Split (Harness FME)
Feature-flag and experimentation platform that ties flags to product metrics, acquired by Harness and folded into its Feature Management & Experimentation module.
Acquired by Harness (2024); impact-measurement-on-flags pioneer.
Ownership
- Acquired by Harness · closed Jun 2024 · source · In the Martech Market Report
Harness, a software delivery platform company, signed a definitive agreement to buy feature management and experimentation vendor Split Software on 29 May 2024 and completed the deal on 11 June 2024.
Closest Split (Harness FME) alternatives
Ranked by reviewed relationship first, then by the capabilities they share with Split (Harness FME).
Wingify (AB Tasty + VWO) experience-optimization platform offering web experimentation, feature experimentation, and…
Enterprise personalization and A/B testing platform for delivering targeted experiences across web, mobile, and email…
Warehouse-native experimentation platform known for rigorous statistics (CUPED, sequential testing), now operating as…
Split (Harness FME) Pros & Cons
Key strengths and limitations to consider, drafted by an AI model from public sources
Strengths
- SDKs evaluate flags locally without a network call per decision.
- Supports backend, web, mobile, and cross-platform experiment assignment.
- Connects exposure data to outcome events for feature-level measurement.
- Runs warehouse-native analysis in Snowflake, BigQuery, and Redshift.
- Adds Harness RBAC, approvals, policies, and audit trails to flag changes.
Limitations
- Requires event and metric instrumentation or a connected event source.
- Lacks a visual WYSIWYG editor for marketer-led web experiments.
- Cloud and warehouse-native experimentation use separate project setups.
- Broadest value depends on adopting adjacent Harness delivery workflows.
Ideal For
Who benefits most from Split (Harness FME), drafted by an AI model from public sources
Quick Analysis
Harness FME (formerly Split) sits in the feature-management-led experimentation market: it uses SDK-based feature flags and deterministic assignment to control releases, capture exposure, and measure outcomes. It is closer to LaunchDarkly, Statsig, and Optimizely Feature Experimentation than to marketer-led visual testing products such as VWO or AB Tasty. Its defining model is experimentation embedded in software delivery rather than page-level optimization.
The product is strongest for engineering-led organizations that need to progressively release backend and client-side changes while measuring product, reliability, or commercial metrics. Local SDK evaluation, cross-SDK sticky bucketing, event ingestion, release monitoring, and warehouse-native analysis are meaningful differentiators versus LaunchDarkly's primarily feature-management focus and Optimizely's broader but more web-optimization-oriented portfolio. Its integration into Harness also makes governance, approvals, audit trails, and deployment workflows more compelling for existing Harness customers.
Evaluate Harness FME when product experimentation must share controls with progressive delivery, especially for full-stack services and regulated release processes. Choose Statsig or Amplitude Experiment when product analytics is the primary experimentation workspace; choose Optimizely or VWO for no-code web experimentation. Before buying, validate event instrumentation quality, metric ownership, cloud-versus-warehouse experimentation design, SDK coverage, and whether Harness platform adoption adds value beyond FME alone.
A SaaS team releases a redesigned onboarding flow to 5% of new accounts and measures activation.
A payments team canaries a backend authorization change while monitoring latency and failure rates.
A mobile product team tests checkout variants with consistent assignment across iOS, Android, and web.
An enterprise migrates read traffic to a new service through dual reads, staged ramping, and rollback.
An AI product team compares prompts, model settings, or temperature values against cost and conversion metrics.
Pricing
Model
custom
Estimated cost
from about $7,200 per year
Third-party estimate of what buyers typically pay. The vendor publishes no price.
Harness lists a $0 Free Plan for Feature Management & Experimentation, aimed at individual developers and small teams; the usage limits aren't published on the pages reviewed.
Checked and not offered
3 capabilities that other experimentation tools are built around, which Split (Harness FME) does not appear to cover. Each was checked against Split (Harness FME)'s own documentation in three separate AI passes, and only a majority verdict is recorded — the count is on each row.
A gap is not a fault — a product that does one job well is a reasonable thing to buy. It is only expensive when you find out after signing.
Integrate & automate documentation lists APIs, webhooks, and integrations but no MCP or agent tool interface.
Checked: https://help.split.io/hc/en-us/categories/360001538192-Integrate-automate· 3 of 3 passes agreed it is absent
Documentation index checked for session recording and replay.
Checked: https://help.split.io/hc/en-us/sections/360003839532-Documentation· 2 of 3 passes agreed it is absent
Documentation index checked for survey, feedback collection, and survey analysis.
Checked: https://help.split.io/hc/en-us/sections/360003839532-Documentation· 2 of 3 passes agreed it is absent
Read from public documentation, and documentation lags shipping. If you work at Split (Harness FME) and this is out of date, and we will record the correction against your name.
Does something in your stack already do what Split (Harness FME) does?
The Instant Stack Audit reads your tools against a capability map and shows where two of them are being paid for to do one job.
Takes about two minutes. Your overlaps and a savings estimate are free — no sign-up. An email address unlocks the costed roadmap.
Add Split (Harness FME) to Your Stack
Use our visual stack builder to see how Split (Harness FME) fits with your other tools. Plan data flows, identify gaps, and share with your team.