Given the enormous amount of domain knowledge that exists in both theory and practice , its sort of amazing that organizations are getting carried away with the sexiness of big data as Jim Goodnight of SAS mentions. You would wonder why all this analysis still leads to the mortgage mess, and debt crisis across many parts of the world. Well that's because there is too much mindless data analysis without an understanding of the core business issues. So why does this happen?
Statisticians and computer experts are sort of forced to come up with models of consumer or business buying insight. These experts are also compelled to model supply chains. The disadvantage is that they have (a) never done these jobs as line managers and/or (b) have no knowledge of what is really known about the topic from published research. Instead, these analysts keep plugging a bunch of predictive variables based on the gut feel of whichever operational manager is willing to talk to them.
Would you allow your doctor to treat you without being current on the latest medical research? I guess not but surprisingly, the more alert B2B marketer or Supply Chain leader is professionally updated on the latest thinking and research just as the better doctors are upto speed for medical research, in their specialty. It's just that line managers don't take the time and effort to tell the statisticians and computer folks all they know.
Take the buying center concept which is that for complicated business purchases there are a bunch of people who are involved in deciding the criteria for supplier selection and what features are really important. Now if you ask your top statistician to model your B2B customer buying behavior and the sales and marketing people do not respond to the statistician's requests for domain briefing (good statisticians never claim domain knowledge), your model is going to be just wrong. This is despite the fact that every B2B marketer understands and lives the buying center concept (its an obvious no-brainer to them) and makes the fatal assumption that others in the organization, like the statistician colleague- would know.
If you are a business leader please do not force the statistician to come up with a model. Your line function leaders must spend time explaining and making sure that the statistician is well briefed on the domain.Better the briefing to the math model builders better your insights from your massive data.