Experiments are the AppActor surface for controlled A/B testing. They are separate from Remote Config. Remote Config delivers stable server-driven values. Experiments decide whether an eligible customer receives an assignment, which variant they receive, and how the result should be evaluated.

​
What to test

  • paywall headline, proof, and CTA copy
  • monthly versus annual-first presentation
  • onboarding placement and screen order
  • offer timing and lifecycle moments
  • holdout groups for measuring incremental impact

​
Experiment anatomy

​
Guardrails

  • Keep premium access tied to entitlements.
  • Test with QA apps or narrow rules instead of unsupported manual assignment overrides.
  • Avoid changing too many payload keys in one experiment.
  • Re-fetch or invalidate app UI state when identity changes.
  • Treat experiment variant as analytics context, not as a customer permission.

Launch Experiments

Plan, configure, test, and start an experiment.

Splitting Rules

Understand traffic, experiment keys, and user assignment.