AI visibility report
Harness ranks #1 in Feature Flags & Experimentation AI search.
Outside the top three on 10 of the 25 prompts buyers actually ask.
Flagsmith is cited on 7 of those losses.
Free trial. Setup comes pre-filled for Harness.
Also benchmarked
Harness appears in another vertical
Track Harness across these prompts daily.
Start free trialBest among 12 vendors · still absent from 64.7% of tracked prompt responses
Top-3 citations across 150 prompt × platform pairs
Peer Ranking
Key Metrics
Platform Breakdown
Grok
Google AI Mode
Gemini Search
Bing Copilot
Perplexity
ChatGPTMost visible, not fully covered. Harness has the most presence among 12 vendors, but appears in only 35.3% of tracked prompt responses. Presence is absolute coverage; share of voice shows how much of the citation pool it owns.
Where Harness is losing
Prompts where competitors are visible and Harness is not.
These prompt-level losses are the first prompts to track and repair.
Where Harness is winning5
Which feature flag platforms handle anonymous visitor evaluation well without identity stitching problems?
Avg # 1.3 · 3 platforms
Which feature flag platforms let product and engineering collaborate on targeting rules without requiring a redeployment every time a rule changes?
Avg # 2.0 · 1 platform
Which feature flag platforms cache the last known flag state locally so applications keep working even if the flag service goes down?
Avg # 2.5 · 2 platforms
Which feature flag tools integrate with incident management workflows so a flag can be killed automatically when an error rate spike is detected?
Avg # 3.5 · 4 platforms
What tools do teams use to set up their first A/B test on a production feature — data layer, targeting, and metrics tracking in one place?
Avg # 3.7 · 3 platforms
Where Harness is losing5
Which feature flag platforms offer a great local development experience without requiring engineers to connect to a remote service every run?
Competitors on 4 platforms
Track this promptWhich feature flag platforms have the best OpenFeature support for teams looking to avoid vendor lock-in?
Competitors on 4 platforms
Track this promptWhich production-grade feature flag platforms offer the strongest SLA and uptime guarantees?
Competitors on 4 platforms
Track this promptWhich feature flag platforms support multi-variate experiments with built-in statistical significance calculations so you don't need a separate experimentation tool?
Competitors on 3 platforms
Track this promptWhat feature flag tools support the full lifecycle — create, roll out, and safely clean up flags — with built-in guardrails for stale flag removal?
Competitors on 3 platforms
Track this prompt
Track Harness daily before the next report refresh.
Track these gapsResearch dossierCapabilities, use cases, sources, reviews, pricing, and FAQ
Overview
Harness is an AI-powered software delivery platform founded in 2017 and headquartered in San Francisco. In June 2024, Harness acquired Split.io—a mature feature management and experimentation platform that had raised over $110M—integrating it as Harness Feature Management & Experimentation (FME). FME combines feature flag delivery and control with built-in measurement and learning tools, supporting continuous and progressive delivery practices. The platform serves over 50 billion flag evaluations to more than 2 billion end users daily via a streaming SDK architecture that evaluates flags locally for performance and data privacy. FME includes cloud experimentation, warehouse-native experimentation on Snowflake and Redshift, and an AI Release Agent for interpreting results. It operates as one module within Harness's broader 15-module SDLC platform, used by enterprises including Experian, ADP, Comcast, SAP, and Salesforce.
Harness Feature Management & Experimentation (FME), built on the acquired Split.io platform, is an enterprise-grade feature flag and experimentation system embedded within the Harness AI DevOps platform. It enables software teams to decouple code deployment from feature release, run controlled A/B and multivariate experiments with a built-in statistical engine, and monitor each gradual rollout for performance regressions—all from a single tool integrated natively into CI/CD workflows and supported by an AI agent for result interpretation.
