The hooks and angles matrix, how long to let a variant run, and the internal habit that ruins tests.
The minimal matrix
Two axes, and nothing more.
The hook: the first three seconds, what is seen and what is heard.
The angle: the promise defended, that is the problem the product solves.
Three hooks on two angles give six variants, which is already too many for a first cycle. Start with three hooks on one single angle: that is the simplest matrix that teaches anything.
One technical clarification: the three variants share the same video body. Only the first second changes, which makes the comparison readable.
How long to let it run
Three benchmarks, in this order.
A spend threshold per variant, defined before launch: that is what decides, not the calendar.
A minimum of three days, below which the figures mean nothing.
And a maximum of one week, beyond which the environment has changed too much.
One organisational consequence: write the threshold into the campaign document. Without that, somebody will stop a variant on the second day, in all good faith, and the whole cycle will be lost.
When to cut
Three reliable signals, and one trap.
The variant has reached its spend threshold with no purchase: it is cut.
The variant has a cost per purchase clearly higher than the control, with sufficient volume: it is cut.
The variant is winning: it stays, and the budget moves towards it gradually.
The trap: cutting on the click rate. It is the most visible and least predictive indicator, because a hook can attract many clicks and no purchases.
What follows a winner
Three steps, in this order.
Describe in writing why it won: which angle, which promise, which person, which setting.
Produce two variants of the same angle with another creator, rather than running the same video longer.
And keep the winner in production as long as it stays profitable, watching its weekly wear.
One constant observation: a winning creative exhausts itself in a few weeks. What is kept is not the video, it is the rule learned.
What is tested, and in what order
| Rank | Variable | Why first |
|---|---|---|
| 1 | The hook | It decides the three-second view |
| 2 | The argument | It decides the full view |
| 3 | The call to action | It decides the click |
What makes a test readable
One variable at a time, an identical budget on each side, and an equal run time. Without those three conditions, the result cannot be attributed to anything.
What is documented for each test
The winning file, the losing variant, the period and the budget. Here, an undocumented test carries no weight at the next approval.
What to write before starting
Five lines, in the campaign document.
The measurable objective, with a figure.
The variants produced, with what distinguishes them.
The spend threshold per variant.
The single person who decides, and their response time.
And the review date, set in advance.
One remark suited to this market: that document is an advantage, not a constraint. In a working culture valuing writing and preparation, a clear test plan gets a budget more easily than a good idea presented orally.
One last marker, valid everywhere but particularly here: the aim of a test is not to validate what the team thinks, it is to discover what it does not know. A test cycle in which every hypothesis is confirmed has learned nothing, and it cost the price of three productions.
