Feature flags + experimentation built for engineering teams — acquired by Harness in June 2024 and now branded Harness Feature Management & Experimentation
Split.io is a feature-flagging and experimentation platform aimed at engineering-led product teams — the closest direct competitor to LaunchDarkly throughout the late 2010s and early 2020s. The product was founded in 2015 in Redwood City, California by Adil Aijaz (CEO), Patricio Echague (CTO) and Trevor Stuart (President), backed by Lightspeed Venture Partners and others, and raised $100M+ in venture funding across Seed through Series D before its exit. On 11 June 2024 Harness completed its acquisition of Split Software (announced 29 May 2024), and the product is now sold inside the Harness platform under the name Harness Feature Management & Experimentation (FME). The standalone Split.io brand and console still work for existing customers, but new procurement now flows through Harness contracts. For Indian product and engineering teams, Split is the right call when you want a mature feature-flag platform with a real built-in experimentation engine, you already use Harness for CI/CD, or you want a credible alternative to LaunchDarkly without going pure open-source.
Split.io — formally Split Software — is a feature-flagging and experimentation platform that pioneered the "feature flags as a measurement layer" idea: every flag is also a treatment in an experiment, and every flag rollout is an opportunity to measure the impact of the change on user-level metrics. This was less common when Split launched in 2015 — most flag platforms (LaunchDarkly included, in its early years) were release-control tools first and measurement tools second. Split flipped that ordering, and the design has aged well; modern feature-flag platforms now uniformly bundle experimentation in some form.
The company was founded in Redwood City, California, by Adil Aijaz, Patricio Echague and Trevor Stuart — Aijaz and Stuart had previously worked together on RingCentral's growth and analytics stack, and Echague brought engineering-platform pedigree from prior roles. Lightspeed led the early rounds, and Split went on to raise $100M+ in total venture capital across Series A through D, including a $50M+ Series D in the early 2020s.
On 29 May 2024, Harness announced its intent to acquire Split, and the deal closed on 11 June 2024. Harness — the developer-platform company founded by Jyoti Bansal (also founder of AppDynamics) — was previously focused on CI/CD, deployment automation and chaos engineering, but had no native feature-management or experimentation product. The Split acquisition closed that gap in one move. The product was renamed Harness Feature Management & Experimentation (FME) and is now both sold standalone and bundled inside Harness's broader developer-platform contract.
For Indian buyers in 2026, the practical change is in the procurement path. If you're already a Harness customer, FME shows up as an add-on to your existing contract — usually quicker and cheaper than buying Split standalone was. If you're not on Harness, the standalone product still works, but the product roadmap is now visibly Harness-shaped (deeper CI/CD integrations, deployment-pipeline-aware flag rollouts), which may or may not align with your stack.
Mature server-side and client-side feature-flag SDKs in 15+ languages including JavaScript, Python, Java, Go, Ruby, .NET, PHP, Node, iOS, Android. Targeting rules support segments, geographies, percentage rollouts, and gradual ramps with automatic guardrails.
Every feature flag can be turned into an A/B experiment. Built-in statistical engine computes treatment effect, confidence intervals and significance against your defined success metrics — no need to bolt on a separate analytics tool.
Send Split your event stream and it computes per-treatment metric impact automatically. Guardrail metrics (reverse-direction signals like errors, latency) are first-class — Split alerts you if a flag rollout is silently degrading a metric you didn't plan to test.
Gradual rollouts (1% → 5% → 25% → 50% → 100%), kill-switch, scheduled rollouts, automatic rollback on guardrail breach. Standard release-engineering workflows now codified inside the platform.
Post-acquisition, Split is being deeply integrated with Harness CD: flag flips can be tied to deployment events, deployment pipelines can pause and resume based on flag-experiment outcomes, and flag changes are tracked in the same audit log as deployments.
SSO/SAML, granular role-based access, audit log, data-residency options for EU customers, SOC2 Type 2. The bar for procurement at large enterprises — important if your buyer is an Indian financial-services or healthcare team.
Split has historically had a free Developer tier (limited keys + monthly tracked users) and per-seat / usage-based paid tiers. Post-Harness acquisition, the published price card has been pulled in favour of custom Harness contracts. Real-world deals in 2025–2026 typically land in the $25K–$150K+ per year range depending on monthly active users, number of feature flags, and which experimentation modules you license. For Indian buyers that's roughly ₹20L–₹1.25Cr+ per year all-in. The page's previous "From ₹2,772/user/month" figure (~$33/user) appears to come from an old Split-standalone Team plan that no longer applies post-acquisition. Always validate via a quote on split.io or through your existing Harness AE.
Split is the wrong call when: you specifically don't want vendor lock-in (use GrowthBook open-source), you need warehouse-native experimentation analysis (use Eppo or Statsig), or your need is pure A/B testing on a marketing site (use VWO or Convert.com).