Remote Config and Experiments should work as one product loop while staying separate in the docs and SDK mental model.
Use Remote Config for the active app behavior. Use Experiments when you want controlled assignment and measurement.
AppActor does not merge Remote Config and Experiments into one SDK method. Remote Config reads return resolved config items. Experiment reads return assignment objects with experiment and variant identity.
1. Start from a configuration
Create or choose a Remote Config that already drives a paywall, onboarding flow, or product-control decision.
Good candidates are configs with:
- one clear business hypothesis
- a small payload surface
- a safe fallback in the app
- an entitlement-safe implementation
2. Create an experiment from the config
Copy the payload shape into experiment variants and change only the fields you want to test.
Keep the control variant close to the current production behavior. Put the hypothesis in one or more treatment variants.
3. Test the variant before launch
Use a QA app, narrow app version, country, or entitlement rule to make the experiment safe to enter from an internal build. Then fetch assignment by experiment key:
Verify the assignment payload renders the same app surface that Remote Config used to control.
4. Launch and monitor
Set traffic, start the experiment, and watch assignment distribution, conversion, trial, refund, entitlement, and revenue impact before declaring a winner.
5. Roll the winner back into Remote Config
When the experiment is finished, turn the winning payload into a normal Remote Config rollout.
Archive or stop the old experiment so two similar payloads do not compete for the same app behavior.
Product loop