The Defense Logistics Agency has five million items in its catalog. DLA handles everything from clothing and food to medical supplies and hardware. For the last category, it deals in some four million items, but doesn't keep all of them in stock. What's more, says the DLA's Robert Carroll, 90 percent of the items that it does stock are difficult to forecast. The agency needed a better means of assessing demand and reducing the risk of both stockouts and overstocks.

Certain items have "high demand variability," making them nearly impossible to forecast. "We either buy too much or not enough," says Carroll. The result in both cases is higher costs.

Many DLA customers are maintenance organizations trying to fix complex equipment. Yet they are affected by factors beyond their control. The situation prompted DLA to come up with a supply chain based on managing risk instead of forecasting.

The agency developed a mathematical algorithm that balances the risk of over- and under-stocking. It focuses on items with frequent but highly variable demand. The new program arose from an effort to understand the customer's environment and sources of uncertainty - "the realization that there are causal factors that can't be controlled," says Carroll.

The program is still in its early stages, but results are already evident. For items with sparse demand, material availability is up about 6 points. For those with more frequent demand, it has improved by 2.5 points.

"It comes from having the right timing of buys," says Carroll. Potential benefits could bring a 20-percent reduction in inventory. That translates into between $1bn and $1.5bn in hardware expense. At the same time, DLA is hoping to focus on improving customer service, with a target of reducing the wait time for items by 20 to 25 percent, and cutting down on backorder processing.

To view the video in its entirety, click here