Beyond Market Segmentation
Beyond Market Segmentation
Beyond Market Segmentation
Operational obstacle: This apparently reasonable approach has run into one major operational obstacle: the fact that consumers do not cooperate. Efforts to use people characteristics to identify groups of consumers with homogeneous purchase behavior have been notably unsuccessful. For example, in a survey of data on market segmentation of this sort, 2 Ronald E. Frank reported that the average co-relation between people characteristics and purchase behavior is lower than 0.2. Using 17 demographic, sociographic, and personality variables, Frank was able to account
2.
In another study, conducted by the Advertising Research Foundation, almost no relationship was found between personality and preference for various types of toilet paper.7 Thus, on the basis of the empirical evidence, one would have to conclude that personality has not been shown to be an effective basis for market segmentation. Still another interesting sociopsychological concept, which interacts with social class, is reference group. Initially articulated by Herbert Hyman, this idea suggests that, for some people, behavior is influenced by their perception of appropriate behavior for members of groups to which they belong or aspire to belong.8 The reference group notion suggests, then, that one's current social class is an index of purchasing behavior only insofar as one identifies himself with it, rather then with another social class. From a marketing point of view, a consumer's reference group may be even more difficult to identify than his social class. The reference group concept suggests strongly that, for specific markets, the analysis of socialclass structure may be misleading. If a product such as a color television set is seen by the bulk of its consumer group (working-class families} as a symbol of upward mobility, portraying it in working-class surroundings might prove disastrous. According to the reference group hypothesis, it would probably be more effective to display the product in an obviously middleclass setting. Family life cycle, another sociographic classification scheme, relates purchasing behavior to the family's stage in the normal life cycle. The major stages of the life cycle are hypothesized to be: (a) single; (b) married no children; (c) married children in the home; (d) married - no children in the home; (e) single widow(er). Obviously, consumption of some products and services, such as diapers and baby food, is going to be directly related to life cycle. Knowledge of the proportion of families in the various cycle groups can therefore aid in estimating sales potential. However, life cycle is too insensitive a measure for establishing preference patterns' within product categories. Usage Patterns: Another method of segmentation is based on patterns of product usage i.e., an analysis of the various uses to which a particular product or brand is currently put. Usage can, of course, vary for several reasons. On the one hand, a product such as all-purpose flour may be used by most consumers for several different applications. On the other hand, it may be used for a single but different application by several groups of consumers. In the latter case, profitable "people" segmentation might be possible, and thus it merits investigation. Segmentation by use is described in an article by Daniel Yankelovich, who recommends analysis of various product markets on the basis of several modes: patterns of usage, values derived from usage, aesthetic preferences, and buying attitudes and motivations.9 Unfortunately, the article includes no information on how one would perform these analyses or on how one would decide on the appropriate method of segmentation. Yankelovich's version of segmentation analysis appears to he largely intuitive. ------------------------------------------------------------------------------------------------------------------------------
As markets have become larger and more competitive, the market planner's primary function has shifted from forecasting the gross size of a market to estimating the probable performance of his current or projected brand in the market. As we have seen, traditional methods of segmenting markets tend to break down when it comes to making these fine distinctions, Recently, a new approachproduct segmentation has been reported in the literature. This method, which segments by perceived product or brand characteristics, shows promise for marketers in predicting brand share. -----------------------------------------------------------------------------------------------------------------------------
Brand Preference The major difference between the traditional techniques and the new approach is suggested by their titles: market segmentation, concentrates on differences among people who comprise markets; product segmentation concentrates on differences among products which comprise markets i.e., compete with each other, For the marketer, however, the important difference is that product segmentation promises to be effective in explaining differences in preference for one brand versus another. In its early applications, it has shown unusual promise for predicting behavior toward new produces. At present, product segmentation appears to require three steps: 1. Learning how consumers differentiate among brands and products they see as constituting a market-that is discovering the product characteristics, real or imagined, which make each brand different from others in the category. Building descriptions of possible new products from new combinations of product characteristics, or from combinations of old characteristics plus feasible new characteristics, and then evaluating consumers' preferences for these descriptions over the current brands in a large-scale national survey. Selecting the new product description that attains the desired preference level and building the new product so that consumers see it as matching the chosen description.
2.
3.
Although the use of this approach for new product development has been too limited to draw firm conclusions about its validity, the mere possibility that new product performance is predictable warrants further consideration. I shall briefly discuss two points of view on product segmentation, Homogeneous Products: In one of the first articles on segmentation by product characteristics Alfred A. Kuchn and Ralph Day suggested that marketers view consumers' preferences for a product, such as chocolate cake mix, as being normally distributed along some key dimension (e.g., "chocolatyness") 11. If consumer preferences are in fact heterogeneous, then some consumers will prefer more of the key ingredient i.e., a more "chocolaty" cake mix than will others. The authors demonstrated that two of the most popular new-product evaluation research practices preference among variants and the paired comparison test tend to lead researchers to create products which are more or less homogeneous. When the first test is used, the most popular variant is generally chosen for production. Competitors' subsequent paired comparison tests wilt result in later entrants being similar. It is important, Kuehn and Day asserted, that key product qualities be distributed among brands in a manner parallel to consumers' preferences for the qualities.
