Design and process solutions for decreasing vendor defects

Abstract

Why do some new initiatives fail while others succeed? This thesis attempts to answer this complex question by investigating the failure of a defect tracking initiative at Amazon and examining how a reintroduction of the process succeeded. Amazon is a very complex and dynamic organization that has both a corporate headquarters as well as a regional fulfillment centers. In distributed organizations like this, successfully implementing network-wide process improvements can be critical to the success of the business. Consequently, Amazon is constantly attempting to roll-out new improvement efforts. Unfortunately, some of these initiatives fail to reach their full potential. Common suggestions for the failure of initiatives include lack of management support, poor technology, limited communication, or lack of vision. This thesis argues that while these suggestions are valid and important when designing a process, these alone cannot by themselves show whether the initiative is destined for success or doomed to failure. To better understand the success rate of the initiative, one must do a deep dive into the incentive structure, motivations, and perceptions of all stakeholders involved in a new process. In 2010, Amazon introduced a defect tracking tool. Adoption grew during the first five months, but declined rapidly during the second half of the year. In 2012, 1 reintroduced the same defect tracking tool but made a number of minor changes to the process. Over the course of 2012, the project resulted in nearly 300,000 defects and nearly one million dollars in vendor chargebacks. More than 50,000 defects were filed in the month of December alone, compared to less than 4,000 in the month January. Approximately 1000 people used the tool at more than 30 different sites. This thesis illustrates how system dynamics modeling of the Amazon defect tracking process can be an effective tool for a more complete understanding of adoption or abandonment rates. At a broader level, this thesis discusses methods for designing new processes or modifying existing ones so they are more likely to succeed.