The Robust Branding experiment method and brief template
Every service on this site uses the same six steps. We write the brief first because it fixes the budget, the measure and the decision before anyone builds an asset.

The six steps of every experiment
Six steps from brief to decision
Robust Branding methodBrief
We agree the business question, the buyer, the budget cap and the decision the result will inform.
One-page briefDirections
Our strategists write three directions that differ in the promise, the audience or the proof.
Three directionsAssets
We build pages, ads and sequences for each direction. AI tools draft variants and our editors apply the house style.
Assets per armControlled run
Every arm gets the same budget, dates and bids. We change one variable per test.
Spend cappedReadout
We report click-through, cost per click, conversion and reply rate per arm with the sample sizes.
Written readoutDecision
Your team scales the winner and retires the rest. We write the next brief.
Next briefThe experiment brief template
We complete this page with every client before a test starts. Your team can use it with or without us.
| Field | What goes in it |
|---|---|
| Question | The decision this test informs, in one sentence. For example, which of three positions should lead the energy vertical page. |
| Buyer | The people who must respond, by role, company size and industry, and the channels they use. |
| Directions | Three directions that differ in promise, audience or proof, each written as the line a buyer would read. |
| Channels | The channels suited to the question, and the channels we leave out with the reason. |
| Cap and dates | Total spend, spend per arm, start date and end date, all set in the platform. |
| Measures | The primary measure for the decision, secondary measures for context, and the sample size per arm. |
| Decision rule | The result that scales the winner, the result that sends it back for rework and the result that ends the line. |
| Owner | The person who signs the brief, reads the readout and acts on it. |
Operating standards for every test
One question per test
A test with three questions produces three weak answers. We keep one.
Platform-held caps
We set every budget cap and end date inside the ad platform.
Every arm reported
We report each arm with its sample size, including the ones we favoured.
Directions before drafting
Our strategists choose the directions before anyone opens an AI tool.
Client-owned assets
Your company owns the ads, pages, lists and raw exports.
Logged changes
We log any change during a test with the reason and the time.
Submit your first experiment brief
Fill in the question, the buyer and the decision. We complete the rest with you in writing.