Amplitude Experiment Overview
Welcome to Amplitude Experiment. This page acts as a quick reference as well as a high level system overview of Experiment's end-to-end feature-flagging and experimentation platform.
-
Getting started
Read the step-by-step guide to get started developing with Amplitude Experiment.
-
System overview
Understand about how Amplitude Experiment works within your system.
-
Experiment SDKs
Find the right Experiment SDK for your app.
-
Experiment REST APIs
Explore the APIs for Amplitude Experiment.
SDKs¶
Client-side SDKs work in a single-user context. Server-side SDKs work in a multi-user context.
Client-side¶
Server-side¶
REST APIs¶
| API | Description |
|---|---|
| Evaluation API | Evaluate a user for the feature flags and experiments assigned to the deployment used to authorize the request |
| Management API (Beta) | Manage or list flags and experiments within your organization. |
System overview¶
When it comes to feature flag delivery and experimentation, one size does not fit all. Amplitude Experiment is built for flexibility to fit in with any architecture and a variety of needs.
That said, you can generally split experimentation and feature-flagging systems into client-side and server-side architectures.
Client-side¶
Easiest way to get started delivering flag and experimenting on your client side application.
Client-side experimentation and feature-flagging involves the client making a request to fetch flags and experiments from Amplitude's remote evaluation servers when the application is initialized.
Server-side¶
Remote evaluation¶
Simple and flexible API & SDKs made to fit into any system.
Server-side remote evaluation involves making a request from your server to Amplitude Experiment's evaluation servers to fetch variants for a user. Use the resulting variants directly on the server or passed back for use on the client (or any other part of your system).
Local evaluation¶
Fast evaluation for performance minded systems.
Server-side local evaluation runs evaluation logic on your server, saving you the overhead incurred by making a network request per user evaluation. The sub-millisecond evaluation is perfect for latency-minded systems which need to be performant at scale. However, because evaluation happens outside of Amplitude, advanced targeting and identity resolution powered by Amplitude Analytics isn't possible.
