Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

GrowthBook

growthbook.io

  • Free plan
  • Open source
  • Web
  • iPhone and iPad
  • Android
  • Browser extension
  • API
  • Self-hosted
8.2 Club score Visit growthbook.io

What GrowthBook is

GrowthBook is a warehouse-native platform for experimentation, feature flags, and product analytics. It supports A/B tests, safe rollouts, and multi-arm bandits, with Bayesian, frequentist, sequential, and CUPED/variance-reduction methods. Its feature-management tools provide unlimited feature flags, percentage rollouts, instant kill switches, and advanced targeting. GrowthBook analyzes data in the warehouse where it already resides and keeps SQL visible. Listed warehouse integrations include Snowflake, BigQuery, Databricks, ClickHouse, Trino, and Adobe Experience Platform. The platform describes more than 24 SDKs, including JavaScript, Python, React, Swift, and Kotlin, and supports web and mobile use, server-side testing, visual editing, and mobile SDKs. Teams can use managed cloud software or self-hosted deployment, including air-gapped installations on major clouds or on-premises. GrowthBook says analysis uses aggregate data without personally identifiable information leaving the warehouse. It lists SOC 2 Type II certification, GDPR, COPPA, and CCPA compliance, plus encryption at rest and in transit. Starter Cloud is free for up to three users and one project; Pro Cloud costs $40 per user per month.

Who it suits

Starter is aimed at individuals or small teams, while Pro suits small and mid-sized teams. Enterprise options address teams operating at scale, including organizations needing self-hosting, custom environments, access controls, or dedicated support.

Good to know, and what to watch

  • Warehouse-native analysis keeps data where it lives
  • A/B tests and multiple statistical methods
  • Unlimited feature flags and traffic
  • Safe rollouts include auto-rollback on Pro
  • More than 24 SDKs
  • Cloud or self-hosted deployment options
  • Starter Cloud allows up to three users
  • Starter Cloud is limited to one project
  • Pro costs $40 per user per month
  • Some rollout and enterprise controls require higher plans

Enterprise

—

Enterprise (Cloud)

—

Enterprise (Self-Hosted)

—

Open Source

Free

Free plan

Getting started

  1. Open the GrowthBook website
  2. Choose managed cloud or self-hosted deployment
  3. Select the free Starter Cloud, free Open Source self-hosted, or Pro Cloud plan
  4. Connect a listed data warehouse
  5. Integrate an available SDK such as JavaScript, Python, React, Swift, or Kotlin

Limits to check

Open Source self-hosted provides one project, while Starter Cloud allows up to three users and one project. Pro Cloud allows up to 30 users and three projects; all three plans list unlimited feature flags, experiments, and traffic.

The Club's review

GrowthBook is a warehouse-native platform that combines experimentation, feature flags, and product analytics. It suits product and engineering teams that need visible SQL, broad SDK coverage, and deployment flexibility. Its strongest case is privacy-conscious experimentation tied directly to existing warehouse data, while its cloud plans impose clear user and project limits.

Overview

GrowthBook analyzes data in the warehouse where it already lives, exposing SQL rather than moving analysis data elsewhere. The company says only aggregate data is accessed and that personally identifiable information stays in the warehouse. That architecture is a practical fit for organizations with established Snowflake, BigQuery, Databricks, ClickHouse, Trino, or Adobe Experience Platform data pipelines.

Founded in 2020, GrowthBook supports web and mobile products through Android, iOS, API, web, and self-hosted options. It offers a managed cloud service or self-hosting, including air-gapped deployments on major clouds and on-premises installations.

It belongs among Experimentation Platforms, A/B Testing Tools, and Feature Flag Software.

Key features

Experimentation and analysis

GrowthBook supports A/B tests and multi-arm bandits, with Bayesian, frequentist, sequential, and CUPED or other variance-reduction methods. That range lets teams choose an analytical approach rather than being locked into one testing model. Server-side testing, a visual editor, audience targeting, and a mobile SDK cover both product interfaces and backend experimentation.

Flags and rollout control

Unlimited feature flags, percentage rollouts, instant kill switches, and advanced targeting support controlled releases without imposing a flag-count ceiling. Pro adds safe rollouts with automatic rollback, while Enterprise adds ramp schedules and approval workflows. Those higher-tier controls matter for teams that need formal release governance; the free tiers are better suited to straightforward flagging and experimentation.

