Predictive Analytics is Simple....But Not Easy

Why can’t my customers get the idea of doing predictive analytics?  I tell them about the computer manufacturer who used clickstream analysis to increase sales by tens of millions of dollars per year…or the one who used advanced algorithms to predict credit card fraud and saved millions….or the manufacturer who did predictive maintenance.  Isn’t this enough to convince them to jump right in?

Well, no, unfortunately.  I meet with hundreds of customers per year and talk about analytics (ok, maybe preach analytics is a better description).  Yet, I am only able to persuade a few each year to define a predictive analytics project.  Through these meetings I  think I have found some of the reasons why so few are convinced to take the leap.

The biggest blockers to predictive analytics in the past have been acquisition costs.  Hardware, software, and people acquisition.  The projects I did 10 years ago required large hardware and software investments in order to handle the volumes of data and advanced algorithms needed to run the analysis.  I like to think that Microsoft (and others of course) have brought down the cost of start up to a nominal cost due to the cloud.  No expensive hardware to buy, just spin up a VM and pay as you go.  Same thing with software.  Even the cost of data scientists can be lowered a little by giving analysts some easier tools to use to get started. 

 However, I still see the biggest blocker is knowing where to start.  As one customer said it best, “How do we get started on something that we never thought we would be able to afford?”  In other words, we have never put any effort into thinking about how we would use predictive analytics.

 I now go into meetings with some starter ideas based on their industry in order overcome these objections and help the customer think in a new way.  I also talk in terms of levers.  It’s an easy concept for people to understand.  There are only certain levers that you can pull to make an impact on your business.  If you manufacture products, then it could be decreasing defects, if you are retail then it could be increasing sales or decreasing cost.  Funny enough, almost every customer I talk to says the weather as a lever.  Yes, weather does affect every business, but can anyone really predict the weather?

 The best example I use with  my customers is one company who just got some people in a room for a few hours and did a high level table diagram of their business and factors that impacted them.  They wound up with a 5 by 5 table with 10 ideas total on levers they could adjust that would impact their business.  They spent a few weeks fine tuning until they decided on one project.  This is why I say that predictive analytics is simple...but not easy.  It is simple to understand your business drivers, but not easy to predict which adjustments you make that would make a huge influence on your company.