Lots of vendors now offer (or are scrambling to offer) analytic software, but the question is, how much expertise will you need to make productive use of the software? As The New York Times recently reported, statisticians and other math whizzes who can handle these techniques aren't having any trouble finding work:
Prebuilt models and off-the-shelf analytic applications might ease the burden for neophytes, but in many cases models and apps work best when they are customized and carefully maintained. Wholesale auto lender Dealer Services tried using prebuilt credit risk models and scoring methods to screen prospective dealer customers, but they just didn't work (as it's not hard to predict that used car dealers present unique risks). Thus, Dealer Services is building its own predictive model based on dealer traits and transaction parameters that have historically worked for the firm.
If you compare the lending risks of today to those of three years ago (when lots of consumers were borrowing against inflated home values), it's easy to see why models have to be adjusted over time. Dealer Services can't take the same risks today that it would have 18 months ago, but at the same time it wants to gain market share by approving as many good risks as it can.
The analytic challenges at Dealer Services are pretty simple, and CIO Chris Brady says she can handle all the modeling, deployment and maintenance challenges with the aid of two assistants. In fact, most IT execs don't seem daunted by the prospect of a talent gap. In a recent InformationWeek/Intelligent Enterprise survey, 34% of the 534 respondents reported they "already have skilled analytics professionals on staff." Nearly half (48%) said they expect to "train in-house BI experts and power users on analytic tools."
Indeed, you don't have to be a statistician to learn how handle analytics. The Times reports that statisticians "are only a small part of an army of experts using modern statistical techniques for data analysis. Computing and numerical skills, experts say, matter far more than degrees. So the new data sleuths come from backgrounds like economics, computer science and mathematics."
Before you wade into analytics and complexities of predictive modeling, I'd be as careful about skills assessments and staff planning as I would be about technology selection and business needs assessment. Use pilot projects to test and prove analytic processes and practices as well as potential business value. Are your people comfortable and confident about what they are doing, or are they struggling and unsure? I'm sure vendors will be only too happy to do lots of hand holding, but at what cost? Can they teach your people how to do it and then move on, or will you need long-term consulting contracts?
I'm not using a model for this analysis (I was an English major), but I'm guessing lots of executives are relying a bit too much on gut feel when investing in analytic software.Are you ready for analytics? Rising ambitions and investment in related software could bring about a renaissance... Or, if it turns out organizations can't muster the talent needed to analyze and predict, we might see a classic Gartner Hype Cycle "trough of disillusionment."