Metrics Thermostat

Abstract

The explosion of information and information technology has led many firms to evolve a dispersed product development process with people and organizations spread throughout the world. To coordinate such dispersed processes managers attempt to establish a culture that implicitly rewards product development teams based on their ability to perform against a set of strategic metrics such as customer satisfaction, time to market, defect reduction, or platform reuse. Many papers have focused on selecting the right metrics and establishing the culture. In this paper we focus on a practical method to fine-tune a firm's relative emphasis on the metrics that they have chosen. In particular, we seek to advise a firm whether to increase or decrease their emphasis on each metric such that the change in emphasis improves profits. Using a thermostat analogy we apply an adaptive control feedback mechanism in which we estimate the incremental improvements in priorities that will increase profits. Iterations of adaptive control seek to maximize profits even if the environment is changing. We demonstrate the metric thermostat’s use in an application to a firm with over $20 billion in revenue. In developing the metric thermostat we recognize that there are hundreds of detailed actions, such as the use of the house of quality and the use of robust design, among which the product development team must choose. We also recognize that they will act in their own best interests to choose the actions that maximize their own implicit rewards as determined by the metrics. Management need not observe or dictate these detailed actions, but rather control the process by establishing the culture that sets the implicit weights on the metrics. The thermostat works by changing those implicit weights. We define the problem, introduce the adaptive control mechanism, modify “agency” theory to deal with incremental changes about an operating point, and derive methods that are practical and robust in light of the data that firms have available. Our methods include statistical estimation and internal surveys. The mathematics identify the critical few parameters that need be determined and highlight how to estimate them. Both the measures and the estimation are illustrated in our initial application to a large officeequipment firm. The metrics thermostat suggests that this firm has about the right emphasis on timeto- market, but has overshot on platform reuse and has lost its focus on customer satisfaction. We describe how the firm reacted to the recommendations and changed its organization. We describe additional ongoing applications with the US Air Force, the US Navy, and a major automobile and truck manufacturer.