Design an experiment before you see the result

1. State a falsifiable hypothesis

Specify who will do what, under which conditions. “Five of twenty qualified prospects will pay a refundable reservation” is testable; “people want healthier food” is not specific enough.

2. Set the threshold in advance

Record the metric, minimum sample, success threshold, timeframe, and stopping rule. Choose a method that measures behavior rather than polite enthusiasm.

3. Document the result honestly

Record the numerator and denominator. For example, 18 of 24 deliveries on time is 75%, below a predeclared 90% target. Save the observed result without changing the target afterward.

4. Select an outcome

Choose Validated, Invalidated, or Inconclusive yourself, explaining uncertainty and limitations. Decide whether to iterate, stop, or test a narrower question.

Use a one-page test brief

• Assumption: what must be true for the venture to work?
• Hypothesis: which person will take which observable action?
• Method: what will you show, offer, or measure?
• Decision rule: sample requirement, threshold, timeframe, and stopping rule.
• Integrity: participant consent, data handling, and limitations.
• Ownership: person responsible for execution and review.

Watch for invalid conclusions

A small convenience sample cannot establish market prevalence. A result below the threshold is not repaired by changing the threshold after the fact. If delivery failed, distinguish failure of the offer from failure of the test setup.

Label an inconclusive result honestly and decide what additional observation would change your view. Record the resulting decision and next task in your workspace. The example numbers above illustrate the method; they are not a recommended universal sample size.