Consumer Self-Selection and Segments of One: the Growing Role of Consumers in Segmentation

David W. Stewart, University of Southern California
ABSTRACT - Variance in consumer behavior is increasing and the source of this variance is increasingly under the control of the consumer. This increase in variance and the degree of control exercised by consumers poses new challenges for segmentation research. Increasingly, consumers are segmenting themselves through a process of self-selection guided by idiosyncratic purposes. The present paper suggests that segmentation strategies should begin with identification of the purposes that consumers seek to achieve through the purchase and use of products. These purposes ultimately lead consumers to place themselves in particular situations in which they seek specific benefits. This view is consistent with control theories of human behavior and recent work on consumer self-selection. The paper offers an approach to segmentation that constructs segments from building blocks defined by the individual consumer in a specific usage context.
[ to cite ]:
David W. Stewart (1991) ,"Consumer Self-Selection and Segments of One: the Growing Role of Consumers in Segmentation", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 179-186.

Advances in Consumer Research Volume 18, 1991      Pages 179-186

CONSUMER SELF-SELECTION AND SEGMENTS OF ONE: THE GROWING ROLE OF CONSUMERS IN SEGMENTATION

David W. Stewart, University of Southern California

ABSTRACT -

Variance in consumer behavior is increasing and the source of this variance is increasingly under the control of the consumer. This increase in variance and the degree of control exercised by consumers poses new challenges for segmentation research. Increasingly, consumers are segmenting themselves through a process of self-selection guided by idiosyncratic purposes. The present paper suggests that segmentation strategies should begin with identification of the purposes that consumers seek to achieve through the purchase and use of products. These purposes ultimately lead consumers to place themselves in particular situations in which they seek specific benefits. This view is consistent with control theories of human behavior and recent work on consumer self-selection. The paper offers an approach to segmentation that constructs segments from building blocks defined by the individual consumer in a specific usage context.

INTRODUCTION

Consumer researchers, like most social scientists are keen observers of behavior. They may differ is what they choose to observe and the method of observation, but observation is the most basic of the research tools used by researchers interested in consumer behavior. Researchers observe responses to paper and pencil tasks such as attitude measurement scales. They observe actual purchase behavior via scanner panels. They observe behavior of individuals who are shopping and who are engaged in various consumption experiences. They even observe consumers observing themselves and reporting on the results of these observations.

Observations, of whatever types, are used to make inferences about behavior in the context in which it occurred. Sometimes the observed behavior can be easily interpreted; sometimes it appears purposeless and random. The latter types of behavior have led to the suggestion that behavior, at least at the individual level, should be treated as a stochastic process (Bass 1974) since no explanation can be discerned. Other researchers have suggested that certain types of behavior are governed by automated processes that involve no conscious thought or decision making (Olshavsky and Granbois 1979) and is, thus, in some sense without explanation. Still others (Belk 1987, Lincoln and Guba 1985) have suggested that explanation should not even be an objective of consumer research. Rather, consumer researchers should be content to describe behavior in great detail. These views of consumer behavior are not comforting for researchers seeking to build deterministic models. These views are outright discouraging for marketing managers seeking some means for influencing the behavior of consumers in the market place.

Even among those researchers who seek deterministic explanations for consumer behavior the results of observation and inferences have been disappointing. Although behavior can often be predicted well, it is seldom explained very well. Even the most controlled of experiments tend to produce results that account for ten percent or less of the variance in observed behavior. Such disappointing outcomes have led to a certain disillusionment with consumer research. This disillusionment, while not universal, has resulted in significant reductions in research staffs and expenditures on marketing research in many organizations.

Among the many tools used by consumer researchers and marketing managers alike, no tool has received as much attention and use as market segmentation. Despite this attention and frequency of use (or perhaps because of it), segmentation has been the object of frequent criticism. Davis (1987) has suggested that segments are almost always too few and too heterogeneous to be of practical value. McDonald (1985) has questioned whether any segmentation system, including recent approaches based on the availability of substantial databases, is useful for developing persuasive marketing programs. There has long been a debate about the best basis for segmentation (Dickson and Ginter 1987, Wind 1978, Wilkie and Cohen, 1977, Frank, Massy, and Wind 1972, Haley 1968), but recent criticism -is directed at the concept of segmentation as a marketing strategy. This raises the question of whether segmentation has ceased to be viable as a marketing strategy, and if so, why? To answer this question a brief discussion of the relationship between variance and segmentation is required.

