Customer-based measures of inventory availability

Journal of Business Logistics, 2002 by Zinn, Walter, Mentzer, John T, Croxton, Keely L

If a firm adopts a 97% in-stock availability policy, does this mean all customers receive 97% service? Or is it more likely that 97% is an average, with some customers receiving above average service and others below average service? In reality, because different items stocked by the firm are usually assigned different availability policies, the service received by an individual customer depends upon the mix of products bought by that customer.

When adopting an in-stock availability policy, firms typically assign higher availability to items with the highest level of either sales or profitability. Firms assign lower availability to slow moving or less profitable items. Therefore, a customer may buy a mix of items that includes a higher proportion of lower availability items and thus receive less than 97% service. In contrast, another customer may buy relatively few of the low availability items and enjoy a higher than 97% level of service.

The availability level experienced by individual customers is not captured by current measures of inventory availability. Current measures are item or order oriented rather than focused on individual customers. However, it is increasingly important to focus on individual customers. Escalating pressures arising from a competitive business environment have forced firms to dedicate more resources to the needs of individual customers. Customer service, for instance, is one of the most visible topics in both the academic and the popular business press (Anderson and Narus 1995). Focus on individual customers is further evidenced by the emphasis on partnerships, alliances, and other forms of close relationships between buyers and sellers in the supply chain (Lambert, Emmelhainz, and Gardner 1999).

To fill the gap between the increased need to measure service to individual customers and the measures of inventory availability in the current literature, single customer inventory availability measures are proposed in this research. The guiding principle is that firms must manage the frequency with which individual customers are faced with a stockout. This principle is particularly important to the management of relationships with key customers. A customer is considered "key" when, for any reason, it is targeted for special service.

The proposed set of single customer inventory availability measures is based on a common premise: What is the probability that an individual customer will be told that a product is unavailable for on-time delivery? hi other words, what is the probability that a specific customer will face a stockout? Four measures are suggested. The first is the probability that an individual customer will be told that the next item ordered is available for on-time delivery. The second measures the probability that an individual customer will be told that the next order is available for on-time delivery. The third is the probability that an individual customer will not be told at least once in the next period (week, month, etc.) that an item is not available for on-time delivery. The final measure is the expected number of times an individual customer will be told during the next period that an item is not available.

These measures use a customer's purchase history to estimate shortages in future purchases. In other words, the measures are designed for suppliers with multiple customers of varying importance, with whom there is an available record of past purchases. It is assumed that customers have a consistent order profile; that is, they tend to purchase the same items in similar quantities and prices. Exampies to which this assumption might apply include packaged goods manufacturers and industrial distributors. These measures were also designed to be easily implemented because they are based on a company's existing records. Most firms have records of their customers' past purchases and have files with the inventory availability policies assigned to each item.

In addition to the measures above, we propose a managerial model designed to assist managers in the implementation of one of the measures, the item availability level promised to a specific customer. That measure is termed the single customer item availability (SCIA). The managerial model includes a mixed integer linear program (MILP) designed to identify the lowest cost combination of item availabilities that will guarantee the SCIA promised to a key customer. It is important to note that the goal of the MILP is to determine the item availability levels needed to support a pre-determined level of service to an individual customer, while minimizing the inventory required to do so. The managerial model also includes a procedure to quantify the level of safety stock needed to service different customers buying the same item but who are offered different levels of service. The procedure is based on setting "trigger" levels for individual customers.

The purpose of these "triggers" is to avoid keeping a separate safety stock for each key customer. Findings from the portfolio effect literature show that a separate safety stock for each customer is unreasonably expensive (Zinn, Levy, and Bowersox 1989). Instead, firms keep a single safety stock, but use the proposed managerial model to determine the "trigger" for each item bought by each key customer. When that level of inventory is reached, firms may select one of several courses of action. They may stop selling the item to less important customers until the inventory is replenished, they may contact the key customer to obtain an update on their purchase intentions until a replenishment arrives or, finally, they may speed up replenishments.