There are three major approaches research designs to identifying and measuring differences between market segments. The first and oldest is the a priori design, which starts by selecting the basis for segmentation using such variables as demographics and then proceeds to collect data such as product usage or ownership media habits and attitudes. The results show how the segments vary with respect to such variables as purchasing behavior overall size and worth and media exposure.
In contrast, the second approach to segmentation uses a cluster based research design, which groups respondents on the basis of their similarities with reference to some set of selected variables – the most common of which deal with attitudes, needs, benefits wanted, and lifestyle. The third approach is a combination of the cluster and a priori approaches. An example of this approach is where a sample of consumer is first divided into users and nonusers of a particular brand, and then the respondents in each segment are clustered on the benefits wanted.
A Priori Segmentation:
This segmentation model selects in advance both the basis for segmentation and the set of descriptions to be used. Thus, the number and type of segments are predetermined. Further, certain assumptions are made about the relationships between the basis for segmentation and the chosen descriptors.
Example of an A Priori Study:
An example of the results of a priori type of market segmentation study based on demographic measures can be done. This study used economic class, occupation of family head, and age of homemaker as bases for segmentation. The percentage of families in each segment who brought the product and each segment’s annual purchase rate (expressed as the number of cases purchased per 1,000 buying families) are shown. While consumption of this product was not heavily concentrated in any one segment, upper income families, families of craftspeople and skilled laborers and families in which the homemaker was 35-44 years old were clearly the largest buyers. With this information as a guide, product manufacturers would concentrate their marketing efforts on these segments.
One of the problems concerning the use of three separate variables as the basis for segmentation is identifying which one best explains product usage. It is likely that some combination of the three would be more powerful in explaining but the number of segments would increase dramatically. If, for example the three variables are combined a total of some 80 segments emerge) for economic classes X five occupations X four age groups)
A priori market segmentation studies require large samples and the use of structured non-disguised data collection forms. Thus, for example, to have any reliability at the individual segment level when many segments are involved would require an overall sample size of thousands. The data from market segmentation studies of this type would probably be collected via telephone, although mail questionnaires could be used under certain conditions, as could purchase diary reports from a consumer panel. The measures needed for such a study are demographic characteristics product or brand purchased and consumption rates for each household.
Evaluation of Priori Segmentation:
In most a priori segmentation studies, the information obtained is relatively useful, easy to obtain and helpful in obtaining a better understanding of the marketplace. It is particularly useful to advertising managers in identifying media vehicles that can best communicate with the firm’s more promising customers.
It should be recognized that a priori studies may not reveal in any precise sense the identity of groups that possess different purchase behavior patterns with respect to a particular product; that is, the descriptions used may not explain the variations between the various segments with respect to the data sought. In a study of market segmentation and total household purchases of a number of grocery products, one researcher defined – on a priori basis – mutually exclusive segments of consumers according to certain household socio demographic descriptors. His analysis showed little homogeneity of purchase behavior within segments and few ascertainable differences between segments – an indication that the socio demographic descriptors used did not represent market segments.
Cluster Based Segmentation:
In this kind of segmentation model the number and type of segments are not known in advance. Rather, respondents are clustered on the basis of their similarities with regard to a selected set of variables such as benefits wanted, lifestyles, and attitudes. In a manner similar to a priori studies, the size and other market characteristics are then obtained.