Run trustworthy experiments across your whole product.
Flags, account-level assignment, revenue-aware metrics and quality checks, with a permanent record of what your team learned and why they decided it.
No winner is ever declared for you. The evidence is shown, the decision stays yours.
Annual plan default
Treatment may be ahead by more than the 2 points you said were worth having, but the range still allows a smaller effect.
Built for how B2B actually works
Most experimentation tools assume one anonymous visitor at a time. Your customers are accounts, with several people inside them.
The right unit
Randomize by account, workspace or user, and keep everyone inside one customer on the same variant.
Honest reads
Effect sizes with their uncertainty, a minimum worthwhile effect you set, and an inconclusive state that is allowed to stay inconclusive.
Debuggable
Trace why one identity saw one variant, catch a broken split before launch, and see which flags are now debt.
Transparent plans
Metered on the identities an experiment exposes. Evaluating a flag locally is free and always will be.
3 projects, Stripe metrics, core diagnostics, 30-day raw retention.
Unlimited projects, sequential/CUPED, warehouse connector, approvals.
SSO, advanced governance, multiple warehouses, first-party relay.