Key Facts
- Founded
- 2017
- HQ
- San Francisco, CA, USA
- Founders
- Jyoti Bansal, Rishi Singh
- Employees
- 1500-1650
- Funding
- ~$570M equity
- ARR
- ~$250M+
- Valuation
- $5.5B
- Status
- Private
Target users
Key Capabilities10
- Feature flags with flexible targeting rules (individual user, segment, percentage-based, and attribute-based rollouts)
- Automated release monitoring with out-of-the-box performance and error metrics tracked per flag from first rollout
- A/B and multivariate experimentation with statistical engine supporting sequential, fixed-horizon, and dimensional analysis
- Warehouse-native experimentation running directly on Snowflake and Amazon Redshift without ETL
- AI Release Agent for natural-language experiment result summarization and guided rollout decisions
- Local SDK flag evaluation for sub-millisecond latency with no sensitive user data sent to the cloud
- Global SaaS architecture serving 50B+ flag evaluations daily to 2B+ end users
- Native CI/CD pipeline integration with built-in governance, change request workflows, and OPA policy-as-code
- Flag lifecycle management with automated cleanup tracking to reduce feature flag technical debt
- Multi-environment management with streaming architecture pushing changes to SDKs in milliseconds
Key Use Cases8
- Progressive delivery and gradual feature rollouts to reduce production release risk
- A/B and multivariate experimentation to measure feature impact on business and guardrail metrics
- Canary and blue/green releases with instant kill-switch rollback capability
- Infrastructure migrations with controlled traffic routing by percentage to minimize disruption
- Beta testing programs targeting specific user segments, accounts, or geographic regions
- Entitlement management and API rate limiting differentiated by customer subscription tier
- Automated detection and alerting of performance regressions during gradual feature rollouts
- Consolidating experimentation across engineering, product, and data science teams into a single platform
Harness customer outcomes
75% faster deployments
United Airlines adopted Harness CI/CD and reported significantly accelerated deployment times, gaining governance policy controls and deployment guardrails for developer teams.
80-to-1 reduction in developer effort for pipeline feature implementation
Ancestry used Harness to implement new pipeline features once and automatically extend them across every pipeline, dramatically reducing the developer effort required.
20–40% increase in post-release support cases reduced to near zero incidents
Adobe Workfront used Split (now Harness FME) to monitor and control feature releases, eliminating the spike in support cases and incidents previously observed during code releases.
Recent Trend
How AI describes Harness3
FeatBit (lifecycle contracts + stale detection + agent‑assisted cleanup) Everything else (Split/Harness, Unleash, Statsig, GrowthBook) handles creation and rollout but does not provide safe, automated stale‑flag cleanup.
What feature flag tools support the full lifecycle — create, roll out, and safely clean up flags — with built-in guardrails for stale flag removal?
Harness Feature Flags — strong CI/CD integration but not warehouse‑native.
Which feature flag platforms integrate natively with popular data warehouses so experiment results flow directly into the analytics stack?
Short answer: The platforms that truly combine feature flags + full experimentation in one unified system are Statsig, Split, Optimizely\_Feature\_Experimentation, GrowthBook, and (to a lesser extent) Harness.
Which platforms combine feature flags and full experimentation in one tool — and when do teams actually need a dedicated experimentation platform on top?
Most cited sources8
- H76
Feature Flag Tools Compared: 10 Platforms for Safer Releases
harness.io·Product Page
- H3
Feature Management & Experimentation | AI Powered | Harness
harness.io·Product Page
- H2
Feature Flags | Harness Glossary
harness.io·Blog Post
- H2
Harness: AI for DevOps, Testing, AppSec, and Cost Optimization
harness.io·Product Page
- H2
Get Started with Feature Flags in Node Blog - Harness IO
harness.io·Blog Post
- D2
Manage stale flags | Harness Developer Hub
developer.harness.io·Documentation
Alternatives in Feature Flags & Experimentation6
Harness FME (formerly Split.io, acquired June 2024) is positioned as the only feature management and experimentation platform natively embedded within a full-stack, AI-powered software delivery suite covering CI/CD, chaos engineering, cloud cost management, and AppSec.
- Core differentiators over standalone feature flag tools include Split's battle-tested statistical engine (sequential testing, fixed-horizon, dimensional analysis), deep CI/CD pipeline integration, a warehouse-native experimentation layer (Snowflake, Redshift), and an AI Release Agent that interprets experiment results and recommends rollout actions.
- Serving 50B+ flag evaluations daily to 2B+ end users, the platform targets enterprise-scale adoption.
