Some suggestions for improving the referenced workspace:
- ability to use custom metrics (doesn't appear to be the case as of now)
- ability to select more than 3 (appears to be the current limit),
- whether or not the confidence calculations adjust for multiple comparisons for tests with multiple challengers