Critical Analysis

Schneider identifies some excellent metrics, and it also makes use of scientific observation by conducting a controlled experiment with a control group and a test group. This alone puts Schneider in a small group of best-in-class IT investment analysts. The company even uses some excellent decision-theory methods that are far too rare in IT. Nevertheless, there is still room for improvement.

First, I’ll address an issue that is a source of great philosophical battle between accountants and economists. Accountants treat depreciation and book value as “real” dollars. Economists, along with management scientists, financial analysts and virtually everyone else, do not. I’m in the latter camp. Book value is not the same as the cash received if a trailer is sold or the cash spent if a new one is purchased. If one of the benefits of Schneider’s proposed investment is a reduction in the number of trailers (or avoiding the purchase of new ones), then they should measure that.

The “technical metrics” identified are good, observable quantities. Methods like what Schneider used for multivariate comparisons have a good foundation in decision theory and, in many cases, have been shown to improve decisions. But these methods are only necessary if the investment is too small to merit development of a more realistic economic model. A $20 million-plus investment does not usually fall into this category. Each of the four comparative metrics could, instead, have been used in a computer simulation of transportation operations. For a fraction of 1 percent of the cost of the investment, Schneider could simulate thousands of shipments on hundreds of routes and tally the effects on profit of latency and the other measures. Profit is the most relevant single “statistical figure of merit” that could have been devised.

Readers of CIO may already know my position on intangiblesnamely, I believe there are none. The same simulation created to assess the effect of latency and other characteristics could model the occurrence of late and lost shipments. Then Schneider could analyze historical shipment and customer data to find statistical correlations of late and lost shipments on repeat business from a customer. The same analysis might be used to correlate driver turnover to billable trailer miles. Drivers and customers have real, observable (and probably already recorded) effects on profit that are no “softer” than revenue per trailer.

Finally, a five-year payback for a large and risky technology investment is really not that good. Once risk is accounted for, many leading-edge investments require paybacks of two years or less. But I suspect that a more complete ROI that includes increased customer and driver satisfaction might achieve the required payback.

Douglas Hubbard is founding partner of Hubbard Ross in Glen Ellyn, Ill. Hubbard Ross uses the scientific and mathematical principles of applied information economics to systematically analyze the economic value of any IT investment. Hubbard is the inventor of applied information economics and has more than 12 years’ experience in IT management consulting. He can be reached at mailto:dwhubbard@hubbardresearch.com.