Summary

Running experiments is such a key component of feature management, and LaunchDarkly provides a good level of functionality to make the most of the opportunity that testing in this way presents. By considering the release of code as an experiment with a sense of how key metric(s) should change, teams can be data-driven in their approach to refining their product.

Separating metrics from experiments and flags themselves emphasizes the importance of knowing what it is that is being measured for an experiment. Once that is known, it is easy to determine whether that metric should increase or decrease to conclude that the experiment has been a success. LaunchDarkly leads us into thinking along these lines when creating a metric, and the separation of metrics from a single flag shows how important a metric could be for multiple features.

You should now be able to create metrics and add them to feature flags to understand which variants of a feature perform the best. You should...