Experimental design in direct mail and the application of Taguchi methods

Abstract

This article considers the use of an experimental design for studying the impact of various factors on response rate in a direct mail application. First we show why factorial designs are superior to “traditional” experimentation in which factors are studied “one-at-a-time.” We then describe 2-level fractional factorial designs. We next introduce Taguchi methods for designing fractional factorial experiments. These methods have been applied primarily in the operations and design engineering areas, but hold great promise for many other settings, including the direct marketing arena. The methods are easy to use and do not require the user to have extensive statistical training.