SEGMENTATION AND VARIANCE IN CONSUMER BEHAVIOR

It is axiomatic that any behavior, including consumer behavior, is interesting only to the extent that it exhibits variance. Indeed, it is only because variations in behavior exist among individuals and within the same individual over time that causal mechanisms can be identified. To a large extent, the very recent origin of the science of consumer behavior may be attributed to the general paucity of variance in behavior among most consumers. When most consumers have little money to spend and few options for spending it, consumer behavior is decidedly uninteresting.

Variance in consumer behavior grew perceptibly after the Second World War as a result of the increase in the affluence of consumers and the increase in the number of alternative products and services available to consumers. During the first forty years after the Second World War, students of consumer behavior and marketing tended to view consumer behavior as a response to marketing stimuli. The variance in consumer behavior tended to be viewed as stemming from variations in the environments of consumers. Over time, marketers came to recognize that systematic differences in the variance of consumers could be identified and market segmentation was born (Smith 1956).

Market segments have traditionally been defined in terms of the differential response (variance) of consumers to marketing stimuli-product, price, market communications. For many products, a relatively small number of segments appeared to capture most of the variance in consumer response. This did not mean that researchers failed to recognize that consumers exercised some judgment in the selection of their response, or that researchers regarded consumers as purposeless respondents to marketing stimuli. Nevertheless, when the alternatives from which consumers may choose are limited in number, it is easy to view consumers as passive respondents to marketing actions. After all, until very recently consumers had a choice of only three television networks, a modest selection of- conveniently located shopping outlets. and a limited number of options with respect to product or service offerings.

In situations characterized by small variance as a result of limited options for response, the number of groupings of homogeneous consumers will, by definition, be small. In fact, in the extreme, there will exist but one group or segment. The latter circumstance may arise because consumers are actually homogeneous, but it is just as likely to arise because consumers do not have alternatives for expressing the very real differences that exist. It follows that as more alternatives are available to the consumer, more groups (segments) may emerge as expressions of the differences that exist among consumers. It is the thesis of the- present paper that current dissatisfaction with segmentation reflects the growth in the number of options available to consumers and the concomitant increase in the number of segments that exist. In fact, the case can be made that the number of options available to the consumer is increasing exponentially. In the limit, the number of potential segments will equal the number of consumers when the options available to the consumer becomes very large. Segments of one are a natural consequence of the growth in alternatives available to consumers.

Segments of one may sound impractical to the marketing manager, and there will certainly be many circumstances where some aggregation is required if efficient marketing programs are to be designed and implemented. Thus, efficiency alone will force managers to look for segments of some type. Nevertheless, the approach to segmentation will need to change in order to accommodate the increase in the variation in consumer behavior. The remainder of this paper will discuss the reasons for the increase in the variance in consumer behavior, and suggest a conceptual outline for a segmentation approach that can accommodate this increase in variance.

CONSUMER ALTERNATIVES, SELF-SELECTION AND CONTROL THEORY

Recent years have seen a proliferation of alternatives from which the consumer may choose. Media options, both in broadcast and print, have grown exponentially over the past several years as cable television penetration has expanded. This proliferation of media is continuing with the advent of international television via satellite and cable, interactive computer networks such as Compuserve and Prodigy, and telephone information services. At the same time, the number of options for obtaining many products and services, the number of points of distribution, has also expanded. Catalogs, direct response, telephone, and interactive computers have taken their place as competitors to the more traditional retail outlets.

