Loading...
LaunchDarkly, Split, and every other major flag provider charge per MAU or per evaluation. Stale flags still evaluate. Every old flag you forget to remove is on their P&L. That is why their dashboards never nag you to clean up.
Every device that hits a flag is a MAU, even if the flag is permanent or stale. Cleaning up flags doesn't reduce MAU; the same users still hit the same SDK. Their P&L doesn't move when you clean up.
Charges scale with both unique users and number of flags. Each stale flag you leave in production is direct revenue. There is no in-product nudge to delete unused flags.
Cleaner usage-based, but stale flags still fire evaluation events. The fewer flags you have, the less they bill. They have the same misalignment, just smaller.
Every major flag provider has a "stale flags" dashboard feature. None of them ever send you a PR. None of them block flag creation when you're over a threshold. None of them refuse to let you create flag #847.
This is not laziness. This is structural. Their head of product cannot ship a feature that reduces their own ARR. The cleanup tool can only be built by a third party whose entire business depends on flags going away.
That is FlagShark. We make money when your flag count goes down. The misalignment only gets resolved by separating the seller of flags from the cleaner of flags.
Keep LaunchDarkly. Keep Split. Keep Unleash, GrowthBook, PostHog, ConfigCat, Flagsmith, Statsig. Keep whatever your team picked. They are good at what they do: targeting, segmentation, rollout control, experimentation.
FlagShark only handles the part of the lifecycle they have no incentive to ship: opening the cleanup PR once a flag is fully rolled out and the code reference is no longer needed.
Install our GitHub Action with 8 lines of YAML. We detect, we open the PR, your team reviews and merges. Your flag provider never even knows we are there.
Free for 3 repos. Works with every major flag provider. Set up in 30 seconds.