Dan Gilmore wrote in “The ‘Probability’ of Supply Chain ROI” propounds properly and rationally the fact that any “forecast,” including forecasts of ROI (return on investment) should not be a single number. Rather, as anyone properly trained in statistical methods will tell you, it should be a range of numbers. The range of numbers would generally be calculated based on a single calculated value plus and minus values that represent the confidence intervals or, simply put, how likely the statistician believes his estimates the calculates will approximate reality. A larger range indicates lower levels of confidence and a smaller range higher confidence levels. Now, while Gilmore is mathematically correct, the fact remains that most small-to-mid-sized businesses (SMBs) simply do not have anyone trained in statistics on their payroll and they are not likely to go out and hire a statistician to produce ROI forecasts for their IT projects – since this would, by definition, automatically reduce the ROI of the enterprise as a whole in the short term.
Back on a growth trajectory
Gilmore makes another comment in his article with which I wholeheartedly agree: “[T]here is some evidence that companies are in fact looking at investments that can help them to get back on a growth trajectory (read: increasing Throughput) without having to add much in the way of head count (read: Operating Expenses) by achieving productivity gains.” Given the world-wide economic malaise that is showing some signs of lessening (for the moment, at least), Gilmore’s description probably suits the vast majority of SMBs across the U.S. and beyond. Furthermore, many others besides me have written that a firm stand on return on investment will be the hallmark of technology spending in the 2011 and beyond. So, I can hardly fault Gilmore for suggesting that SMB executives and managers need to become increasingly sensitive to and realistic about ROI for every kind of investment in their firms’ futures.
Too much complexity already
Despite my agreement with Gilmore on theoretical grounds regarding forecasts – including ROI forecasts; and despite my agreement with him regarding the goal of companies to get back on a growth trajectory through wise investment of capital resources, I must disagree with him on the matter of adding useless complexity to the return on investment forecasting process. Allow me to explain why I use the harsh term “useless” to describe such an effort in the development of a ROI forecast for an IT project.
First of all, let me say that statistical methods ought to be applied where they make sense. Statisticians generally agree that a valid statistical sample must contain at least 30 members. This works great where you have 30 dogs, 30 cows, 30 houses, 30 automobile, 30 miles of roadway, and so forth for comparison. Then, of course, you need to factor for environmental differences. Thirty or more cows all in the same pasture, eating the same foods, and enjoying the same climate would make a pretty good statistical sample for some studies of cows. On the other hand, three Holstein cows in northern Minnesota, two long-horns in west Texas, 15 black whiteface cows in eastern South Dakota, and ten mixed-breed cows in central Florida are not likely to constitute a good “sample” for cow studies.
Simply because there are too many environmental dissimilarities surrounding the cattle. By the time these factors were accounted for, (generally speaking) any results would have such a large confidence interval as to make any prediction almost meaningless. When considered as a whole, a typical SMB has tens of thousand of variable at work within the enterprise. Any number of those variables are likely to dramatically separate it from any “sister” enterprises in a sample group used to forecast ROI outcomes. Of course, the fact that traditional ERP – Everything Replacement Projects – are going to affect the whole enterprise is a big part of the problem of predicting ROI outcomes. With tens of thousands of variables at play, picking the winning number is far more challenging than winning the lottery.
Reducing the scope reduces the complexity
First of all, a good many SMBs today have a “pretty good” ERP system in place – regardless of its brand. Unless there is some pressing reason to undertake a traditional ERP – Everything Replacement Project, it is probably a far better idea to consider a New ERP – Extended Readiness for Profit project instead. Narrowing the scope of the project reduces the complexity. And, reducing the complexity increases the likelihood that your ROI forecast will be more on-target. Allow me to give you a couple of examples:
Warehouse automation – Suppose your Current Reality Tree (CRT) shows you that, in order to increase Throughput, you need to enable your warehouse to be able process 80 to 100 shipments per hour, rather than the present 30 to 40 per hour. Furthermore, you need to do this while holding the line onOperating Expenses.
Business intelligence – After reviewing their Current Reality Tree and laying out a Transition Tree (TrT), your marketing and sales team has assured you that, if they are provided tools with which to better understand customers and markets, they will be able to segment the marketThroughput by 130% in two years. No additional staff or other increases in Operating Expenses and create UROs (un-refusable offers) and increase would be required.
If your executive management team were to elect to pursue either of these projects – or both – the goals are specific and measurable – as would be the expected outcomes. ROI calculations become simple:
Where delta-T = the estimated change in Throughput (Revenues less Truly Variable Costs), delta-OE = the estimated change Operating Expenses, and delta-I = the estimated Investment (including inventories).
Simple. Elegant. And, provided reasonable consideration is put into the estimates for the variables involved, such ROI calculations are far more likely to be right than any calculation around traditional ERP – Everything Replacement Projects.
©2010, 2011 Richard D. Cushing