Information Systems: A Management Perspective

useful cases from previous editions

American Express Authorizer's Assistant

Charge card companies require that merchants receive an authorization before accepting a card for a purchase. Granting credit that should be denied results in annual losses of several hundred million dollars. Incorrectly denying credit reduces sales commissions and annoys customers who choose the American Express card to avoid having a preset credit limit.

Credit requests go from a merchant to an authorization center. Most requests can be granted automatically by a computerized system that uses statistical criteria to detect unusual activity on the card. The 5% of requests that require direct human attention go to credit agents who have 90 seconds to make a decision (the response time guaranteed by American Express). Prior to 1988, the agents had to decide by quickly looking at information in up to 12 different databases.

American Express implemented an expert system called the Authorizer's Assistant to improve the authorization process. (An expert system is an information system that supports the work of professionals in well defined situations requiring expert knowledge.) The Authorizer's Assistant analyzes the 5% of requests requiring human attention, using over 800 rules to simulate the decision process of experienced credit agents. Within 4 seconds, the agent sees a recommendation plus an explanation of the reasoning behind it. The remaining 86 seconds is available to think about the situation and ask any necessary questions. The system has increased credit agents' productivity by 20%, recommended 33% fewer credit denials, and improved the accuracy of predictions of credit and fraud losses. Although the Authorizer's Assistant is widely cited, it is only one part of a large business process that includes a large-scale transaction processing system for receiving credit requests from merchants, processing those requests, and transmitting the approval or denial back to the merchant.


  1. Describe the parts of this system's operation that do and do not depend on artificial intelligence on the results of artificial intelligence research.

  2. Explain why you agree or disagree with the following assertion: Although this system demonstrates a modicum of intelligence by the way it uses information, it is not as intelligent as any alternative system that would provide positive identification of the card holder.
Sources: Dzierzanowski, James M. et al. "The Authorizer's Assistant: A Knowledge-Based Credit Authorization System for American Express." In Herbert Schorr and Alain Rappoport, Proceedings of the First Annual Conference on Innovative Applications of Artificial Intelligence. Menlo Park, CA: American Association for Artificial Intelligence, 1989, pp. 168-172.;
Leonard-Barton, Dorothy and John J. Sviokla. "Putting Expert Systems to Work," Harvard Business Review, Mar-Apr 1988, pp. 91-98.

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