Ship only the right features

Split brings together feature flags and your feature performance data to turbocharge software development.

  • Faster development: Separate code deploys from feature releases
  • Safer releases: Reduce the MTTR of failed features and instantly kill them
  • Greater impact: Experiment to measure and maximize data-driven results
Feature Delivery Platform
Split combines feature flags and performance data to manage, monitor, and experiment with every feature you ship.

Go from code to customer faster

Streamline code deployment with feature flags. Keep new features dormant until you’re ready to roll out.

Once you are ready to release, Split’s sophisticated targeting makes it easy to test in production and then release a feature to segments of your user base.

Explore Feature Flags
Feature flags and progressive delivery
Feature flag control that progressively rolls out a feature to 40% of users who have access to the v3 API.

Make changes without pushing new code

Feature flags sometimes need to do more than just toggle on/off. Need to swap out copy? Alter a search algorithm?

Dynamic configurations alter feature attributes on the fly, in production, without a code change.

By abstracting changes from the code, you can also allow any approved team member (aka, a product manager) to adjust feature attributes.

Explore Dynamic Configurations

Know when it’s safe to release

Releasing new features doesn’t need to be risky. Split does what your APM can’t, telling you which feature is changing your metrics.

Split monitors every feature you release and instantly alerts you to a problem. Catch issues before users notice, and reduce the blast radius of a failure.

If you do have a failed feature, use the kill switch to disable it. You’ll never need to rollback a change.

Explore Alerting
Release alert warning
An alert caused by a feature with degrading metrics in the past 5 minutes.

We measure so you can manage

Split calculates the impact of each feature on each user and each metric. Know if your change should be rolled out widely or turned off immediately.

Split pairs your feature flags with data from most popular data pipelines. Send us any event—clicks, page load times, shopping cart values, errors. We’ll map it to the user and feature involved.

Explore Data Import + Export
Feature flag integrations
Integrations with Google Analytics, mParticle, Segment, and Sentry.

Turn any and every feature into an experiment

Feature flags? Check. Data? Check. We’ve got all the ingredients you need to A/B test new features and measure impact.

Split randomizes users, automatically attributes data, and calculates impact for every feature you put behind a flag. We figure out what works, so you can focus on the results.

With built-in statistical best practices, any member of your development team can test out new ideas directly with users.

Explore Experimentation

Fits the way development teams work

Scale across multiple teams with environments and permissions to define ownership.

Control changes and reduce errors with approval flows. And track it all with audit logs that keep a record of every change.

Integrate with popular tools—Slack, Jira, Datadog, and more—so you always know where features are in your release process.

Explore Developer Workflows
Feature flag workflow
Feature flag audit log with feature created, rolled out to QA, rolled out to 5% of users, and killed.

Built for enterprise performance, security, and resilience

Split’s high-performance, streaming architecture pushes flag and experiment changes via CDN to our SDKs in milliseconds.

Our SDKs target and evaluate feature flags locally, so customer data is never sent over the internet.

We build resilience into every aspect of our service. Our SaaS app, data platform, and API span multiple data centers. And our SDKs cache locally to handle any network interruptions.

Explore Platform

Deliver impact with every feature you build

Speed up development cycles, reduce release risk, and focus your team on the features that create maximum impact.