One of my favorite aspects of my job is getting to build snazzy models to…well, predict things. So, it was with some interest that I encountered a reference to predictive modeling in Friday’s Wall Street Journal (subscriber link). The article discusses one auto dealership’s efforts to get Detroit to actually equip vehicles the way consumers want to actually buy them:
At one AutoNation Inc. location in Delray Beach, Fla., scores of “orphan” vehicles have been sitting on the lot for months. One hulking silver Dodge Ram pickup has languished unsold for 237 days, an eternity by automotive standards. The problem? Chrysler equipped the truck with a V6 engine instead of the V8 requested by most buyers of big trucks.
Parked nearby is a red Jeep Grand Cherokee with four-wheel drive, a feature popular in snowy climes but not sunny Florida. One Chrysler Sebring convertible is so loaded with options that its sticker price is $32,000 — nearly as much as a BMW 3 Series.[...]
Frustrated the Big Three aren’t moving fast enough, Mr. Jackson is taking matters into his own hands. About a year ago, AutoNation hired McKinsey & Co. and two other consulting groups with retail expertise to mine consumer data. The goal is to identify the few versions of every vehicle that are big sellers among the thousands of possible variations. GM’s Mr. LaNeve says his company is seriously considering joining with Mr. Jackson to create a “predictive modeling” system.
What’s surprising is not that this is happening…it is that it’s taken this long for Detroit to have started moving towards this sort of analysis.