Problem
We all agree that correlation does not equal causation. In order to find out how something influences or causes a change in behavior we must measure lift. So should we just write off correlative data analysis? Why even measure click rate and other KPIs if they don't directly support causation? When might this information be helpful?