Without a doubt, justifying the expenditure for new projects or new technologies is one of the most challenging tasks that any professional will ever undertake. It is particularly onerous in a down economy, when senior management is likely to scrutinize every nickel and dime the company is spending and to demand an extremely quick return.

At first glance, coming up with the data necessary to calculate a project's return on investment (ROI) may seem like the trickiest aspect of the cost-justification exercise. I wouldn't want to wade through accounting books myself. But while it's true that such things as huge overhead categories and intangible benefits can make for some murky numbers, the real challenge lies in managing the uncertainty that is an inherent part of life itself.

That uncertainty lurks at key junctures across the supply chain. No matter how good you think your forecasting ability is, a vendor's inability to obtain raw materials can cause unexpected project delays. Transportation snags can disrupt even the most carefully planned production schedule. Future demand can plummet. Unanticipated engineering changes or bad parts can blow the budget. The list goes on and on.

So, although managers should hope for the best-case scenario in putting together the cost justification for a project, they need to plan for the worst-case scenario.

Consider, for example, the case of an earthmoving equipment manufacturer in South Africa that purchased a new type of screw for the axle assembly. Shortly after the equipment went into production, the screws began failing in the field in an unusual way. Instead of shearing at the head, which is the more typical mode of failure, the screws were splitting down the middle. Failure analysis later revealed a material flaw that was undetectable by the naked eye and therefore not revealed in visual QA inspections. The upshot was that the company had to recall parts from around the world, replacing the faulty hardware with a different screw at great cost—a costly reverse supply chain nightmare. Bye-bye, original ROI.

Or what about the company that predicted blockbuster sales of a new medical device and used the expected incoming cash flow to justify an entire new manufacturing plant? Unfortunately, failure to secure FDA approval dragged the product introduction out for 18 months beyond the expected launch date. Sales after that were slower than expected, to boot. So long, original ROI.

Engineering change orders are another project bugaboo. The mantra for many engineering departments today is "Do it right the first time." But the fact of the matter is that the average number of change orders on any given product design project can easily run up into the double-digit range—and that is assuming most everything goes as planned! Smart companies budget for engineering change orders. Otherwise, see ya later, original ROI.

Despite the uncertainty, I am amazed how often people (including me) obsess over getting the "right" answer. Some even go so far as to calculate a project's ROI out to three significant decimal places, even when the data are shaky to start with. And few of them expect anything to actually go wrong, even when they take shortcuts like cutting corners on the design, purchasing substandard components, or going with the "lowball" carrier.

Project managers could make far better use of their time by examining the expected financial performance of a proposed investment over a wide range of conditions. The interest or discount rate selected, the estimated annual savings and costs, and even the project schedule can completely transform the economics of a proposed project. By doing this type of sensitivity analysis, project managers can gain insight into how "wrong" the data can be and still remain confident that the project is financially viable. And, frankly, that's precisely what senior management really wants to know.

Simply put, sensitivity analysis involves plugging different numbers into the net present value (NPV) calculation. The formula is relatively easy to crank through (see the accompanying graphic)—any good financial calculator today will do it automatically. The challenge still involves coming up with the specific numbers, because one thing sensitivity analysis does not involve is simply pulling numbers out of the air.

The first step is to identify the baseline case, which includes the best-guess estimates for the data. The data should be conservative—meaning that it is close to best case but assumes a few minor bumps in the road along the way.

The second step is to define a range for each data item, bounded at the top by the best case and at the bottom by the worst. Conservatism is the operative word here, too. A snowstorm might be factored in as a potential impact on a project's timetable, but a 100-year-flood or invasion of locusts would be taking things a bit too far.

The net present value for several data points within this range is then calculated. This proves to be the most illuminating step because it helps the project manager identify which conditions affect a project's net present value the most. For example, the higher the interest rate—which is used to convert the expected cash flow from each year of the project to present dollars—the higher the project's net present value. On average, companies assume an interest rate that is between 10 and 20 percent. And, of course, not discounting the cash flows at all will result in a severe overestimate of a project's ROI.

The amount of the estimated cash flows and the timetable for receiving those cash flows have a strong impact on a project's ROI. Project delays are costly because they push the expected annual savings further into the future, when the value of money is lower. Likewise, unexpected cash infusions into a project push the ROI out. Should quality problems arise only after the product hits the market, sales could plummet far below the initial projections—and this does not even include repair, replacement, and warranty costs!

In the end, one can hope that most project managers will never have to deal with any worst-case scenarios. But by performing some sensitivity analysis, the prudent ones will have covered all of their bases in advance and still achieved a reasonable ROI—even if Murphy's law does hold true!