GrowthBook

GrowthBook is an open-source web app and server that provides tools for feature flagging, A/B testing, and product analytics. It allows teams to manage feature rollouts and conduct experiments to make data-driven product decisions. The software is deployed as a server and can be run via Docker Compose for self-hosting or used through a cloud account.

The platform functions as a warehouse-native tool, meaning it connects directly to existing data sources rather than requiring data to be ingested into a proprietary database. This architecture allows teams to define metrics using SQL and analyze experiment results based on their own stored data. It is designed for developers, data scientists, and product managers who need a centralized system for remote configuration and statistical analysis across different environments.

Key features

• Feature flags with advanced targeting and gradual rollouts
• Support for 24 SDKs including React, Python, Android, and iOS
• Stats engine supporting CUPED, Sequential, Bayesian, and Bandits
• Warehouse native connectivity for BigQuery, Snowflake, and Databricks
• SQL-backed metric definitions for conversion rates and quantiles
• Built-in product analytics suite for creating team dashboards
• Documentation support with screenshots and Markdown
• REST API and webhooks for custom workflow integrations
• MCP server for managing features and cleaning stale flags

GrowthBook integrates with 11 different data sources, including Redshift and ClickHouse, to power its analytics. It provides a comprehensive suite for experimentation that includes Sample Ratio Mismatch (SRM) checks and post-stratification to ensure statistical accuracy. The system uses a combination of a web interface for management and a wide array of SDKs for implementation across various frontend and backend environments. By utilizing a warehouse-native approach, the software avoids the need for separate data pipelines for experimentation metrics.

This software is an open-core solution for companies seeking the flexibility of an in-house experimentation platform without the overhead of building one from scratch. It targets organizations that want to maintain control over their data while utilizing professional grade statistical tools.