Create an experiment card
Record the hypothesis, participants, recruitment method, task, timing, owner and stopping conditions. Set evaluation criteria before viewing results to avoid choosing favourable criteria afterwards.
Measure behaviour
Prioritise task completion, requests for help and reasons for abandonment. Record the numerator, denominator and collection method for every rate. Observe the baseline before setting targets.
Reduce bias
Use consistent instructions across sessions. Note when a facilitator helps. Separate internal tests from sessions with target users instead of combining them to improve the reported result.
Interpret the results
Report whether the hypothesis was supported, unsupported or still lacks evidence. Explain sample limitations, technical failures and contextual factors. Link conclusions to notes or original data you have permission to retain.
