Look for Analysis to Get Vertical Soon
Historically, the analytical tools sold to a customer in one industry were the same tools provided to those in another—there was little or no tailoring of tools for particular industries. A chi square was a chi square, a logistic regression in retail was a logistic regression in banking. It was up to smart analysts to figure out how data from a specific industry could be analyzed in a meaningful fashion for that industry.
This has begun to change, but many analytical environments are still industry-generic. This is despite the fact that each industry has business problems that are best solved with a particular analytical approach. And transaction software (e.g., ERP) vendors have long customized their products by industry.
While a solution to almost any industry-specific problem can be cobbled together with generic analytical tools, the skills to do this are not widely available. The result of an industry-generic approach to analytics is that many problems in many industries remain unaddressed. It’s also a good bet that many analytical tools are misused because they are applied where they don’t fit particularly well. People who want to use regression analysis to study stock returns in financial data, for example, may not know that heteroscedasticity (http://en.wikipedia.org/wiki/Heteroscedasticity)–you gotta love Wikipedia for having an entry on it—will be a potential problem that needs to be corrected for.
In the future, I’ll wager that we will see a lot more industry-specific analytical applications. If you want to do some analysis, you will need to know your industry and the kinds of decisions that need to be made in it, but you won’t need to know the details of how to do the analysis. OK, you’ll always need to know something about statistical inference and significance testing, but most college graduates know something about that already. And the internal and external data you need to do the analysis will be either bundled with the app subscription, or easily extracted from your transaction systems and internal or external databases.
So if you’re in health care, get ready for the analytical app that tells you whether or not your coronary thrombosis treatments are yielding the right level of patient survival. If you’re in consumer banking, you’ll be able to use an app that tells you how likely a checking account customer is to depart for another bank. If you’re in a shipping company, you’ll have a handy app that tells how to optimize the loading of your trucks. None of these analyses will require Ph.D.s in statistics to help you create the application or interpret the results.
We’re already seeing this trend emerge from existing vendors. Check out this list (http://www.sas.com/industry/financial-services/banking/index.html), for example, of SAS banking applications. I think the trend will accelerate, however, and the solutions will become narrower and more industry-focused. Of course this will have implications, as some recent comments to previous posts have pointed out, for how our business processes work and how our analysts are educated and deployed. On the whole it will be for the better, but there will undoubtedly be some dislocations.
