رویکرد برای پیوند خصوصیات مشتری برای تصمیم گیری موجودی

Differentiation of logistics services through distinct inventory policies is analyzed. A framework is presented in which product-based information is supplemented by customer-specific characteristics when deciding on the inventory policy for a specific product. Product-based information includes sales volume and variation by product. Customer information consists of purchase volume and its growth potential, and evaluation of the effects that service level has on purchases. Delivery time is used as service measure and inventory policy consists of selecting the production mode from make-to-order, make-to-stock, or one of their variations. As a result, distinct inventory policies are formed.

One possibility for a company to differentiate itself from the competitors is to offer better logistics services. For a traditional make-to-stock company one alternative to carry out this value-added logistics strategy is to move the location of inventories at least partially towards the customers. In order to shorten delivery times for some of the products and customers, market area stocks or even customer-dedicated consignment stocks can be established (e.g. Meijboom, 1999). On the other hand, it could be feasible, from the cost-efficiency point of view, to operate with some of the products and customers on a make-to-order basis. Therefore, it could be justified in many circumstances to design product and customer-specific production mode and inventory location policies (Giesberts and van der Tang, 1992; Fuller et al., 1993). Classification of decision-making situations according to some identifiable criteria is a popular approach for designing policies that match the requirements of specific situations. A well-known classification scheme in production and inventory management is ABC analysis, in which the classification is based on the monetary value of item sales or usage. However, many decision-making situations require more refined classification according to more than one criterion (Vollman et al., 1988). The traditional approach has been expanded by including, e.g. item criticality as additional classification criteria (Flores and Whybark, 1989). Cohen and Ernst (1988) have presented a clustering method with which multiple criteria can be used in classification. With multiple criteria approach, criteria other than product-based criteria can be included in the analysis. E.g. in a physical distribution application Fuller et al. (1993) used criteria related to sales volume, order size, coordination, delivery and handling requirements. De Leeuw (1996) used characteristics of products, processes and markets to discriminate distribution control situations. For designing differentiated logistics service arrangements van der Veeken and Rutten (1998) determined separate customer order profiles according to three types of attributes: general customer attributes (sales), customer order attributes (e.g. number of deliveries), and product logistics attributes (product value, product size). When designing differentiated policies, it is important to have a balanced view of both the effectiveness of service improvements and how efficiently they can be achieved. In this paper we make an attempt to link customer analysis into inventory policy decision making so that aggregate demand information is supplemented by customer-specific characteristics when deciding on the inventory policy for a specific product.

The presented framework gives one way to handle the complexity and interdependencies of the required product and customer analyses while determining appropriate inventory and production control policies for a firm. It is necessary to link customer characteristics to pure product-based inventory policy analysis in order to create distinct customer service policies to different customers. This way it is possible to allocate the logistics resources to customers with highest profitability expectations and growth potential. In our framework, the product-based analysis is supplemented by evaluating a customer's business importance and determining a subjective estimate on the effects of the service changes on the purchased volumes. The classification methods used in the product and customer analysis above have some limitations that restrict their usefulness in complex situations. It is difficult to combine and understand the effects of more than two factors at the same time if a traditional matrix presentation is used. In many cases, especially when the number of the products and customers is very high, it is also difficult to define, for example, the relevant grouping of customers beforehand. Another important limitation is that the boundaries in the classifications separate objects (products or customers) strictly into different categories regardless of the real difference of the objects from each other. This may cause the situation that very similar objects near the boundary are located in separate categories and become subject to very different policies. To overcome these restrictions, e.g. statistical clustering techniques can be applied to support the analysis. There has recently emerged another technique, artificial neural networks, which may be used to deal with the complexities in classification problems (Venugopal and Baets, 1994). Its applicability in the problem presented in this paper is an interesting subject for further research.