Get started with Optimizely Feature Experimentation

Optimizely Feature Experimentation lets you run experiments anywhere in your technology stack. Use it to release features safely behind feature flags, run A/B tests to compare variations, and roll out or roll back changes instantly, without redeploying code.

Before you begin

To use Feature Experimentation, you need the following:

• An active Optimizely account with Feature Experimentation enabled.
• Developer access to the application you want to instrument.
• One of the supported SDKs installed in your codebase (see the following Feature Experimentation SDKs section).

For information on how to start building and running experiments using the Feature Experimentation APIs and SDKs, see the Feature Experimentation developer documentation.

To learn more about the differences between Optimizely Web Experimentation and Optimizely Feature Experimentation, see What to expect when you migrate from Optimizely Web Experimentation to Optimizely Feature Experimentation.

Get started

Follow these steps to run your first experiment:

1. Install an SDK for your language or platform (see the following Feature Experimentation SDKs section).
2. Review the core concepts of Feature Experimentation.
3. Create a feature flag in your Optimizely project.
4. Implement the flag in your code using the decide method.
5. Define your audience and configure a flag rule (A/B test or targeted delivery).
6. Analyze results on the Experiment Results page.

Feature Experimentation SDKs

Choose the SDK that matches your application's language or platform. Each SDK exposes the same core methods (decide, track event, and others) so the concepts in this documentation apply regardless of which SDK you use.

• Android SDK
• C# SDK
• Flutter SDK
• Go SDK
• Java SDK
• JavaScript SDK – SDK version 6 and above
• JavaScript (Browser) SDK – SDK versions 5.3.5 and below
• JavaScript (Node) SDK – SDK versions 5.3.5 and below
• PHP SDK
• Python SDK
• React SDK
• React Native
• Ruby SDK
• Swift SDK

Define and debug experiments

• Design and run A/B tests and flag deliveries.
• Understand the difference between feature flags and flag variations.
• Define your audiences and attributes.
• Roll out and roll back features to subsets of your customers using targeted deliveries.
• Fine-tune your experiments by previewing your experiment variations.
• Learn how to use allowlisting.
• Learn how to troubleshoot Optimizely Feature Experimentation.

Get results

• Learn how to read and analyze the Results page.
• Choose the right metrics for your experiment.
• Integrate external analytics platforms,, like Google Analytics or Amplitude.

Best practices

• Leverage CDNs.
• Find out how datafile versioning and management works in Optimizely Feature Experimentation.
• Work with microservices.
• Create and manage staging environments.