LaunchDarkly

LaunchDarkly is the leading feature management platform enabling engineering teams to safely deploy code behind feature flags, run targeted feature rollouts, and conduct A/B experiments at any scale. Founded in 2014 by Edith Harbaugh and John Kodumal, LaunchDarkly is used by over 5,500 organizations including IBM, Microsoft, NBC, and Atlassian for continuous delivery and experimentation workflows. LaunchDarkly's feature flag system supports percentage rollouts, user-attribute targeting (user plan, location, device type), multi-variate flags (A/B/C/D testing), and kill switches for immediate feature disabling without redeployment. Flags are evaluated locally by LaunchDarkly's SDK using a streaming connection to the edge relay proxy — meaning flag evaluations execute in <1ms without a network round-trip to LaunchDarkly servers. LaunchDarkly Experimentation connects flag variants to business metrics (conversion rates, revenue, error rates) for statistically rigorous A/B testing. LaunchDarkly Data Export streams flag evaluation events to your data warehouse (BigQuery, Snowflake, Redshift) for custom analysis.