Case Study #3: L.L. Bean
Question 1: How does L.L. Bean use past demand data and a specific item forecast to decide how many units of that item to stock?
L.L. Bean uses several different calculations in order to determine the number of units of a particular item it should stock, whether it is a new item or a never out item. The first piece of data that is required is a point forecast for the item in the future period. This comes from the forecasting department, and is based off of the book forecast and past demand data. For a new item, there is a lot more judgment involved, especially with trying to estimate how much demand this new product will generate. This point forecast is then used in conjunction with the A/F ratio, which looks at an individual item’s past season’s forecast and actual demand. By calculating the A/F ratio, L.L. Bean can find the range of inventory that the product will be in the upcoming season after converting the point forecast into a demand distribution. For example, if there was a 50% chance that the forecast errors for last season were between .5 and 1.5, then it follows that those same distributions would occur in the future period. So in this example, the stock amount to order would be between 500 and 1500 units of that item. The third step in forecasting demand is to find the service level based off a profit margin calculation. L.L. Bean wants to look at what the probability of each unit bought is versus the amount they would lose if the unit were to be liquidated. They can then use this to calculate a fractile, which is used to determine the actual order size as long as it falls within the past period’s distribution. The fractile calculation must be done so we can see at what point it is optimal to hold the stock in order to balance overstocking and understocking costs, which then determines the number of units to stock.
Question 2: What item costs and revenues are relevant to the decision of how many units of...