Flag keys, types, and defaults ship in the same pull request as the code they gate.
Statsig alternative
Flags in git. Not a stats engine.
shipd is for teams who want release control in the repository. Statsig is for teams whose product is experimentation.
Statsig is built around metric gates, sequential testing, and warehouse-aware experiments.
Git source of truth
0 extra hops
PR environment plan
$0 to start
Where the products differ
Statsig is an experimentation platform with flags attached. shipd is the inverse: flags as code, local evaluation, no claim on your stats stack.
| Topic | shipd | Statsig |
|---|---|---|
| Job to be done | Release control in git, evaluated in-process. | Experimentation and metrics. Flags are how tests ship. |
| Flag keys | Defined in shipd.config.ts and reviewed in git. | Gates and configs are authored in the Statsig console. |
| Targeting | Dashboard today. Targeting rules in code is on the roadmap. | Targeting is tied to experiment assignment and Statsig user objects. |
| Evaluation | SDK loads a signed snapshot and resolves in-process. Postgres is not on the request path. | Statsig SDKs evaluate locally after downloading configs. The source of truth is the Statsig console. |
| Statistics | No. Variants only. | Stats engine, pulse reports, and warehouse-native experiment analysis. |
| Pricing | Free tier, then seat-based Pro from $15 per member / month. | Platform pricing. Confirm current rates on statsig.com. |
Choose shipd when
- You need flags reviewed in pull requests, not a metrics console.
- Evaluation must sit inside the request with no extra hop.
- You already have a warehouse or stats workflow and do not want another.
- You want a free tier before committing to an experimentation platform.
Stay on Statsig when
- The reason you buy flags is to run trustworthy A/B tests.
- You need sequential testing, CUPED, or warehouse experiment pipelines.
- Product and data teams live in Statsig pulse reports.
Be precise
What shipd does not claim
Targeting still lives in the dashboard
Flag keys are config as code today. Targeting rules in the repository are coming. Until then, rollouts and audiences are edited in shipd, not in git. Evaluation is still local: see local evaluation.
Not an experimentation suite
shipd evaluates variants. It is not a stats engine, and it is not trying to replace LaunchDarkly experiments, Statsig pulse, or PostHog analytics.
Not a Statsig replacement
shipd evaluates variants. It does not compute significance, holdouts, or warehouse experiment results.
Questions
- Is shipd a Statsig alternative?
- For feature flags as code, yes. It is not an alternative to Statsig's experimentation stats engine or warehouse workflows.
- Does shipd run A/B tests?
- You can serve variants. shipd does not tell you if the test won. Pair it with your own analytics if you need that.
- Does shipd evaluate flags locally?
- Yes. The SDK downloads a snapshot and checks flags in memory. Your app does not call shipd on every request.
- Is there a free plan?
- shipd has a free plan: 5 projects, 100 flags, 5 members. Pro is $15 per member / month.
Define once. Evaluate everywhere.