| name | growth-experiment |
| description | Design a growth experiment with a clear bottleneck, hypothesis, intervention, metric, guardrails, and learning plan. |
Growth Experiment
Use when improving acquisition, activation, engagement, retention, referral, or monetization through measured iteration.
Procedure
- Identify the funnel or behavior bottleneck using evidence.
- State a falsifiable hypothesis describing why the intervention should change behavior.
- Choose the smallest intervention that meaningfully tests the hypothesis.
- Define primary metric, guardrails, audience, duration or sample logic, and stopping conditions.
- Identify confounders and instrumentation needed before launch.
- Run the experiment without changing multiple causal variables unnecessarily.
- Analyze effect size and uncertainty, then record what was learned even if the result is negative.
Quality gate
An experiment is successful when it produces reliable learning, not merely when a metric moves.