GrowthBook is an open source feature flagging and A/B testing platform you can self-host — a LaunchDarkly and Optimizely alternative that runs experiments against your own data warehouse with a built-in statistics engine.

What is GrowthBook?

GrowthBook is an open source platform for feature flags and A/B testing. You wrap code in flags to turn functionality on or off without redeploying, then run experiments and measure them with a built-in statistics engine that queries your own data warehouse — so the numbers come straight from your data, not a vendor’s black box.

What is GrowthBook best for?

Product and engineering teams that want to ship behind flags, roll out changes gradually, and prove their impact with rigorous experiments. It fits especially well for data teams that already have a warehouse — BigQuery, Snowflake, Databricks, or Postgres — and want experiment analysis layered on top of it instead of buying a separate analytics SaaS.

What can GrowthBook do?

Manage feature flags with targeting rules, environments, and percentage rollouts
Run A/B tests and multi-arm bandits on a Bayesian and frequentist stats engine (CUPED, sequential testing, SRM checks)
Query 11+ data warehouses directly — “warehouse native,” so your event data never leaves your infrastructure
Define metrics in SQL and build product analytics dashboards
Drop into your stack with 24 SDKs (JavaScript/React, Python, PHP, Ruby, Go, Java, Kotlin, Swift, and more)
Import existing flags from LaunchDarkly, and automate via REST API, webhooks, and an MCP server

Is GrowthBook free?

Yes — GrowthBook is free to self-host under the MIT license, with unlimited feature flags, experiments, and traffic. GrowthBook Cloud adds a free Starter plan for up to 3 users; Pro is $40 per seat per month (up to 50 users) and unlocks advanced stats, the visual A/B test editor, and product analytics, while Enterprise is custom-priced with SSO/SAML, audit logs, and SLAs.

Where does GrowthBook fall short?

It’s open core, not fully open source — the visual A/B test editor, advanced statistics, SSO/SAML, and audit logs sit behind a paid Pro or Enterprise license.
Self-hosting requires MongoDB (and optionally Redis) for the control plane, which is more moving parts than flag-only tools that run on Postgres alone.
Flag management is solid but less deep than dedicated flag tools — complex approval workflows, scheduling, and flag dependencies are thinner.

What does GrowthBook replace?

GrowthBook is a self-hosted alternative to LaunchDarkly for feature flagging and Optimizely for experimentation. It covers both jobs in one tool — flags plus a real stats engine — and because it queries your warehouse directly, you avoid per-seat SaaS pricing and never ship event data to a vendor.

FAQ

Is GrowthBook open source? Mostly — the bulk of the code is MIT licensed and free to self-host, but some features (the visual editor, advanced statistics, SSO) live in directories under a separate commercial license. It’s open core rather than strictly OSI open source.

Can I self-host GrowthBook for free? Yes. The self-hosted open source edition is free and includes unlimited flags, experiments, and traffic. You only pay for Enterprise features or for using the managed Cloud beyond its free tier.

Is GrowthBook a good LaunchDarkly alternative? For teams that also want experimentation, yes — it does feature flags plus warehouse-native A/B testing in one tool and can import your existing LaunchDarkly flags. For pure flag-management depth (approvals, scheduling), LaunchDarkly is still ahead.

What do I need to run GrowthBook? Docker, a MongoDB database for GrowthBook’s own data, and a connection to your data warehouse (BigQuery, Snowflake, Postgres, etc.) if you want experiment analysis. Redis is optional and only needed for scaling.