You've changed an engineering process and want to know whether it actually improved outcomes. You need a way to separate real improvement from normal week to week noise, and to look beyond a single headline metric.
How do you measure whether a process improvement actually worked?