Product and service options have also proliferated as firms have made use of computer assisted-manufacturing to provide more variations of basic designs. For example, R. R. Donnelly has been a pioneer in "selective binding" whereby magazines are "customized" for individual consumers in terms of both editorial material and advertising. Automobile manufacturers are experimenting with the production of customized option packages that will be available for delivery within 72-hours. Insurance companies now offer "cafeteria" plans that allow consumers to construct their own benefits from an extensive menu of offerings. Davis (1987) has refereed to the proliferation of product and service options as mass customization. Such mass customization has been made possible by the merger of information, manufacturing, and service delivery technologies.

As noted above, one important outcome of these increasing options is that the variance in consumer behavior is increasing and an increase in variance carries with it an increase in the number of potential segments within the market. More important, a significant increase in the variance of consumer behavior may require a change in the philosophical orientation of segmentation research. It may no longer be sufficient to ask how consumers may differ in their response to marketing stimuli. It may be more important to ask to which marketing stimuli consumers will attend. Another way of saying this is that it may no longer be sufficient to ask what marketing stimuli do to consumers. Rather, it may be more appropriate to ask what consumers do to marketing stimuli. Segments of consumers will no longer be defined by marketing stimuli to which they respond. Consumers will segment themselves based on their own self-selected behaviors. This change in philosophical orientation is consistent with the growing body of literature on consumer self-selection (Zimmer and Dorfman 1985). This literature makes it very clear that consumers scan their environments for personally relevant stimuli to which they then respond (Cotton, 1985, Pechmann and Stewart 1990, Tolley 1991). Self-selection is, itself, a process. This process, while clearly an empirically demonstrable phenomenon, is not well understood. It is obvious that an individual will selectively attend to personally relevant stimuli. The more important questions are how this process of self-selection works and how this process might be captured in segmentation research. A potentially useful framework for understanding the process of self-selection is found in a psychological theory called control theory (Powers 1973, 1978).

Control theory suggests that it is inappropriate to make inferences about human behavior solely on the basis of the observation of behavior. Rather, it is important to understand the purpose behavior serves for the actor, or consumer. Control theory suggests that the observation of behavior, without an understanding of purpose is meaningless. Indeed, behavior may appear meaningless, even random, without an understanding of purpose. Control theory suggests that human behavior is purposeful and that consumers attend to personally relevant stimuli and engage in strategic behaviors that are designed to realize specific goals. Further, it suggests that behavior, in and of itself, is of no importance. Behavior occurs only for the purpose of controlling consequences that affect the actor (Powers 1978). Note that this is a very different view from the Skinnerian view that behavior is controlled by its consequences (Skinner 1938).

Control theory suggests that an explanation of behavior must include definition of the purpose of the actor. Two identical behaviors may, in fact, be carried out for very different purposes. Two highly dissimilar behaviors may, in actuality, be designed to achieve the same outcome. The interpretation of the meaning of the behavior is not possible without knowledge of the purpose of the behavior, which serves as a kind of reference signal. Marketers and consumer behavior researchers have not ignored the purposefulness of consumer behavior or the self-selection that characterizes much of consumer response to marketing stimuli. Recognition of the purposes that provide the foundation of consumer behavior has led to the use of means-ends analyses (laddering) (Gutman 1982, Reynolds and Gutman 1988) as an approach for linking behavior to consumer purposes. Nevertheless, such analyses are infrequent and have seldom been systematically linked to market segmentation. More often than not, purpose is inferred from the point of view of the observer, rather than from the point of view of the actor. This has tended to be the case regardless of whether experimental observation or more naturalistic observation served as the basis for the inference.

