Manufacturers generally classify products in terms of broad product lines, developing a single marketing strategy and production plan for each line. That makes sense for marketing, but it's a mistake for production. Different SKUs [stock-keeping units] within a product line can have very different inventory needs.
Take, for example, a large American manufacturer of men's blazers. As part of our research into lean retailing, we tracked the demand for different sizes of a blue blazer. Far from a trendy fashion item, the blue blazer is a staple of the wardrobes of millions of men. But from the perspective of actual consumer buying patterns, a blazer in an atypical size actually has more in common with a fashion-driven product than with the same style jacket in a popular size. For example, sales for 46-regular, one of the most popular sizes, vary only by twice the average weekly demand, while sales for 43-regular vary as much as four times the average demand. A rare size, such as 43-long, would vary even more. To satisfy retail customers, the manufacturer must hold a proportionately larger inventory of 43-regular, even though in absolute terms it will hold much more of 46-regular. But most manufacturers, including this one, tend to assign the same inventory policy for all products in a product line.
By fine-tuning inventories according to SKU-level demand, a manufacturer can increase profits and reduce inventory risks.
By fine-tuning inventories according to SKU-level demand, a manufacturer can increase profits and reduce inventory risks. To demonstrate that improvement, we ran a computer simulation that tests various inventory policies for three groups of SKUs in the same product line — one group with low variance in demand, another with medium variance, and the third with high variance. (See the exhibit "A Better Way to Manage Inventory.")
The first test shows a scenario in which a manufacturer is most concerned about keeping its big retail customers happy by maintaining very high order fulfillment rates. The manufacturer sets a single inventory policy to ensure that its highest variance SKUs have plenty of finished goods on hand — say nine times the expected weekly demand for those SKUs. Following that inventory policy, the other two groups of SKUs in that product line also carry inventory of nine times the expected weekly demand even though their variation is never more than four times the average.
The second test reflects a manufacturer whose concern is maintaining inventories at a level appropriate for its high-volume, low-variability SKUs — say three weeks of demand. That means much lower inventories in general and a savings in working capital and risk. But the trade-off is that the manufacturer frequently runs short on its medium and especially its high-variability items. That means lost sales and maybe a canceled contract with a prized customer.
In the third test, the manufacturer focuses on balancing the costs of stockouts and inventory by setting a single inventory policy for all SKUs at seven weeks. In the case of blazers, the inventory of the 43-regular is just about right, but there are too many 46-regulars and stockouts of 43-longs.
The better approach, of course, is for the manufacturer to assign an individual inventory policy for each SKU. The fourth test optimizes the profit of each SKU according to the estimated costs of stocking out versus holding inventory. Inventories for some SKUs go up, while others go down, but overall inventories fall. And net profits rise.
We know of no manufacturers that have fully implemented what we propose. Yet lean retailers like Home Depot and Wal-Mart already incorporate some SKU-level analysis in their own inventory decision making. Calculating SKU-level variation can be done on a simple spreadsheet, so moving toward this type of inventory policy should be quite feasible.