Feature flagging has grown from “a way to hide unfinished code” into a full experimentation and release-management category, and the vendor landscape shifted again recently: Split Software, once a standalone leader, was acquired by Harness and rebranded as Harness Feature Management & Experimentation (FME). That consolidation, alongside LaunchDarkly’s move to usage-based pricing and Statsig’s analytics-plus-flags bundle, makes this a good moment to compare where each platform actually stands.
At a Glance
| Platform | Best For | Starting Price | Pricing Model | Public Self-Serve Pricing? |
|---|---|---|---|---|
| LaunchDarkly | Teams wanting flags, experiments, and app/LLM observability in one product | Free (Developer tier); usage-based above that | Usage (MAU, connections, AI runs, etc.) | Yes |
| Split (now Harness FME) | Organizations already consolidating onto Harness’s DevOps platform | Contact sales | Custom, module-based | No |
| Statsig | Teams that want flags and product analytics/experimentation bundled | Free (Developer tier); Pro $150/mo | Event volume-based | Yes |
Short Verdict
LaunchDarkly offers the most generous named free tier and the clearest self-serve usage pricing of the three, now expanded well beyond flags into agent/LLM observability. Split no longer operates as an independent, self-serve product — as Harness FME, it’s sold as one module inside Harness’s broader DevOps platform, with pricing only available by contacting sales. Statsig is the one to consider if you want feature flags and product analytics (events, session replay, experimentation) from a single vendor at a flat monthly starting price.
What Each Platform Actually Is
LaunchDarkly is a feature management platform built around flags, targeting rules, A/B testing, and multi-armed bandit experiments, and it has pushed further into AI-era tooling with agent configuration, LLM observability (with LLM-as-judge evaluation), session replay, and error/trace monitoring built into the same product. Pricing is unusual in that seats are unlimited and free; you pay for usage instead, measured across service connections, client-side MAU, AI runs, session replays, errors, and trace/log volume.
Split was a standalone feature-flagging and experimentation platform until Harness acquired it in 2024; the product is now called Harness Feature Management & Experimentation (FME) and is sold as one of several modules inside the Harness DevOps suite (alongside CI, CD, Infrastructure as Code, and others). Its official documentation confirms the rebrand directly. There is no public self-serve price list for FME — both the “Essentials” and “Enterprise” Harness bundles route to “Contact Sales,” and FME itself isn’t included in the lower-cost Essentials package at all.
Statsig combines feature flags, A/B testing and experimentation, product/web analytics, and session replay in one platform, aimed at teams that want flag rollouts and the analytics to evaluate them without separate billing relationships. Its free Developer tier is usage-capped by monthly events rather than seats, and the paid Pro tier scales with additional event volume.
Pricing Breakdown
| Plan | LaunchDarkly | Split / Harness FME | Statsig |
|---|---|---|---|
| Free tier | Developer: $0 — unlimited flags, 1K client-side MAU, 5 service connections, 5K AI runs, unlimited A/B tests; 1 project, 3 environments only | Not publicly offered | Developer: $0 — 2M events/mo, unlimited flag checks, 50K session replays/mo, unlimited seats |
| Mid tier | Foundation: pay-as-you-go — unlimited projects/environments; e.g. ~$58/mo (annual) for 10 connections + 2K MAU | Not publicly offered | Pro: $150/mo — 5M events included (then $0.05/1K extra), 100K session replays/mo, unlimited retention on analytics |
| Top tier | Enterprise: custom — SSO/SAML, SCIM, audit logs, custom roles, compliance certifications, named CSM | Enterprise (Harness bundle): custom — module selection, premier support, dedicated account manager | Enterprise: custom — warehouse-native deployment, SSO, RBAC, HIPAA eligibility (BAA) |
The gap in the middle column is a material finding of this research, not an oversight: Harness’s public pricing page lists no prices for any module, including FME, and its own documentation and comparison pages from third parties consistently describe FME as quote-only. Teams evaluating Split/FME today should expect a sales conversation rather than a self-serve checkout.
Pros
- LaunchDarkly: Unlimited free seats plus a genuinely usable free usage tier, with transparent published rates for every metric once you exceed it.
- Split / Harness FME: Strong fit if you’re already standardizing on Harness for CI/CD, since flags, pipelines, and infrastructure-as-code live under one contract and admin console.
- Statsig: One bill covers flags, experimentation, and the analytics needed to judge them, instead of paying a flagging vendor and an analytics vendor separately.
Cons
- LaunchDarkly: The free Developer tier is capped at a single project and three environments, and lacks SSO, approval workflows, or audit logging — all common requirements once more than one team touches flags.
- Split / Harness FME: No published self-serve pricing at all, and it isn’t included in Harness’s lower-cost “DevOps Essentials” bundle, so even small teams must go through a sales cycle to get a number.
- Statsig: The Pro tier caps you at one project per subscription, and heavy event volume can push costs up quickly at $0.05 per 1,000 events beyond the included 5 million.
Who It’s a Good Fit For
LaunchDarkly fits teams of almost any size that want to start free and scale into usage-based billing with full transparency, especially ones also experimenting with AI agents and wanting observability for them in the same tool. Split/Harness FME fits organizations already committed to Harness’s DevOps platform who are comfortable negotiating an enterprise contract. Statsig fits product and growth teams who want flags and the analytics to measure their impact without juggling two vendor relationships.
Who Should Look Elsewhere
Teams that need enterprise controls like SSO on a tight budget may find LaunchDarkly’s jump from the free tier to Enterprise pricing jarring. Teams that want to compare self-serve pricing between vendors before talking to anyone will find Split/Harness FME frustrating, since there’s simply no public number to compare. And organizations that specifically don’t want their analytics and feature-flag data living in the same platform may prefer to keep those functions separated rather than adopting Statsig’s bundled approach.
Bottom Line
LaunchDarkly and Statsig both compete openly on published, usage-based pricing, which makes them easy to evaluate side-by-side. Split’s absorption into Harness FME has effectively removed it from that kind of self-serve comparison — it’s now a sales-led enterprise module, which may still be the right choice for Harness customers but changes the buying process entirely.
FAQ
Is Split still a separate product from Harness?
No. Harness’s own developer documentation states that Harness acquired Split in June 2024 and the product is now called Harness Feature Management & Experimentation (FME).
Does LaunchDarkly charge per seat?
No — all three LaunchDarkly plans include unlimited seats at no per-seat fee; billing is based on usage metrics like client-side MAU and service connections instead.
What counts as an “event” on Statsig’s free tier?
Statsig’s Developer plan includes 2 million events per month alongside unlimited flag and config checks; the event count specifically covers analytics/logging events rather than flag evaluations.
Can I get a trial of Harness FME without a sales call?
Based on Harness’s public pricing page, no — every paid tier, including the module containing FME, routes to “Contact Sales,” with no listed self-serve trial.
Which platform is easiest to start with as a solo developer or small team?
LaunchDarkly’s and Statsig’s free tiers are both usable without a credit card and don’t require a sales conversation, making either a lower-friction starting point than Split/Harness FME.