SDKs, security, and privacy

GrowthBook describes more than 24 SDKs, including JavaScript, Python, React, Swift, and Kotlin, giving mixed web, mobile, and service stacks a broad integration surface. It says the service is SOC 2 Type II certified and compliant with GDPR, COPPA, and CCPA, with encryption at rest and in transit. Keeping personally identifiable information out of analysis is a meaningful advantage for privacy-sensitive data teams.

Pricing

GrowthBook uses a freemium model, with paid plans starting at $40/user/mo.

Open Source (Self-Hosted)

This plan costs nothing and includes one project, unlimited feature flags, unlimited experiments, and unlimited traffic. It is the clearest choice for organizations willing to operate the platform themselves and that need no paid cloud support.

Starter (Cloud)

Starter costs nothing and allows up to three users, one project, unlimited flags, unlimited experiments, and unlimited traffic. It is aimed at individuals or small teams. The three-user ceiling and single project make it a useful evaluation or small deployment, not a broad departmental rollout; community support is included.

Pro (Cloud)

Pro costs $40 a month, billed per seat per month. It supports up to 30 users and three projects, while retaining unlimited flags, experiments, and traffic. Premium support and safe rollouts with auto-rollback make it the practical tier for small and mid-sized teams, but seats and projects remain capped.

Enterprise

Enterprise Cloud and Enterprise Self-Hosted use custom pricing tailored to scale and licensing requirements. Cloud adds custom environments, SSO and SCIM provisioning, dedicated support and training, and a 99.99% uptime SLA. Self-hosted adds custom environments, advanced access control, SSO and SCIM provisioning, and dedicated support and training. These are the options for large or tightly governed deployments, rather than buyers seeking a published price.

Platforms

GrowthBook is available on Android, iOS, web, API, and self-hosted deployments. Its cloud or self-hosted model, including air-gapped and on-premises operation, gives infrastructure teams more control than a cloud-only service. The SDK coverage is especially relevant when experiments span browser, mobile, and server code.

Who it's for

Starter fits an individual or small team with one project and no more than three cloud users. Pro is better for small and mid-sized teams that need premium support, multiple projects, and safer automated rollouts. Enterprise suits organizations operating at scale that require custom environments, identity provisioning, governance, support, or a formal uptime commitment. Self-hosting is the sensible route for teams that must keep deployment inside their own infrastructure.

Pros and cons

Pros

  • Warehouse-native analysis keeps SQL visible and avoids moving personally identifiable information out of the warehouse.
  • More than 24 SDKs and named JavaScript, Python, React, Swift, and Kotlin options cover varied product stacks.
  • Unlimited flags, experiments, and traffic appear across the free and Pro plans, so usage volume is not the pricing limiter.
  • Self-hosted, air-gapped, and on-premises deployment supports strict infrastructure requirements.
  • Pro and Enterprise add progressively stronger rollout safety, governance, support, and access controls.

Cons

  • Starter Cloud is limited to three users and one project, which quickly constrains growing teams.
  • Pro's $40 per-seat monthly pricing and 30-user, three-project caps make expansion a budget and administration decision.
  • Enterprise pricing is custom, so larger buyers cannot estimate cost from a published rate.
  • Starter relies on community support; premium or dedicated assistance requires a paid tier.

Alternatives

Confidence by Spotify is a better fit for teams wanting a free plan with unlimited seats, 75,000 monthly credits, up to five surfaces, and Google SSO.

Statsig suits teams prioritizing a free Developer tier with 2M events per month, unlimited seats, session replays, and one-year analytics retention.

Marpipe is an alternative for catalog and generative creative workflows.

AB Tasty is a paid alternative with a free trial and plans listing unlimited experiments, variations, and metrics.

Bucketeer is the simpler open-source, self-hosted choice with no per-seat pricing.

Amplitude is suited to teams seeking a free analytics plan capped at 2M events per month with session replays.

RunPivot is a web-focused option with a Startup plan priced at $99 a month when billed annually.

VWO Feature Experimentation is a paid, trial-based alternative with feature flags, environments, reporting, webhooks, and basic support.

