Run an A/B Test / Experiment

Background: Native [platform] app with [remote config / experimentation
tool / none yet]. The change I want to test: [describe—e.g., new
onboarding, different paywall, button copy]. Hypothesis: [what I expect
and why].

Use case: Design a sound A/B test so I learn the truth instead of
shipping on a hunch.

Implementation details:
- A clear hypothesis, one primary success metric, and guardrail metrics
  (so a win on one thing doesn't quietly harm another)
- Variant delivery via remote config / experiment framework; consistent
  assignment per user; respecting that app cohorts update slowly (not
  everyone is on the new build)
- Sample size / duration reasoning so I don't call it too early
- Avoiding common mobile pitfalls: version-skew confounds, store-review
  delays, novelty effects
- A decision rule defined before launch (ship / kill / iterate)

Deliverable: An experiment design—hypothesis, primary + guardrail
metrics, variant setup, the assignment/exposure approach, a rough
sample-size/duration estimate, and the pre-committed decision rule.