Experimentation & feature flags · GrowthBook

GrowthBook

Open-source, warehouse-native experimentation and feature-flag platform offering both Bayesian and frequentist sequential analysis with CUPED.

GrowthBook is an open-source (MIT-licensed) experimentation platform that reads results directly from a customer's own data warehouse rather than requiring data to flow through a separate event pipeline, alongside a feature-flagging system for controlling rollouts. It supports two statistical engines chosen at the org or project level: a Bayesian engine using an improper uninformative prior by default (or an optional weakly-informative Normal prior), reporting "chance to win" and full uplift distributions as violin plots; and a frequentist engine running two-sample t-tests with optional sequential testing to guard against inflated false positives from peeking. Both engines support CUPED variance reduction, and the platform includes automatic checks for sample ratio mismatch and suspicious uplifts. It can be self-hosted for free or run on GrowthBook's managed cloud, appealing to teams that want warehouse-native experimentation without vendor lock-in.

At a glance

Vendor GrowthBook
Pricing model Free tier + paid plans
Free tier Yes
Deployment Cloud, Self-hosted
Open source Yes (MIT)
Best for Teams wanting warehouse-native, open-source experimentation with a choice of statistical engine.

Pricing

Free self-hosted open source, or free Starter cloud plan (3 users); paid Cloud Pro is $40/seat/month; Enterprise is custom for cloud or self-hosted.

Plan Price Notes
Starter (Cloud) $0 up to 3 users, 1 project, unlimited flags and experiments
Pro (Cloud) $40/seat/month up to 30 users, 3 projects, AI visual editor, multi-armed bandits
Open Source (self-hosted) $0 1 project, unlimited flags/experiments, community support
Enterprise custom cloud or self-hosted, SSO/SCIM, approval workflows, 99.99% SLA

Prices read from the vendor's own page on September 21, 2026. Vendors change prices; check the source before you budget.

Features

  • Warehouse-native experiment analysis (Snowflake, BigQuery, etc.)
  • Choice of Bayesian or frequentist statistical engine
  • Sequential testing in the frequentist engine
  • CUPED variance reduction on both engines
  • Feature flagging with targeting rules
  • Sample ratio mismatch and data-quality checks
  • Self-hosted (MIT license) or managed cloud

Integrations

Profile last reviewed September 21, 2026

Head to head

Alternatives

GrowthBook in the index now

Terms to know

Related guides