Article ID: | iaor19982151 |
Country: | Netherlands |
Volume: | 53 |
Issue: | 2 |
Start Page Number: | 119 |
End Page Number: | 139 |
Publication Date: | Nov 1997 |
Journal: | International Journal of Production Economics |
Authors: | Kingsman Brian G., Souza Antonio Artur de |
Keywords: | artificial intelligence: decision support |
Versatile manufacturing companies make mainly customised products, competing for each order with other supplier companies on the basis of price, technical expertise, delivery time and reliability in meeting due dates. They include engineer-to-order and make-to-order companies. Versitality is required in continually having to design and configure how to manufacture new or modified products, having continually to deal with varying production loads and having to deal with each customer order individually, even if it is for a very similar product to one sold earlier. A major problem is determining the cost of producing the order and then the price to be quoted. The standard textbook approach of first making a cost estimate, using activity-based costing for example, and then adding some pre-determined profit margin is not how companies do, and have to, operate in practice. Estimation on the basis of the times required for the various production processes needed plus the costs of the materials is only a starting point. Research into a number of such companies has shown that cost estimation and pricing has to be regarded as a single process, it cannot be separated into two distinct activities. It is a complex process, not only requiring the manipulation of known information, but also requiring extensive use of managerial experience and judgement. A model of the cost estimation and pricing process is presented, focusing on the factors influencing the process at the different decision stages in the treatment of a customer enquiry and the rules the cost estimators and bid managers apply when using their judgement to decide about these factors. The main sources of bias and errors that cost estimators make are also discussed. Almost 200 heuristic ‘expert’ rules have been identified. A prototype decision support system incorporating the process model and the rules has been developed. The general types of rules and the form of the DSS are reviewed in the paper.