Who GrowthBook is for#
Product teams running A/B tests from warehouse data
GrowthBook lets teams define metrics against their existing data warehouse and connect those results to feature flag rollouts.
Skip if:
Skip it if your team has no event pipeline or wants a fully managed analytics suite with minimal SQL involvement.
Engineering teams replacing managed feature flags
Teams can self-host flag infrastructure, use SDKs across services, and automate workflows through APIs and webhooks.
Skip if:
Choose a managed flag vendor if uptime responsibility and SDK support should sit fully with an external provider.
The problem it solves#
Feature delivery and experimentation often split across multiple paid tools: one for flags, one for A/B tests, one for analytics, and another for data warehouse analysis. That fragmentation makes it harder to connect releases to product impact, and per-seat or event-based pricing can punish teams as usage grows.
Teams with strong data infrastructure also need experimentation logic that respects their metric definitions. If metrics live in a warehouse, exporting data into a vendor dashboard creates governance and accuracy problems.
How it solves it#
Feature flags with targeting
GrowthBook supports advanced targeting, gradual rollouts, and experiments, so teams can release changes to segments before full production exposure.
24 SDKs
The README lists 24 SDKs, including React, Python, Android, and iOS, which helps teams run consistent flags across web, backend, and mobile surfaces.
Warehouse-native experiments
GrowthBook queries 11 data sources, including BigQuery, Snowflake, Databricks, Redshift, and ClickHouse, so experiment analysis can use existing warehouse data.
Advanced statistics engine
GrowthBook documents CUPED, sequential testing, Bayesian methods, post-stratification, bandits, and SRM checks for teams that need more than basic conversion-rate comparisons.
Product analytics and API access
Built-in analytics, webhooks, and a REST API let teams build dashboards and automate flag workflows around the same experimentation data.
Strengths and trade-offs#
Strengths
- Flags and experiments in one workflowGrowthBook connects rollout control to measurement, so product teams can test a feature and analyze impact without moving between unrelated vendors.
- Works with existing warehousesWarehouse-native analysis is a strong fit for teams that already govern metrics in BigQuery, Snowflake, Redshift, Databricks, or ClickHouse.
- Self-hosting path is documentedThe README gives a Docker Compose quick start, which makes local evaluation and self-hosted deployment easier to validate than API-only experimentation tools.
Trade-offs
- -Open-core licensingGrowthBook says the bulk of the code is MIT licensed, but some directories use a separate commercial enterprise license. Teams need to check whether required features sit in the open or enterprise tier.
- -Data warehouse setup affects valueGrowthBook is strongest when metrics already live in a supported data source. Teams without clean event tracking or SQL-owned metrics will need data work before experiments are useful.
GrowthBook vs alternatives#
GrowthBook vs LaunchDarkly
GrowthBook and LaunchDarkly both support feature flags and controlled rollouts, but GrowthBook adds warehouse-native experimentation and a self-hosted path. LaunchDarkly is a managed proprietary feature management service with a mature hosted operations model.
GrowthBook is the better fit when a team wants feature flags tied to its own warehouse metrics and wants the option to self-host. LaunchDarkly is still the safer choice when the organization prioritizes vendor-managed reliability, enterprise support, and a fully hosted flag service over data ownership.
Install and self-host#
git clone https://github.com/growthbook/growthbook.git
cd growthbook
docker compose up -dWhat it's built on#
- Languages
- PythonTypeScript
- Frameworks
- ExpressNext.jsReact
- Databases
- MongoDBMySQLPostgreSQL
- Infrastructure
- AWS
FAQ#
Can GrowthBook be self-hosted?
Yes. The README includes a Docker Compose quick start and links to full self-hosting instructions.
Does GrowthBook only do feature flags?
No. GrowthBook combines feature flags, experimentation, product analytics, metric definitions, APIs, and warehouse-native analysis.
What is GrowthBook's license model?
GrowthBook is open core. The project states that most code is MIT licensed, with separate commercial licensing for some enterprise directories.
Similar open-source tools#
OpenFlags
Self-hosted feature flags for JavaScript with zero dependencies
Flipt
Feature flags that store and evaluate on your own infrastructure
Flagsmith
Open source feature flags and remote config, self-hosted
Unleash
LaunchDarkly alternative with SDK for 30+ languages, self-hosted
CocoIndex
Incremental data framework for AI agents.
OpenObserve
Self-hosted observability for logs, metrics, and traces