Run a growth experiment that teaches you what to do next.

1. Name the constraint and the behavior

Start with a point in the customer journey where progress breaks down. Prospects may not respond, trial users may not complete setup, or first-time buyers may not return. Choose one behavior you can observe and explain why it matters. Write a hypothesis that connects a proposed change to that behavior: if we clarify the first setup task, more relevant users may complete it without assistance. Include the customer group and context. Avoid testing an entire growth strategy at once, because a broad result will not tell you which assumption was wrong or what to change next.

2. Establish a usable baseline

Record how the process works now and how you will measure it. Define the start and end events, time window, eligible users, and exclusions. A conversion rate without a denominator or a clear event definition can conceal more than it explains. For a small business, a carefully maintained manual log may be enough to begin. Check whether a change in traffic source, customer mix, season, or operating conditions could affect the result. You do not need to pretend a small experiment is statistically definitive. State what the evidence can support and where uncertainty remains.

3. Keep the test small and operable

Choose a change the team can implement and observe without disrupting essential customer work. Give it an owner, a start date, a review point, and a clear stopping condition. Include any customer experience or operational measures that should not deteriorate. If testing a new onboarding prompt, watch support requests and task errors as well as completion. Keep a record of what changed during the test. Unplanned adjustments may be necessary, but they affect interpretation and should not disappear from the report.

• Hypothesis: what you expect and why.
• Audience: who is included and excluded.
• Change: the specific intervention.
• Primary measure: the behavior that answers the question.
• Guardrails: outcomes that should not worsen.
• Decision: what you will do for each plausible result.

4. Review the result and make the next decision

Compare the observations with the baseline and review the context before claiming an effect. Ask whether the change reached the intended users, whether tracking worked, and whether other conditions changed. Read customer feedback alongside numbers. Decide whether to keep, revise, stop, or repeat the test with a better design. Write a short outcome note: what you tried, what happened, what remains uncertain, and what you will do next. A test that disproves an assumption can be successful learning. Avoid attributing a later revenue or funding result to one experiment unless the evidence supports that specific relationship.

Questions, answered.