AN ILLUSTRATIVE TALE

Imagine an individual sitting in front of a computer screen and at apparently random intervals moving a control stick in some direction. The probability of moving the control stick increases with the amount of time since the last movement, and the direction of movement, measured in degrees, follows a uniform distribution. Now assume a whole room of such individuals sitting at a computer screen and moving a control stick. Using models that are quite familiar to mathematical modelers of consumer behavior, it would be possible to predict accurately the number of individuals at any given point in time who will move the control stick and the direction of that movement. Missing from this analysis are the reasons the individuals are sitting at a computer screen and moving a control stick. Neither is there any information about how the probability of movement or the direction of movement might be influenced. Dissatisfied with mere prediction, a researcher observing this situation may decide to construct an experiment. He obtains his own computer and control stick. He brings in respondents and instructs them that their task is to keep a pointer on the computer screen within a circle that is moving about the screen. After forty respondents have participated in the task, the researcher observes that there is a direct correspondence, subject to some error perturbation, between the movement of the circle on the screen and the frequency and direction of a respondent's movement of the control stick. The researcher shouts "Eureka!" and races out to observe the circles to which all of the earlier subjects were responding Much to the researcher's chagrin, he finds no circles moving about the computer screens. He knows that moving circles influence the movement of a control stick; he has empirical evidence from the laboratory. But wait, this is a clever and creative researcher. Upon pondering the situation he concludes that the circle does not have to be physically present on the screen. He posits a "latent circle" that exists in the mind of the actor. He reasons that this latent circle guides the behavior of the control stick.

The clever researcher sets out to test his new theory of latent circles. Being well trained at a leading research university he knows that he will need to carry out several studies employing different methodologies if he is to establish the credibility of his theory. He designs two studies. In the first, he observes behavior -and asks a representative group of respondents where they would place a circle that would guide the movements he has observed. With the exception of a couple of troublemakers who insisted they did not use circles, all of the respondents place the circle in the expected location. The data from the two troublemakers cannot be included in the analysis, and they are dropped from the sample. A retrospective study is subject to various confounding effects, however, so the researcher conducts a second study.

In the second study, respondents are asked to imagine that a circle exists on the screen. The task of the respondents is to use a control stick to keep the cursor within the imagined circle on the screen. The researcher finds that there is a direct relationship between the location respondents reported for their latent circles and their movement of a control stick. The researcher concludes that his theory is supported, writes an award winning journal article, receives a chaired professorship, and spends the rest of his days training students to do research on latent circles and pondering the metaphysical implications of latent circle theory.

Unfortunately, all is not well. Latent circle theory just does not help predict how the average person sitting at a computer will move a control stick. If respondents are asked about circles, the theory seems to work retrospectively. If respondents are asked to specify the location of a circle immediately prior to moving a control stick, the theory works well too. But, when measures of latent circles are obtained several days prior to the interaction with the computer they do not predict very well, even when the respondent claims to be using them. Even more troubling is the finding that distractor tasks, like computer games, seem to eliminate the influence of latent circles.

A second researcher, after years of work on latent circles, decides it is time for a new paradigm. This second researcher hypothesizes that latent circle theory is little more than the stage management of behavior. This researcher sets out to visit people who interact with computers and use control sticks. He obtains a large grant from a well known institute, solicits the cooperation of a number of other social scientists who are disillusioned with latent circle theory, and sets out to find users of computers. As he and his band of researchers cross the country, they interview numerous users of computers; some are interviewed even as they use the computer.

The outcome of the second researcher's odyssey produces a remarkable variety of human responses. One respondent used the control stick to play a computer game that demanded considerable eye-hand coordination. This respondent reported an affinity for the game that bordered on the religious. A second respondent, an accountant, reported use of the control stick to access menu commands on a spreadsheet program. This respondent suggested that he only engaged in such behavior because he was paid to complete a particular task, a decidedly secular motivation. Other respondents were interviewed, and each offered his or her own idiosyncratic response.

The second researcher, upon pondering all that he has learned, concludes that latent circle theory is vacuous; it does not capture the richness of responses he has discovered. More important, there are such interesting idiosyncrasies among respondents. Each has his or her own unique interpretation of behavior. Even the same behavior, using a computer and a control stick, can be explained in numerous ways. There is no single explanation, no single reality. He declares the old paradigm dead; long live the GO Observe Demonstrable Verbal Interactive Boundary Expanding Solutions (GOOD VIBES) Paradigm.

The second researcher writes several widely acclaimed papers, develops a cult following in the field, receives a chaired professorship from his university, and spends the remainder of his days enjoying conversations with interesting individuals who are not university professors, have no aspirations to be university professors, and have found meaning in life outside the confines of a university campus.

