| Decision analysis (DA) has been used successfully to analyze complex problems in strategy and policy, but has seen less application in operational decision making. High-technology product transition decisions are an especially fertile area for the application of DA to operations because of their complexity, frequency, and the stakes attached to them. My thesis is that DA principles and techniques allow high-tech operators to communicate and manage risk while providing a more powerful means of marshalling available data, information, and knowledge to support decision-making on a fast, flexible basis.; I begin my dissertation with a strategic overview of product transitions in high technology. I present a simple model, what I call the Four Factors Model, to describe the thrust, frequency and direction of product transitions as shaped by the strategic intent of a high-tech manufacturing enterprise. Next I describe a framework based on DA principles, techniques, and tools called SCRAM (Supply Communication/Risk Assessment and Management), that enables marketing, materials, operations, sales, and engineering groups in a high-tech enterprise to work together to attack complex operational problems. The aim of SCRAM is to provide a basis for credible analysis in a manner that allows a firm to retain, record, and grow its base of organizational knowledge.; Drawing on methods from statistical decision theory such as conjugate probability distributions and preposterior analysis, I next develop an information value measure called the Expected Value of Customer Data (EVCD). Firms can use the EVCD to guide data-mining through large corporate databases, to price market intelligence, to structure informated interactions with customers, or to identify and resolve organizational disagreements with large impacts to high-stakes transition decisions.; Finally I develop an options model, what I call a Data-Generating Option, where demand and the first-period signal on demand are represented by a conjugate Gamma-Poisson pair from Bayesian statistics. The practical application of the data-generating option leads to a business methodology I call Supply Response. Supply Response suggests new ways to design layered or phased architectures in a company's supply net, supported by informated linkages, to help the firm achieve greater responsiveness to market pulse while reducing the total cost of delivering risky products to market. |
