How do you walk an interviewer step-by-step through a root-cause analysis of a failed experiment without getting defensive?