Alas, there is still discontent in the land. A middle-aged (in years only. Psychologically he/she is in her/his late 20's) marketing researcher sits reading a series of memoranda from corporate management. One informs him/her of the downsizing of the marketing research department (she/he has been spared, for the moment). A second memorandum asks the researcher whether she/he might be interested in a transfer to sales management at a substantial increase in compensation. A third memorandum is from an obviously peeved vice president of marketing who has just caught grief from other members of the corporate operating committee for a decline in sales and an increase in marketing costs. This vice president suggests that the problem is the current segmentation system, which appears no better and perhaps worse than the last two. She is particularly peeved that the advertising designed to attract new users of the corporation's services seems to have high recall among only current users, and on the persuasion measure the current strategy seems to produce a net switch to competitors. On the other hand, the price promotion designed to hold current users in the face of competitive price promotions seems to have attracted some new customers, but has caused some existing customers to complain about reductions in the quality of service delivery and, in some cases, actually switch to a competitor.

This industry researcher is dismayed by the frequency with which the firm has changed segmentation strategies. She is also dismayed by the lack of direction provided by academic researchers. She is a student of researcher one, the founder of latent circle theory. She helped obtain support from her corporation for the work of professor two, who founded the GOOD VIBES movement. She appreciates the contributions or both researchers, but she finds no solution to her firm's segmentation and marketing problems in their work. She wonders if there is a way to combine elements of the two approaches, so she visits some users of computers in her own firm. She asks each in turn what they are doing. Like professor two, she receives a variety of responses. Each user seems to have some unique reason for using the computer. Even those individuals who claim similar purposes have selected different computers, different control devices, and different software. On the other hands several users who report very different purposes are using identical control sticks, identical computers, and even identical software. She also observes that all individuals use a control stick. Apparently, the control stick is used to move a pointer around the screen.

The corporate researcher decides to ask a few more questions. After asking the more general question about purpose, the researcher probes. An interesting pattern begins to emerge. The type of computer, control stick, and software selected appear to be related to each individual's general purpose. Each individual can explain the purpose behind their use of the computer Individual variations in equipment and software used for the same or similar purposes are explained as the outcome of slight differences in what the individual users are trying to accomplish and by differences in the characteristics of the users.

The industry researcher continues to ask questions. She observes each individual moving the control device and asks what the movement is designed to accomplish. Each respondent indicates that the benefit they are seeking from the movement of the control stick is the placement of the cursor at a certain point on the computer screen (could this validate latent circle theory?). The placement of the cursor is not random, however. It is related to particular outcomes that the individual user is seeking, and constraints that are imposed by the software, hardware, and control stick. For example, one individual frequently moves the control stick forward because, he explains, this is the way to move the cursor to a set of menus located at the top of the computer screen. The pattern was complete. The individual behaviors of individual respondents were the outcome of a complex process. The process was guided by the self-selection of specific purposes. These purposes in turn, guided the selection of tools with specific characteristics. The characteristics of the tools, in turn, constrained the nature of specific observable behaviors.

The industry researcher now knows how to segment markets. She rushes to explain her insights to the vice president for marketing. After listening patiently, the vice president of marketing dismissed the plan as too expensive and time consuming. The industry researcher also shared her insights with her mentor, the founder of latent circle theory. He listened patiently and complimented the researcher for her insights. He also suggested that the behavioral process she had identified could never be submitted to empirical test; there were too many variables to control and subjects would have too much latitude for influencing outcomes. The industry researcher also shared her insights with the founder of the GOOD VIBES movement. He also listened patiently and complimented the researcher's insight. He told her that her view was certainly one of many possible truths, but he found that the process she described lost too much of the rich artistic component of behavior for him to find it useful.

The disappointed industry researcher left her job and founded a consulting firm that applied her insights to the analysis of markets. She grew wealthy and traveled throughout the world (since most of her clients are European and Japanese). At last report she was splitting her time between her Beverly Hills home and her chalet outside Geneva.

