Information Systems: A Management Perspective

useful cases from previous editions

Digital Equipment Corporation: Building an Expert System

Digital Equipment Corporation's business strategy in the late 1970s and 1980s was selling customized VAX computer systems. Although the strategy had many competitive benefits, it gave rise to a large number of valid configurations of a technically complex product set. In too many cases, customer shipments were delayed by the lengthy final assembly and test process, and the installation at the customer site was delayed because of missing or incorrect parts. After trying typical data processing approaches to this problem, Digital turned to an expert system approach.

Digital's XCON (expert configurator) system was one of the first expert systems in daily production use in industry. Starting with a customer order and configuration guidelines, XCON adds or deletes components necessary to make the order complete and correct. It also assigns each component to a location, configures cable layouts and connections, and creates other information needed for installation. Its outputs include a line-item summary of the order; detailed configuration drawings; and explanatory text concerning parts not needed, spare parts, unused capabilities, and other information. Prior to XCON, this work was done by people called technical editors. According to some surveys, their work was completely accurate only 65 percent of the time. The impetus to build XCON came from the cost and delays of testing the configuration by using it to assemble a computer in a plant that covered 13 acres and contained $20 million in inventory. Company growth projections indicated that these costs would become even larger.

In 1979, AI researchers at Carnegie Mellon working with Digital personnel created a prototype of XCON containing 210 rules. It was validated using 50 representative cases on which it made two significant errors and ten minor errors. After a pilot implementation in 1980, the system was in regular use by 1981. XCON largely automated the work of the technical editors. In 1981, 23 technical editors created 19,000 system configurations by hand. In 1986, 17 editors created 60,000 configurations with better quality and detail using XCON, although the XCON system eventually required a staff of 40 programmers to maintain the rules it contained. By the end of the 1980s, XCON had grown to more than 18,000 rules, 40 percent of which changed each year. Benefits include supporting three times the order level with fewer mistakes. Estimated net return to Digital was more than $40 million per year.

  1. Use the WCA framework to organize your understanding of this case and to identify important topics that are not mentioned.

  2. Explain how the development of this expert system fits into the various system life cycle approaches.

Sources: Barker, Virginia E., and Dennis E. O'Connor. "Expert Systems for Configuration at Digital: XCON and Beyond." Communications of the ACM, Vol. 32, No. 3, Mar. 1989, pp. 298-312;
Sviokla, John J. "An Examination of the Impact of Expert Systems on the Firm: The Case of XCON." MIS Quarterly, June 1990, pp. 127-142.

Back to Useful Cases from Previous Editions


Main Page About the Book Student Resources Instructor Resources Contact Us

©1999 Prentice-Hall, Inc.
A division of Pearson Education
Upper Saddle River, New Jersey 07458

Legal Statement