Verdict

Choose GrowthBook when experimentation, feature management, and analytics must connect directly to warehouse data, especially across web, mobile, and server stacks or controlled infrastructure. Its transparent SQL, privacy positioning, unlimited core usage, and deployment choices are compelling. Look elsewhere if three Starter users are insufficient, if you need Pro controls without paying per seat, or if you require a predictable Enterprise price.

The verdict

Choose GrowthBook if you need experimentation, feature management, and analytics tied to warehouse data. It fits teams that value visible SQL, broad SDK coverage, and cloud or self-hosted deployment. Look elsewhere if your team needs more than three Starter users or wants Pro-level controls without a paid plan.

Questions people ask

Is GrowthBook free?

Yes. Open Source self-hosted and Starter Cloud are free plans. Pro Cloud costs $40 per user per month.

Which deployment options are available?

GrowthBook is offered as managed cloud software or self-hosted software, including air-gapped deployment on major clouds or on-premises.

What data warehouses integrate with GrowthBook?

Listed integrations include Snowflake, BigQuery, Databricks, ClickHouse, Trino, and Adobe Experience Platform.

Does GrowthBook support mobile products?

Yes. It supports web and mobile use and lists mobile SDK support, including Swift and Kotlin options.

What security and privacy measures are listed?

GrowthBook lists SOC 2 Type II certification and GDPR, COPPA, and CCPA compliance, with encryption at rest and in transit. It says only aggregate data is accessed for analysis and personally identifiable information stays out of the warehouse analysis flow.

What the maker publishes

API
The project documents a full REST API, separate management and SDK endpoints, webhooks and OpenAPI documentation. source
Audience
The pricing page describes Starter as for individuals or small teams, Pro for small and mid-sized teams, and Enterprise for teams operating at scale. source
Chrome extension limits
GrowthBook DevTools requires the GrowthBook JavaScript, React or HTML Script Tag SDK installed on the site. source
Cloud hosting
GrowthBook Cloud is hosted on AWS and Enterprise Cloud includes a 99.99% uptime SLA. source
Company facts
LinkedIn lists GrowthBook as privately held, founded in 2020, headquartered in Palo Alto and employing 11–50 people. source
Company legal name
The Chrome Web Store publisher and website footer identify the maker as GrowthBook Inc. source
Current service status
The official status page reports the Cloud App, API and CDN operational with 100.0% uptime and no notices in the past seven days. source
Data architecture
GrowthBook says its warehouse-native architecture analyzes data where it lives and makes SQL visible. source
Deployment
GrowthBook is available as a managed cloud service or self-hosted, including air-gapped deployment on major clouds or on-premises. source
Developer integrations
GrowthBook lists Slack, Discord, GitHub, Datadog and Vercel Marketplace integrations. source
Documentation depth
The docs include quickstart, API, webhooks, integrations, security, self-hosting, SDK, tutorials and agent guides. source
Event integrations
Event and tracking integrations include FullStory, Firebase, Segment, GA4 and RudderStack. source
Experimentation
The platform supports A/B tests and lists Bayesian, frequentist, sequential testing, and CUPED/variance reduction among its statistical methods. source
Experimentation
The experimentation product supports multivariate, sequential and bandit tests with Bayesian or frequentist analysis. source
Feature flags
Feature flags support targeting, gradual rollouts, low-latency delivery and rollback control. source
Feature management
Feature management includes unlimited feature flags, percentage rollouts, instant kill switches, and advanced targeting. source
Founded
2020 source
Funding
Crunchbase lists two funding rounds and three investors, naming Y Combinator and Data Tech Fund among the most recent investors. source
Installation
The repository provides Docker Compose self-hosting instructions that start the public image and expose GrowthBook locally on port 3000. source
Integrations
Listed warehouse integrations include Snowflake, BigQuery, Databricks, ClickHouse, Trino, and Adobe Experience Platform. source
License
GrowthBook uses an open-core model: most code is MIT licensed while some directories use the commercial GrowthBook Enterprise License. source
Limit: technical learning curve
G2 reviewers report that experiment setup, data validation and onboarding require technical comfort and can take time. source
Limit: UI and documentation
G2 reviewers mention confusing or barebones UI, difficult initial integration and gaps in documentation for multi-organization setups. source
MCP server
The MCP server can create features, start experiments, clean stale flags and query product analytics from MCP clients. source

Checked against the maker's pages on 15 Oct 2026.