What is the point of these stories? It is quite simple. Consumer behavior is complex and it is largely under the control of the consumer. Even the constraints on behavior that may be present are frequently the result of a process of self-selection. Consumers choose situations that suit particular purposes and use products and services that are appropriate for achieving these purposes in given situations. Consumers behave in order to achieve specific consequences that are consistent with specific purposes. This view of consumer behavior has some important implications for segmentation research.

SOME NORMATIVE IMPLICATIONS OF CONTROL THEORY AND CONSUMER SELF-SELECTION FOR MARKET SEGMENTATION

Consumer purposefulness and self-selection suggest that individuals interact with their environment to achieve specific consequences. This means that genuinely useful market segmentation must explicitly incorporate consumer purposefulness and the interaction of the individual consumer and his/her environment. The importance of the interaction of person and environment for segmentation research has been discussed elsewhere (Dickson 1982, Punj and Stewart 1983). Rather than repeat this discussion, the remainder of the present paper outlines a conceptual approach for segmentation research.

1. Segmentation should begin with discovery of the purpose(s) for which consumers acquire and use individual products and services. It is the consumer's purpose that drives the processes of information search and acquisition, weighting of information, formation of impressions and attitudes, decision making, and post-purchase evaluation. Research designed to discover consumer purposes is inherently qualitative in is early stages, but in its latter stages, it can also include the construction of taxonomies of purposes, benefits, and use occasions. Consumers self-select from an array of purposes those which require action over the short, intermediate, and long term. Understanding of any particular behavior will be incomplete without an understanding of purpose and the consumer's process of self-selection.

2. Purposes occur within the context of situations. These situations include potential use occasions, benefits sought, and consequences to be avoided. Consumers engage in a conscious process of selecting situations that are consistent with a particular purpose and avoiding those situations that are inconsistent with a particular purpose. A mapping of situations (usage and purchase situations) into purposes can provide insight into the consumer's process of self-selection.

3. Benefits sought from products and services grow directly from purposes and usage contexts. Specific product benefits are, except at the most abstract level, insufficient for determining the purposes of consumers. Benefits (consequences) that are desirable for some purposes and for particular situational contexts may be undesirable and actively avoided for other purposes and in different situations. A mapping of benefits into purposes and use occasions is necessary if consumer behavior is to be predicted over the long term. Benefit segmentation (Haley 1968), is a necessary, but insufficient basis for segmentation, particularly over the long term.

4. The relative importance of product features and attributes are key determinants of consumer preference and choices. The importance of product features and attributes will vary both across and within individual consumers as a function of benefits sought, usage situation, and purpose. Thus, the importance of product features and attributes should be measured for each benefit and use occasion for which the product is appropriate (in the view of the consumer). These importance ratings should ultimately be mapped into consumer purposes.

5. The set of products that the consumer considers appropriate for providing particular benefits and for realizing specific purposes will vary as a function, of purpose, use occasion, and benefits sought. Thus, for each identifiable use occasion, a set of products the consumer views as substitutes should be identified. This set of products should include those that the consumer produces for himself or herself.

6. The consumer's evaluation (or expectations) of product performance will vary across purposes, use occasions, and benefits sought. Evaluations of the relative performance of products will also vary by purpose, use occasion, and benefit sought. Thus, performance evaluation cannot be measured in an absolute sense; it can only be obtained relative to the use of other products or services that the consumer considers appropriate for a given purpose, benefit, and use occasion.

7. A frequent assumption of many choice modeling exercises is that relative ratings of product performance may be combined with consumer's importance weighting to describe and ultimately predict choice. There is a general recognition in the literature that this type of exercise must be done at the individual level (Shocker, Stewart, and Zahorik 1990a&b) because importance weights may differ from individual to individual. On the other hand, most choice modeling exercises make an important simplifying assumption. This assumption is that all consumers use the same decision rule, namely, a compensatory rule. Research on consumer decision making is quite clear, however, that single-stage compensatory rules are used by consumers less frequently than other rules.

