Every test goes down the same path. Answer each question yes or no and the path lights up: what we test, how we build it, which creative we need, what it costs and how we read it. The test card at the bottom fills in as you go.
Four questions every Monday, in this order. The first yes sets the week's test.
Now we know the topic, the shape of the test follows from what we want to learn. Same rule: stop at the first yes.
Every version needs four things decided: its opening (the first 3 seconds: shot, words on screen, first line), its format, its proof (why anyone should believe it) and whether we have it, can re-cut it, or need to make it.
A past ad or shoot says this message as it is. Use it.
Right footage, wrong opening or old offer text on it. Edit it.
Write a brief: the opening, the format and the proof.
Answer step 2 and the build appears here.
Inside one idea, versions differ in how they show it, never in what they say.
One ad is seen by people at every stage, so the first 3 seconds can't assume they know us.
Not a label. The material, the maker, the offer.
The picture already shows the product. The copy says why it's worth it.
Every ad set gets about 3× the target cost per sale. If it spends 2× with no sale, we stop it early.
Answer step 2 to see the budget for this build.
We judge the idea against the other ideas, not the single ads inside it. Versions with little spend are unread, not losers.
Hit the target cost per sale. The version Meta favoured is copied into Scale as the same post and the weakest Scale ad comes out.
Close to target, or too little spend to judge. Top it up, or change one thing and run it again.
2× target with no sale, or clearly behind. We write down what it taught us.
This is what goes in the log. The top half is filled in before anything goes live; the bottom half after the read.
See it filled in for a real Hyde & Hare test: the beanbag example →