## To-dos - [ ] Figure out the standard tooling set - Tooling should support and simplify adherence to the process - Need to research performance concerns of using Unleash on .com - Both [Unleash](https://github.com/unleash/unleash#what-is-a-feature-toggle) & [Flipper](https://github.com/jnunemaker/flipper/blob/master/docs/ui/README.md) have UIs that can be used to view existing experiments & adjust their rollout strategies & targets (e.g. users, groups, etc.) - [Unleash has variants built in](https://unleash.github.io/docs/toggle_variants), Flipper does not appear to natively support variants - Unleash server is node module, whereas Flipper is Ruby gem - Using Flipper’s built-in `%-of-actors` makes sense once we know we’ll have a logged in user - Using Flipper’s built-in `%-of-time` and converting that to individual users via a cookie (i.e. as in the `Gitlab::Experimentation` framework) makes good sense when we’re not sure we’ll have an authenticated user or when we know that we won’t have an authenticated user (i.e. sign-up & start-a-trial flows) - It would be nice if PMs could directly control the rollout timing & strategy of experiments ## Alternative Solutions - [Intuit’s Wasabi A/B Testing Platform (deprecated)](https://github.com/intuit/wasabi)