- Unlike pure-play specialists such as LaunchDarkly or Statsig, Harness bets on platform consolidation across the entire SDLC.
- Compared with open-source alternatives such as Unleash, Flagsmith, or GrowthBook, Harness FME is a commercial SaaS-only offering, trading self-hosted flexibility for enterprise reliability and integrated experimentation analytics.
Reviews
Praised
- Ease of use and intuitive flag management interface
- Strong feature flag creation and targeting capabilities
- Quick setup and developer onboarding
- Easy integrations with existing CI/CD and observability tools
- Streamlined A/B testing without requiring dedicated data science headcount
- Accessible for both technical and non-technical team members
- Strong multi-environment control and rollout controls
Criticized
- Steep learning curve for complex configurations
- Cluttered or difficult-to-navigate UI
- Missing features compared to pure-play feature flag specialists
- Complex and opaque enterprise pricing model
- Documentation can be challenging to navigate as platform evolves
- Cloud-only; no on-premise or self-hosted deployment option for FME
On G2, the Harness Platform listing (which encompasses FME) holds a 4.6/5.0 rating from 277 reviews, ranking 4th in the Feature Management category behind LaunchDarkly, Statsig, and PostHog by review volume. Enterprise and mid-market users consistently praise ease of use, feature flag functionality, quick setup, and strong integrations with existing DevOps toolchains. Common criticisms include a steep learning curve for complex configurations, UI navigation challenges, and missing features compared to pure-play feature flag specialists. On Gartner Peer Insights, Harness FME holds a 5.0/5.0 rating but with only 1 published review as of mid-2025, making category-level conclusions limited for the FME-specific product.
Pricing
Harness FME offers a free plan accessible via the Harness platform free tier and an Enterprise plan with custom pricing. The pricing model is usage-based, calculated on the number of active feature flags and managed users, with monthly and annual billing options. Advanced experimentation and enterprise-grade governance controls are available in the Enterprise tier. Enterprise plan pricing is not publicly disclosed; per Octopus Deploy citing Vendr data, a 200-person organization may pay approximately $23K–$41K annually for the full Harness platform. A free trial is available for evaluation.
Limitations
- Harness FME is a cloud-only SaaS offering with no self-hosted or on-premise deployment option, which may present data residency or compliance challenges for highly regulated industries.
- G2 reviewers report a steep learning curve for complex configurations and flag setups, and some note the UI can be cluttered or difficult to navigate.
- Enterprise pricing is not publicly disclosed and has been described by some users as complex or misaligned with modern deployment patterns.
- Unlike open-source alternatives such as Unleash, Flagsmith, or GrowthBook, FME has no community-maintained self-hosted version.
- Teams seeking a standalone, lightweight feature flag tool may find the broader 15-module Harness platform to be excessive overhead.
Frequently asked questions
Topic coverageCoverage by buyer topic
Topic Coverage
Prompt-Level Results
| Prompt | Perplexity | ChatGPT | Gemini Search | Bing Copilot | Google AI Mode | Grok |
|---|---|---|---|---|---|---|
Capability5/5 cited (100%) | ||||||
Which enterprise feature flag platforms offer the most flexible targeting — user segments, percentage rollouts, and custom attributes? | ||||||
Which platforms combine feature flags and full experimentation in one tool — and when do teams actually need a dedicated experimentation platform on top? | ||||||
Which feature flag platforms support multi-variate experiments with built-in statistical significance calculations so you don't need a separate experimentation tool? | ||||||
Which feature flag platforms handle anonymous visitor evaluation well without identity stitching problems? | ||||||
Which enterprise feature flag platforms offer the best audit logs, approval workflows, and change management for regulated industries? | ||||||
Developer Experience5/5 cited (100%) | ||||||
