In a recent blog post from Vinnie Mirchandani on why companies aren’t ready for analytics, he noted that “most analysts focus on slice and dice, not decisions.” If you’ve read my previous posts on this site, you know that I could not agree more.
Decisions require a little more understanding than the rear view offered by the standard management report that gets described as analytics. Decisions are about the future, not the past, and analytics need to be about the future as well.
At FICO, when we start an analytics project we don’t start by seeing how much data we can pull together, or what new technologies can be invented to analyze that data. We start with the decision, then determine what analysis would improve that decision, then identify the data and technology that would be needed for the analysis. In a sense, we work backward from the decision, rather than working forward from the data in hopes that we’ll meet a needy decision along the way!
Take one of the areas we’re working on today: Forecasting if a person will take their prescribed medication. This turns out to be a tractable problem and an important one to many people in the field of healthcare. The decision – who to intervene with to change their behavior, and how best to impact that behavior – requires the analytics that tell you who needs the intervention.
Because we’re dealing with decisions, not just analytics, there’s another critical factor involved: Experimentation. Are we making the right decisions on the right people? What happens when we change the actions, or the audience, by changing the thresholds for the analytics we use to determine what to do with whom? What trips many companies up is moving from analytic adoption to analytic experimentation.
The good news is that, unlike Vinnie, we see more and more companies that ARE ready for analytics-guided decisions. But other companies that are just jumping on the analytics bandwagon will soon discover how important it is to start with the decision.