Measure the impact of every feature
You built it. But did it work? There’s no need to guess over results. Split automatically calculates the impact of every feature on every metric.
Forgot a metric? We can even calculate it for you after the fact.
Assign key metrics for the primary goals you want to achieve. Iterate until you nail it.
Guardrail metrics protect you from unintended consequences, revealing when new features are worse than the original.
Make every feature an experiment
Split fully integrates feature flags with the data you need to run and analyze experiments. It’s full stack, so you can support frontend, backend, and mobile teams with one platform.
Split automatically creates an experiment for every feature you code. Support simple feature rollouts, randomized experiments, and advanced user segmentation.
Experimentation best practices
Learn the requirements to experiment at scale, from statistical methods to measurement strategies.Download the O'Reilly eBook
Rigorous data you can take action on
Software is complicated. With so many variables, causation—not correlation—is what matters.
Split measures statistically significant impact in controlled, randomized experiments that you can confidently base business decisions on.
With built in best practices such as experimental review periods and sample ratio checks, you don’t need to be a data expert to run a sound experiment.
Connect your data, we’ll handle the rest
Capture data from any source with out-of-the-box integrations—Segment, mParticle, Google Analytics, and Sentry—in addition to tracking data directly from the SDK or via API.
Split manages data attribution with an advanced platform that maintains consistent experiment design across multiple devices, concurrent features, and version changes.Read the docs
Foster a culture of experimentation
Go Daddy created a culture of engineering experimentation with a rigorous rollout and impact review process across dozens of teams.Read the Case Study