Which feature flag platforms offer a great local development experience without requiring engineers to connect to a remote service every run? | ||||||
Which feature flag platforms let product and engineering collaborate on targeting rules without requiring a redeployment every time a rule changes? | ||||||
What feature flag tools support the full lifecycle — create, roll out, and safely clean up flags — with built-in guardrails for stale flag removal? | ||||||
What feature flag platforms make it easiest to write unit tests for feature-flagged code paths without making tests brittle? | ||||||
Which feature flag platforms have the best tooling for preventing flag sprawl and keeping the flag inventory manageable as the codebase grows? | ||||||
Integrations & Ecosystem5/5 cited (100%) | ||||||
Which feature flag platforms integrate best with container-native progressive delivery pipelines for safe canary and blue-green deployments? | ||||||
Which feature flag platforms integrate natively with popular data warehouses so experiment results flow directly into the analytics stack? | ||||||
Which feature flag tools integrate with incident management workflows so a flag can be killed automatically when an error rate spike is detected? | ||||||
Which feature flag platforms can push flag state changes to a data lake so experiment assignments can be joined with downstream conversion events? | ||||||
Which feature flag platforms have the best OpenFeature support for teams looking to avoid vendor lock-in? | ||||||
Performance & Reliability5/5 cited (100%) | ||||||
Which feature flag platforms cache the last known flag state locally so applications keep working even if the flag service goes down? | ||||||
Which feature flag platforms are best for server-side evaluation at scale — and which are optimised for client-side evaluation in a high-scale SaaS app? | ||||||
Which feature flag platforms handle millions of flag evaluations per second without adding latency to hot paths? | ||||||
Which feature flag platforms add the least latency per synchronous flag evaluation call at high request volumes? | ||||||
Which production-grade feature flag platforms offer the strongest SLA and uptime guarantees? | ||||||
Setup & First Run5/5 cited (100%) | ||||||
What are the best feature flag platforms for migrating away from hardcoded environment variable toggles without breaking production? | ||||||
What's the quickest feature flag platform to add to an existing Node.js backend without a major SDK rewrite? | ||||||
Which feature flag platforms work well across a monorepo serving both a React frontend and multiple microservices from a single integration? | ||||||
I'm evaluating feature flag platforms for a 5-engineer startup — what are the real tradeoffs between self-hosted and managed options at this stage? | ||||||
What tools do teams use to set up their first A/B test on a production feature — data layer, targeting, and metrics tracking in one place? | ||||||
Turn this matrix into daily prompt monitoring.
Track prompt changesVertical Ranking
| # | Brand | PresencePres. | Share of VoiceSoV | DocsDocs | BlogBlog | MentionsMent. | Avg PosPos | Sentiment |
|---|---|---|---|---|---|---|---|---|
| 1 | Harness | 35.3% | 10.3% | 8.0% | 30.0% | 0.0% | #22.1 | +0.41 |
| 2 | GrowthBook | 33.3% | 10.7% | 4.7% | 0.0% | 43.3% | #20.2 | +0.44 |
| 3 | Unleash | 32.7% | 14.3% | 18.7% | 24.7% | 47.3% | #20.9 | +0.42 |
| 4 | LaunchDarkly | 32.7% | 20.4% | 0.0% | 26.0% | 64.0% | #27.4 | +0.45 |
| 5 | Statsig | 26.0% | 15.5% | 4.7% | 4.0% | 34.7% | #33.1 | +0.44 |
| 6 | Flagsmith | 25.3% | 13.6% | 6.7% | 22.0% | 43.3% | #33.7 | +0.38 |
| 7 | ConfigCat | 20.0% | 6.9% | 2.0% | 16.0% | 23.3% | #33.5 | +0.40 |
| 8 | Kameleoon | 17.3% | 2.7% | 0.0% | 16.7% | 9.3% | #13.9 | +0.42 |
| 9 | DevCycle | 8.0% | 2.0% | 0.7% | 3.3% | 6.0% | #24.6 | +0.51 |
| 10 | Optimizely | 5.3% | 1.5% | 0.7% | 0.7% | 14.0% | #23.9 | +0.39 |
| 11 | Eppo | 5.3% | 1.4% | 2.7% | 4.0% | 3.3% | #39.8 | +0.24 |
| 12 | VWO | 3.3% | 0.6% | 1.3% | 1.3% | 0.0% | #19.7 | +0.35 |
Turn this into your team dashboard
Sign up to unlock project-level analytics, daily tracking, actionable insights, custom prompt configurations, adoption tracking, AI traffic analytics and more.
Free trial. Setup comes pre-filled from this report.