what makes the result attributable. If the hook changes and the body changes, a difference in performance tells you nothing you can reuse next quarter.
In practice that means recording all the openings in one sitting against one body. It takes minutes rather than a production cycle.
Record the body once, then run the six straight after it while your voice is still warm. Six back to back is a normal target for a single session.
Keep the openings roughly the same length as each other. A three-second hook against a seven-second one is a length test with a hook test hiding inside it. Neither result transfers.
Write the six the night before. Say each one out loud before it goes on a card. An opening that sounds written is an opening you cannot deliver six times.
Then name the files as you record them. Concept, hook number, take number. A folder of unnamed clips is the most common way a hook test dies before it launches.
Judge on the three-second view rate, then confirm on the money
A hook affects who stays past the first seconds. Its effect on purchases is real and heavily diluted by everything downstream. Waiting for conversion data to judge a hook is slow and expensive.
The three-second view rate accumulates hundreds of times faster than purchases. It also measures the exact decision the hook is trying to influence.
Then verify the survivor downstream. A hook that wins on view rate and loses on cost per acquisition is attracting the wrong attention. That is useful to know and impossible to see from one end.
Cheap metric to eliminate, expensive metric to confirm. Never the reverse. Optimising to an early metric alone reliably produces ads that stop scrolls and sell nothing.
- Stage one: eliminate on early attention metrics, quickly and cheaply.
- Stage two: run the two or three survivors long enough to read conversion cost.
- Never promote a hook on the basis of stage one alone.
Two stages, and the direction that never reverses
Eliminate on the cheap metric, confirm on the expensive one. The other way round buys scrolls that never convert.
Six openings live
Same body, funded equally
Cut on view rate
Bottom of the batch goes now
Two or three survive
Now they get real budget
Confirm on cost per acquisition
The only promotion metric
This is the whole editor
Highlight a phrase and a clip lands on those exact words. No timeline, no keyframes, no layers.
Ninety percent of a hook test is media, so cheaper filming saves almost nothing
Do the arithmetic in media, because production is the small number here.
Say you need a set number of impressions per hook to read a view rate reliably. Call your cost per thousand impressions C. Six hooks costs six times the impressions, times C, divided by a thousand.
That figure is your true hook-testing budget. It flips the usual instinct. Cutting production cost saves you little.
Cutting the impressions required saves you a lot. You do that by judging on a metric that accumulates fast. Choose the cheap metric and you can afford twice the hooks for the same money.
That is the whole trade. It is also the opposite of where most teams negotiate.
Work the number out before you launch. Impressions per hook, times six, times your cost per thousand, divided by a thousand. If that lands above a week of spend, run four hooks rather than six.
Where a six-hook test's budget goes
Production is the bar nobody should be negotiating. The impressions are where the test is won or lost.
- Recording the openings5%
- Cutting six variants5%
- Impressions to a verdict90%
Four ways a hook test quietly becomes six files and no lesson
The bodies drift. Someone re-records the body for one variant because it felt flat. Now the test measures two changes. Cut every variant from the same body take.
The hooks are too similar. Six polite rewordings of one idea produce six similar results and one conclusion. The conclusion is that the test was wasted.
The test never ends. Set the impression threshold beforehand, call it, log the result, move to the next layer.
The lesson is not recorded. The point of a hook test is a transferable rule, not a winning file.
Write the sentence down. For this audience, openings that name a price beat openings that name an outcome. That sentence is what you paid for. The file is a by-product.
Four failures, and the fix for each
None of these costs money to prevent. All of them cost a fortnight to discover afterwards.
| How it goes wrong | What prevents it | |
|---|---|---|
| Body drift | One variant gets a fresh body recording | Cut every variant from the same body take |
| Timid variation | Six rewordings of one idea | Six different shapes, deliberately spread |
| No end point | It runs until somebody gets bored | An impression threshold set before launch |
| No lesson | You keep a winning file and nothing else | One written sentence you can reuse |
Six variants for about 60 credits is what keeps the method running
The mechanics only work if producing six near-identical ads is easy. Give each variant a revision round with an editor and coordination eats the return. The test quietly stops happening.
This is the specific job Cutroom does. One take goes in, up to three minutes. Back comes a finished 9:16 MP4 you direct on the transcript.
Trim the opening. Delete a line you do not want. Write the opening headline, or pick one it drafted. Then export.
At 20 credits per output minute, six thirty-second variants is about 60 credits. Basic carries 2,500 credits a month. The sixth variant costs what the first did, which is not true anywhere a person is doing the assembly.
The facts to plan around. Somebody has to be on camera. Three minutes is the ceiling. Every export is a 9:16 MP4, and there is no timeline underneath.
The recording session still outranks the editing session. Six genuinely different ideas come from twenty minutes with a pen the night before.
Questions people ask
- How many hooks should be in one test?
- Four to six. Fewer and you are not exploring enough different shapes to find an outlier. More and each one gets too little budget to produce a readable number, unless your spend is large.
- Can I test hooks by editing an existing winner?
- Yes, and it is the best place to start. A body that already converts is a proven asset, and swapping openings on it tells you about attention without risking the part you know works.
- Is the first frame part of the hook?
- Yes, and it is often the larger effect. The first frame is what a scrolling viewer processes before any words arrive, so treat the opening image as part of the variable you are testing.
- What if all six hooks perform the same?
- Then the constraint is somewhere else, usually the offer or the audience. That is a useful result. It stops you spending another month writing openings for a problem that openings cannot solve.
- Who should not run a hook test at all?
- Anyone who cannot afford the impressions for six arms to reach a readable view rate. Run two arms instead, or put the money on the offer. The method works the same way at two.
Same body, six openings, judged on the metric the opening controls. About 60 credits of production, because the sixth variant costs exactly what the first one did.