Using LaunchDarkly to experiment
When working with experiments within LaunchDarkly, we will still be using feature flags, as we explored in Chapter 3, Basics of LaunchDarkly and Feature Management. As we progress through this section, you will want to create some new flags to set up experiments and get to grips with the functionality that LaunchDarkly offers.
The experimentation functionality within LaunchDarkly couples a feature flag with some form of metric that can prove the success or failure of the test. In that regard, we will be using the rollouts from Chapter 4, Percentage and Ring Rollouts, to deliver two or more variants of a feature to the users. In the scenario of running an experiment, one of those variants should be the control (baseline or benchmark), which is the current experience that's available to customers. In any experiment, the existing implementation (including when there is not an existing feature) should be considered the baseline. The new implementation...
