Split
Feature delivery with automatic impact measurement on every flag.
What it is
Split pairs feature flags with an attribution engine that automatically measures the effect of each release on your metrics, including guardrail alerts when a rollout degrades a key indicator. That measure-everything-by-default posture is its main differentiator. Harness acquired it in 2024.
Best for
Where it is strong
- Automatic metric impact on every flag catches regressions nobody thought to measure.
- Guardrail alerts stop a bad rollout before it reaches everyone.
- Strong data-platform integrations.
Where it falls short
- Post-acquisition roadmap now sits inside Harness's broader platform strategy.
- Requires meaningful event instrumentation before the impact analysis is useful.
- Opaque enterprise pricing.
Pricing
Free tier available
Free tierFree tier: Free tier with limited monthly tracked keys.
What drives the bill: Paid tiers are quote-based, priced by monthly tracked keys.
Check current pricing on Harness’s siteFigures are indicative and were last reviewed August 2026. Vendors change pricing often; confirm before you commit.
Works with
What people who use Split say
No reviews of Split yet. If you have used it in anger, yours would be the first.
Alternatives to Split
If Split is not the right fit, these solve the same problem differently.
LaunchDarkly
A/B & Flags
The enterprise feature-flag platform, with experimentation layered on top.
Statsig
A/B & Flags
Feature flags, experimentation and product analytics in one warehouse-friendly platform.
Eppo
A/B & Flags
Warehouse-native experimentation with a serious statistics engine.