Applying Archetypal Analysis in Marketing Research
Applying Archetypal Analysis in Marketing Research
© Paul Riedesel, Action Marketing Research 2008
Archetypal analysis is a mathematical procedure for decomposing a multivariate dataset as a function of a set of underlying archetypes or ideal types. While used for many years in the physical sciences, these methods were first introduced to marketing research practitioners by Louviere and Carson at the 1998 Advanced Research Techniques Forum. Little public use has been made in our field since then. 'Tis the pity. Archetypal analysis is an interesting alternative to the more-familiar methods of cluster analysis used to represent consumer heterogeneity (segments). And it avoids a major fallacy in virtually all segmentation solutions—one we suspect every good researcher is aware of, but few talk about. This note seeks to elevate the practice of archetypal analysis within marketing research by: Explaining its rationale and contrasts to conventional segmentation Reviewing the algorithm Illustrating its application via a major study of American Baby Boomers Discussing practical issues in analyzing and reporting archetypal data Why Bother with Archetypes? Our argument is not against cluster-based segmentation, which can be very useful. Rather, our argument is for adding a new tool that can be a respectable alternative to understanding and acting on consumer heterogeneity. The basic rationale for archetypal consumer analysis and cluster-based segmentation is the same. Consumers (a term meant to include business buyers) differ in their attitudes, needs, behavior, and other characteristics. Because it is usually impractical for a firm to create a unique marketing mix for each consumer, firms rely on various means of grouping consumers. Some segments may be given higher priority. Different products, distribution channels, marketing communications, etc. may be aligned with each segment. A fundamental problem faced by researchers as they process segmentation data is that the
© Paul Riedesel, Action Marketing Research 2008
Archetypal analysis is a mathematical procedure for decomposing a multivariate dataset as a function of a set of underlying archetypes or ideal types. While used for many years in the physical sciences, these methods were first introduced to marketing research practitioners by Louviere and Carson at the 1998 Advanced Research Techniques Forum. Little public use has been made in our field since then. 'Tis the pity. Archetypal analysis is an interesting alternative to the more-familiar methods of cluster analysis used to represent consumer heterogeneity (segments). And it avoids a major fallacy in virtually all segmentation solutions—one we suspect every good researcher is aware of, but few talk about. This note seeks to elevate the practice of archetypal analysis within marketing research by: Explaining its rationale and contrasts to conventional segmentation Reviewing the algorithm Illustrating its application via a major study of American Baby Boomers Discussing practical issues in analyzing and reporting archetypal data Why Bother with Archetypes? Our argument is not against cluster-based segmentation, which can be very useful. Rather, our argument is for adding a new tool that can be a respectable alternative to understanding and acting on consumer heterogeneity. The basic rationale for archetypal consumer analysis and cluster-based segmentation is the same. Consumers (a term meant to include business buyers) differ in their attitudes, needs, behavior, and other characteristics. Because it is usually impractical for a firm to create a unique marketing mix for each consumer, firms rely on various means of grouping consumers. Some segments may be given higher priority. Different products, distribution channels, marketing communications, etc. may be aligned with each segment. A fundamental problem faced by researchers as they process segmentation data is that the