medium10 pts · first answer countsexperimentationdata-quality

An uneven 50/50 split

A test was set to split traffic 50/50, but control has 52,000 users and treatment has 48,000. What should you do?

A.Nothing; the split is close enoughB.Treat it as a sample ratio mismatch: investigate the assignment or tracking bug before trusting any resultsC.Remove 4,000 control users at randomD.Declare the treatment the winner

Discussion 0

Join free to post in the discussion.

An uneven 50/50 split

Discussion 0

Related questions