Write the decision before launching the experiment. If the team cannot say what it will do with the result, the test is probably underspecified.
I would keep the brief short enough to review in one sitting. The purpose is to expose assumptions and prevent the goalposts from moving once results arrive.
Copy this brief
Problem: What is happening in the business, and what evidence shows it?
Hypothesis: Why might the proposed change improve the outcome?
Audience: Who is included, who is excluded, and how will they be reached?
Change: What exactly differs from the current approach?
Primary outcome: Which behavior or business result determines the decision?
Guardrails: What must not deteriorate beyond an acceptable level?
Resources: Budget, people, implementation time, and follow-up capacity.
Timing: Launch date, outcome maturity period, and review date.
Decision rule: What happens after a positive, negative, or inconclusive result?
A worked example
Suppose a fictional B2B product receives relevant traffic but few visitors request a demonstration. The hypothesis is that the page explains features without showing the workflow the buyer cares about.
The test replaces a feature-heavy section with a short workflow demonstration. The primary outcome is qualified demo requests per eligible visitor. A guardrail is the proportion of requests that become appropriate sales conversations. The team sets the observation period and minimum useful evidence before launch.
A higher form-fill rate with lower qualification would not automatically justify the change. It may mean the new explanation attracts more interest while creating the wrong expectation.
Separate the result from the explanation
Record what happened before explaining why. If several things changed or the audience shifted, state that limitation. A useful result can be inconclusive about causality while still revealing an implementation issue or a better research question.
Do not hide failed tests. Keep the hypothesis and result searchable so another person does not repeat the same work without understanding what was learned.
End with an owned action
Choose whether to adopt, revise, stop, or gather more evidence. Name the person responsible and the next review point. An experiment creates value when learning changes execution.
The 90-day growth plan should contain the prioritized experiments, not duplicate every detail of these briefs. That separation keeps the plan readable and the work accountable.
Co-founder and CEO of Stackmatix, startup advisor, and former Head of Sales at MightyHive. Connect with Matt on LinkedIn ↗ · More about Matt →