Feature flags and A/B testing for small SaaS teams means either LaunchDarkly pricing or rolling your own from scratch

Feature flags, gradual rollouts, and A/B testing are standard practice for any SaaS product with more than a handful of users, but LaunchDarkly charges enterprise prices (- per seat per month, minimum /month) that are prohibitive for bootstrapped or early-stage products. The alternative is building your own, which takes 2,000+ hours for anything production-grade. GrowthBook launched on HN as an open-source A/B testing framework to 199 points in August 2021 and their YC W22 launch got 148 points, both showing sustained community demand for an affordable middle path. As of 2026, the market still splits between expensive SaaS platforms and complex self-hosted stacks, with no true plug-and-play option at sub-/month for small teams. This gap sharpens as AI product teams need instant killswitches for model changes and percentage-based rollouts.