2.
When a description of a proposed new product indicates definite consumer preference for the new product over current brands, it is possible to capitalize on that market potential by developing the product so that it is perceived by consumers as matching its description. The contrast between this approach and market segmentation should become obvious on reflection. Consumers' preference is measured in terms of product characteristics rather than the characteristics of a hypothetical consumer population. (The term "product characteristics" is used here in the broad sense and includes "perceived" as well as "real" properties.) Thus the process of segmentation becomes a search for new, and as yet nonexistent, combinations of product characteristics for which there is significant unmet consumer demand; who prefers the new brand or product is relatively less important. When viewed in this manner, product segmentation is a systematic and superior way of searching for
Research techniques Volney J. Stefflre and I have argued that there are four distinct research steps involved in discovering and evaluating product segments in the typical consumer packaged goods market.12 Further thought, however, has made it clear to me that the philosophy of product segmentation is more important than any specific set of techniques. While the techniques I shall discuss in this section have been successful in predicting new product performance, it is certainly possible that alternative techniques could lead to equal accomplishments. Our four research steps include: 1. 2. 3. 4. A "brands and usage" or "item-by-use" study to determine which brands are seen as competing for the same uses, A judged-similarity study to determine which brands arc seen as similar, and what characteristics of brands consumers use in deciding whether brands arc similar. A small-scale preference study to discover the proportion of consumers who like particular current brands, and who also prefer the proposed new products described. A large-scale national probability preference study ID discover, in the case of each of several new product descriptions, the proportion of consumers who prefer it to their current brands.
Specific steps Here is how each of the four research objectives just outlined is accomplished. First, determine what brands and products consumers see as competitive with one another. Consumers may or may not agree with the manufacturer's view of the market, or with the brands and products assumed by professional service organizations to constitute the market. Researching the interaction between brands in a market and the uses to which such products are put allows one to arrive quickly at a list of the items consumers see as comprising the market. Second, collect consumer judgments of similarity among the products comprising the market and their reasons for making such Judgments. The similarity data are used as input for statistical computer programs, such as factor analysis or multidimensional scaling. When analyzed, these data can show, by means of physical models, how the many brands and products in the market are perceived in relation to each other. Generally three-dimensional configurations, these models are physical representations of what I have earlier called the market structure. Since the reasons consumers give for perceiving competitive brands as similar are recorded during the judged similarity interview, it is possible to apply these criteria to the model of the market structure. In so doing, the marketer can begin to understand what perceived characteristics determine the positioning of each brand.
For example, if Brand A is seen as most similar to Brands D, F, H, and P, one can analyze the reasons consumers give for the A-D, A-F, A-H, and A-P similarity judgments. By considering together the various characteristics they list, one gains insight into the conglomerate characteristics Brand A is seen as having and begins to gain an understanding of the way in which consumers segment competitive products. With this accomplished, it is then possible to hypothesize new combinations of current characteristics, and/or combinations of current characteristics and other feasible characteristics not currently found in the product category, thereby identifying possible new product opportunities. Third, determine where these new product segments or descriptions, "fit" in the current market structure by doing small-scale preference research. For example, consumers are asked to rank current brands in the order of their preference for them, and then to rank the descriptions of the proposed new products in the same manner. Subsequent comparison of these rankings provides a rough indication as to the brands with which each product description is competitive. Once all the descriptions are ranked in the current market structure, the marketer can choose for further evaluation those descriptions which appear to fit in areas of high potential. Fourth, conduct a probability preference study among a projectable national sample of consumers in order to make a final evaluation of the new product descriptions. In such a study, the end result is a representative measure of the proportion of consumers who prefer each new description to their current brand. Thus the profit-oriented manager can determine two important things about each new product description tested: 1. 2. What total preference exists for each possible new brand. What incremental preference his company would enjoy with each new product description. (He can anticipate the favorable shifts that would likely occur from both competitive brands and his company's own current brands, if any, to his proposed new products.)
Product bundle Once a new product description that promises to meet the desired marketing objectives is identified, the company has to develop the actual new product so that consumers will see it as the best existing example of the successful description. Briefly, this process includes discovering what psychological stimuli (colors, smells, shapes, textures, scenes, and so on) are associated with the descriptive words, and developing, through an evolution of consumerperception testing, each of the integral components (i.e., package, product, name, and advertising). The resultant "product bundle" is then tested to determine whether consumers see it as the best currently available example of the target description, and if they behave toward it as they did toward its description.
10
11
Share of total market Share from competitive products Brand A Brand B Brand C Brand D Brand E Brand F Brand G
12