Consumers often use multi-stage decision strategies, screening out many alternatives on the basis of one or a few attributes. Simple heuristics, brand based or not, are among the more common decision rules used by consumers. For example, research has shown the lexicographic rule to be one of the more common rules employed by consumers (Reilly and Holman 1977). Screening strategies and heuristics used by consumers tend to be highly idiosyncratic and are driven by both the consumer's purpose and situational constraints such as budgets, time, and product availability. Individual difference characteristics, factors such as expertise, need for cognition, and involvement, have figured prominently in research on consumer decision making. An important implication of consumers' use of idiosyncratic decision rules for market segmentation is that modeling of choice behavior must be at the level of the individual and must provide a means for capturing idiosyncratic choice rules used by consumers. This is a far more complex and formidable task than modeling choice based on a uniform decision rule. Nevertheless, it is an approach that is consistent with the purposeful self-selection of consumers-consumers select alternatives for comparison, they select attributes on which to base their decisions, and they select decision rules which suit their purposes, their circumstances, and their individual characteristics.

Identification of individual choice rules is the last stage in the segmentation process. Note that choice rules are specific to individuals and are context bound, that is, the rules apply only for choices among a specific set of alternatives that are evaluated within a specific usage or purchase situation in order to acquire specific benefits for a specific purpose. The results of this operation are occasion specific market segments consisting of a single consumer. In fact, each consumer may represent multiple statements, one segment for each occasion. This represents the most basic unit for segmentation research.

Obviously, for most products and services, individual by use occasion segments are impractical as units to which marketing actions can be targeted. Some type of aggregation will be necessary to provide a smaller number of segments for which marketing plans can be formulated and implemented. The issue is not whether aggregation should occur, but what the building block for such aggregation should be. The foregoing discussion suggests that the building block for segmentation should be the individual within use of occasion.

SOME ADVANTAGES OF INDIVIDUAL BY OCCASION BUILDING BLOCKS

For many products and services, those with single and simple uses, the individual by occasion building block will reduce to the simple, individual level approach to segmentation that has been typical of past research. Aggregation may be quite straight-forward in these circumstances. More complex products and services, that offer multiple benefits and have the potential to serve multiple purposes, will offer more interesting opportunities for complex segmentation typologies based on individual by occasion building blocks. Such segmentation offers a number of potential advantages for the marketer and the consumer:

1. The approach provides a basis for offering individually customized products and services where it is economically justifiable (and this is increasingly the case in numerous product categories);

2. It directly maps product attributes, benefits, and use occasions into one another and links them to purposive consumer behavior:

3. It explicitly recognizes that consumers may use information in different ways. Consumers selectively attend to information and frequently adopt idiosyncratic decision rules. Knowledge of this selectivity and idiosyncrasy provides a better understanding of consumer behavior and a foundation for more effective marketing programs;

4. It shifts the focus of segmentation away from marketing actions and stimuli, and to the consumer and his or her purposes. As such, the approach is more consistent with the concept of consumer sovereignty. It also explicitly recognizes that consumers segment themselves through a process of self-selection. Marketers and consumer researchers merely discover these segments.

The question may arise as to whether the approach to segmentation outlined above can be implemented in any practical and economical fashion. The experience of a number of firms suggests the answer is yes despite the greater demands for data collection and analysis the procedure imposes.

CONCLUSION

The increasing number of options available to consumers has resulted in an exponential increase in the variance of consumer behavior. This increased variance is at least partially the result of consumers' exercising greater self-selection. This self-selection, in turn, has moved the locus of segmentation from the marketer to the consumer. Discovery of these consumer defined segments requires a more complex approach to segmentation that recognizes the purposefulness of consumer behavior and uses individual by use occasion as the basic building block for segmentation.

REFERENCES

Bass, F. (1974), 'The Theory of Stochastic Preference and Brand Switching," Journal of Marketing Research, 11 (February), 1-20.

Belk, R. (1987), "The Role of the Odyssey in Consumer Behavior and in Consumer Research," Advances in Consumer Research, Vol. 14, (Provo, Utah: Association for Consumer Research), pp. 357-361.

