A Cure for the Decision-making Disorder
Five centuries ago, the Roman Catholic Church introduced the devil’s advocate for the express purpose of countering arguments in favor of canonization. It’s too bad that Time Warner, Yahoo, and George W. Bush did not have devil’s advocates at hand when they decided to buy AOL, spurn Microsoft’s purchase offer, and invade Iraq. Toss in decisions to invest in subprime mortgages and hedge risk with credit default swaps, and it is the contention of Thomas H. Davenport that we are living in an era of decision-making disorder. It is a disorder that Davenport, in a recent article1 in Harvard Business Review, lays squarely at the door of our willingness to view decisions as the prerogative of the individual. The author characterizes this process as little more than a black box. Businesses, including medical organizations, must take a less idiosyncratic and individual approach to decision making. Break down the decision-making process into its components: the decisions that need to be made, the information required, and who needs to do what. These are the stakes: From the chair of the department to the chief clinical officer, decisions will make or break an imaging service, particularly in this fast-moving and rapidly evolving environment. Davenport suggests four steps in his framework for improving decision making. First, list the decisions that need to be made and identify which are most important. Otherwise, all decisions will be treated equally, and some deserve more time and attention than others. Second, take inventory of the factors that go into each decision, such as who plays what role in each decision, how often is it encountered, and what information is available to support the decision. Third, plan the intervention. Design the organizational roles, processes, systems, and behaviors that should be used to make decisions. The author contends that the approach to effective decisions must be broad and must include all aspects of the decision process, including the much-overlooked execution of decisions. Fourth, this process needs to be institutionalized because managers need the tools and assistance to make decisions on an ongoing basis. Chevron, for instance, has a decision-analysis group with members who coordinate data, build decision models, and assist and assess managers in the decision process. Organizations that make this effort must take the final step and assess the quality of the decision at a reasonable interval after the decision has been made. For Example In a highly competitive market with price pressure (such as imaging), pricing strategy is extremely important. Davenport describes the approach to pricing taken by tool maker Stanley Works, which has operated a pricing center of excellence since 2003. Pricing (along with sales and operational planning, fulfillment processes, and lean manufacturing) was one of several decision domains that the company identified as critical to its success. This center is staffed by a director, internal consultants from each business unit, and IT and data-mining experts, and over time, it has instituted a number of pricing methodologies. Currently, the center is focusing on pricing optimization, with the recommendation that the business-unit managers be given more responsibility in pricing their products. The center regularly schedules what it calls gross margin calls to share successes and failures among business units. A measure of automation has been added, including a process for authorizing promotional events. In addition to participating in project startups, coaching, and mentoring, the center uses white-space (or gap) analysis to analyze customers’ sales data and identify opportunities. The results have been impressive: Gross margins have grown from 33.9% to 40% in six years, and the company estimates that the changes have delivered more than $200 million in value to Stanley Works. Davenport contends that analytics and decision-making automation are among the most powerful tools available to help organizations improve decisions. Nonetheless, these tools cannot operate in the absence of human intuition and judgment. For example, decision automation is used in nearly all mortgage and insurance-policy decisions in the United States, but also was a likely contributor to tremendous bank losses associated with subprime loans. To guard against excessive reliance on these tools, companies should warn their managers against building and using models that they don’t understand; should make the assumptions behind those models abundantly clear, so that when changes take place, the model will become suspect; should track how well the models are working; and should cultivate human backups who can revise decision criteria over time and can understand when an automated algorithm is obsolete. They should also be aware of when analytics and quantitative models don’t work well: when a decision needs to be made very quickly, or in a rapidly changing environment (such as health care today).