Call it the mathematics of slack: For each hour gained through productivity-enhancing software, less than an hour of additional labor usually gets done. Or, as Rebecca Wettemann, director of research at Wellesley, Mass.-based Nucleus Research, puts it, "Time saved does not equal time worked." Based on its observations of many real-life project implementations, Nucleus has created a trademarked database of "correction factors" to account for how well various kinds of employees actually make use of productivity gains. Not unexpectedly, the most efficient workers are usually those whose output is concrete, such as a line worker, or whose pay is tied to commission.

The formula below shows one way of using correction factors to adjust expected savings. In this example, a midsize packaged goods company plans to try out a knowledge-management system at a branch office. The company, using a straight, time-is-dollars equation, expects the system to save each marketing employee about four hours out of a 40-hour workweek (a 10% raw productivity gain).