Cotton, John L. (1985), "Cognitive Dissonance in Selective Exposure," in D. Zillman and J. Bryant (Eds.), Selective Exposure to Communication, (Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.), 11-34.

Davis, S. M. (1987), Future Perfect, (New York: Addison Wesley).

Dickson, P. P. (1982), "Person-Situation: Segmentation's Missing Link," Journal of Marketing, 46 (Fall), 56-64.

Dickson, P. P. and J. L. Ginta (1987), "Market Segmentation, Product Differentiation, and Marketing Strategy," Journal of Marketing, 51 (April), 1 -10.

Frank, R. E., W. F. Massy, and Y. Wind (1972), Market Segmentation, (Englewood Cliffs, N.J.: Prentice-Hall).

Gutman, J. (1982), "A Means-End Chain Model Based on Customer Categorization Processes," Journal of Marketing, 46 (Spring), 60-72.

Haley, R.I. (1968), "Benefit Segmentation: A Decision-Oriented Research Tool," Journal of Marketing, 32 (July), 30-35.

Y. S. Lincoln and E. G. Guba (1985), Naturalistic Inquiry, (Beverly Hills, Calif.: Sage).

F. P. McDonald (1985), "Whither the New Segmentation Systems," Marketing and Media Decisions, 20 (6), 94, 96.

Olshavsky, R. W. and D. H. Granbois (1979), "Consumer Decision Making - Fact or Fiction," Journal of Consumer Research? 6 (September), 98-100.

Pechmann, C. and D.W. Stewart (1990),-'The Role of Comparative Advertising: Documenting Its Effects on Attention Recall, and Purchase Intentions," Journal of Consumer Research, 17 (September), }80-191.

Powers, W. T. (1973), "Feedback: Beyond Behaviorism," Science, 179, (Jan. 26), 351-356.

Powers, W. T. (1978), "Quantitative Analysis of Purposive Systems: Some Spadework at the Foundations of Scientific Psychology," Psychological Review, 85, 417-435.

Punj, G. N. and D. W. Stewart (1983), "An Interaction Framework of Consumer Decision Processes," Journal of Consumer Research, 10 (September), 181 -196.

Reilly, M. and R. Holman (1977), "Does Task Complexity or Cue Intercorrelation Affect Choice of an Information Processing Strategy: An Empirical Investigation," in W. D. Perreault, Jr. (Ed.), Advances in Consumer Research, Vol. 4 (Chicago: Association for Consumer Research), 185-190.

Reynolds, T. J. and J. Gutman (1988), "Laddering Theory, Method, Analysis, and Interpretation," Journal of Advertising Research, 28 (February/March), 10-31.

Shocker, A. D., D. W. Stewart, and A. J. Zahorik (199Oa), "Mapping Customer Perceptions of Markets," Journal of Management Issues, 2 (Summer), 127-159.

Shocker, A. D., D. W. Stewart, and A. J. Zahorik (1990b), "Mapping Competitive Relationships: Practice, Problems, and Promises," in G. Day, B. Weitz, and R. Wensley (Eds.), The Interfaces of Marketing and Strategy, (Greenwich, CT: JAI Press).

Skinner, B. F. (1938), The Behavior of Organisms, (New York: Appleton-Century-Crofts).

Smith, W.R. (1956), "Product Differentiation and Market Segmentation as Alternative Marketing Strategies," Journal of Marketing, 21 (July), 3-8.

Tolley, R.S. (1991), 'The Search: Patterns of Newspaper Readership," in E. Clark, T. Brock, and D.W. Stewart (Eds.), Advertising and Consumer Psychology, (Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.).

Wilkie, W. L. and J. B. Cohen (1977), "An Overview of Market Segmentation: Behavioral Concepts and Research Approaches," Working Paper, (Cambridge, Mass.: Marketing Science Institute).

Wind, Yoram (1978), 'Issues and Advances in Segmentation Research," Journal of Marketing Research, 15 (August), 317.

Zimmer, D. and J. Bryant (1985), Selective Exposure to Communication, (Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.).

----------------------------------------