Professor Art Swersey's New Book Takes Experimental Design from the Factory Floor to the Frontlines of Marketing
New Haven, Conn., June 1, 2007 — Statistical experiments are powerful and practical tools in business. A well-designed experiment can help companies be more competitive by pinpointing areas to reduce costs, increase productivity, and improve quality. Although much has been written about the use of experimental design in solving manufacturing problems, little attention has been given to its non-industrial applications.
In the new book Testing 1-2-3: Experimental Design with Applications in Marketing and Service Operations (Stanford University Press), Yale School of Management Professor Art Swersey and co-author Johannes Ledolter of the University of Iowa demonstrate how the same experimental design methods long used to optimize manufacturing processes can also add value to marketing and service activities such as website design, internet advertising, and direct mail marketing.
“Business leaders are beginning to realize that experimental design has widespread applications to management decision making, particularly in service organizations,” said Swersey, a professor of operations research.
The authors also make the case that the way that most experimentation is done in business today – by changing one factor at a time while holding other factors constant – is highly inefficient and can lead to wrong conclusions. The better method, as the authors show, is to test all factors simultaneously. Doing so not only reduces the time and cost of experimenting, but also provides the decision-maker with better information.
Swersey and Ledolter demonstrate the benefits of multi-factor experimental design in marketing and services through real-world examples and case studies. They show, for example, how a major magazine publisher increased its sales in a supermarket chain by 20% by simultaneously testing the effects of 10 different factors, such as placing additional display racks in the store or using on-shelf advertising. In another example, subscription-based long-distance service provider PhoneHog improved the customer click-through rate on its website by 35% through an experiment that varied its website design using 45 test web pages. Without a multi-factor experiment, 1,658,880 test web pages would have been required to test all possible combinations of the website design.
Other cases in the book examine how experimental design boosted the circulation of Mother Jones magazine by improving its direct mail response rates, and how a leading office supply retailer attracted more small business customers to its retail and online stores by testing and improving its email advertising copy.
In their years of experience in teaching operations research, Swersey and Ledolter have found that the best way to learn about how to design experiments is by solving exercises, analyzing real data sets, and designing and carrying out experiments. In the book, they include exercises in each chapter that allow readers to test the tools and concepts for themselves.
Arthur J. Swersey is Professor of Operations Research. Professor Swersey’s expertise is in quality management, operations management, and mathematical modeling. He has done research on siting vehicle emissions testing stations, and school bus scheduling, and has developed queuing models for the New York City Fire Department. In his current research he has devised a mathematical model for predicting the severity of prostate cancer based on biopsy results and prostate specific antigen (PSA) levels.