You work on a subscription-based creative software product and want to test a redesigned first-run onboarding flow that simplifies workspace setup and highlights key actions earlier. The team believes the change will increase the share of new users who activate within their first week, but there is concern that a more aggressive onboarding may increase early abandonment or support contacts. You need to design the experiment so the team can decide whether the expected lift is large enough to justify launch.
How would you design this A/B test, including the primary and guardrail metrics, the unit of randomization, the sample size needed, and the analysis plan you would pre-register so the team can make a launch decision confidently?