Polygamous Loyalty and Varying Utility Function: an Exploration of Brand Switching in Frequently-Made Purchases

Citation:

Fred Selnes and Sangeeta Singh (2001) ,"Polygamous Loyalty and Varying Utility Function: an Exploration of Brand Switching in Frequently-Made Purchases", in AP - Asia Pacific Advances in Consumer Research Volume 4, eds. Paula M. Tidwell and Thomas E. Muller, Provo, UT : Association for Consumer Research, Pages: 162-166.

Asia Pacific Advances in Consumer Research Volume 4, 2001      Pages 162-166

POLYGAMOUS LOYALTY AND VARYING UTILITY FUNCTION: AN EXPLORATION OF BRAND SWITCHING IN FREQUENTLY-MADE PURCHASES

Fred Selnes, Norwegian School of Management, Sandvika, Norway

Sangeeta Singh, Norwegian School of Management, Sandvika, Norway

INTRODUCTION

Customer loyalty has been the focus of much discussion over the past decades and managers now commonly accept that the key to a firm’s success, and consequently profitability, is more dependent on the retention of existing profitable customers rather than trying to woo new ones (Reichheld 1996; Dowling and Uncles 1997; O’Brien and Jones 1995). For obvious reasons companies prefer that customers are loyal only to them, and thus not have relationships with their competitors. However, most companies find that many of their customers have relationships also with their competitors, and thus do not have monogamous relationships. Not only do companies find that customers have more than one relationship, but that their customers have strong and lasting relationships with multiple suppliers. One question that arises is why customers have relationships with more than one supplier within a category? And, if customers have polyamous relationships, what factors determine the choice within the set of alternative relationships?

Traditionally, brand loyalty has been defined as the nonrandom selection of one or more brands from a set of such brands as a result of evaluative decision making processes (Jacoby and Kyner 1973). However, others (e.g. Winter and Rossiter 1989) have found that consumers in high frequency categories, do not confine their choice to one or two brands, but instead distribute it over several brands.

This research attempts to demonstrate that in the case of frequently purchased, low involvement purchase situations, consumers are likely to display preferences for more than one alternative (brand or supplier) and that the choice on each occasion is going to alternate between these alternatives. In addition, relative satisfaction with the different alternatives and the variability in the utility function across purchase situations are used to explore the composition of the choice set and explain how choices are being made. We first develop a theoretical rationale for a set of hypothesis which are tested based on a survey of consumers’ behavior in grocery shopping. Our findings are discussed in terms of managerial implications and suggestions for future research.

THEORETICAL BACKGROUND

When consumers are faced with a decision involving several alternatives, the decision process is likely to have two stages. The first stage is eliminating unacceptable alternatives, and the second stage is involving a more detailed evaluation of the reduced set (Lussier and Olshavsky 1979). In high involvement or less-frequent purchases, the choice decisions in the final stage are thought to be elaborate and compensatory in nature (Bronnenberg and Vanhonacker 1996). Although two-stage choice models have been extensively studied in brand choice decisions (e.g. MacAlister 1982; Nedungadi 1990), little attention has, to the best of our knowledge, been devoted to understanding choice behavior when choices are made repetitively. The evidence seems to suggest that consumers do engage in only limited information acquisition in repetitive and low involvement purchases (Hoyer 1984; Lehman and Moore 1980; Swan 1969). Therefore, in repetitive choices it is reasonable to assume that consumers have the information they need to make their decision, and thus we may assume that they make informed decisions. It is further reasonable to assume that consumers store evaluations (e.g. satisfaction) of the alternatives in their choice-set, and that these evaluations contain information about each alternative’s value (benefit and cost). The selection of an alternative from the stored set of alternatives is then made by maximizing value after a cost-benefit analysis (Roberts and Lattin, 1991).

Stigler (1961) demonstrated that for a rational consumer, the expected utility of search for additional alternatives decreases and once the 'best’ brand is identified, the search is completed. This brand that best satisfies the consumer’s needs then becomes the only considered brand (Hauser and Wernerfelt, 1990). When the consumers have a utility function (based on costs and benefits associated) that remains constant across purchases, there is no reason for the consumer to have more than one alternative unless the utility is derived by altering among the alternatives (e.g. variety seeking). Therefore, in grocery shopping, the consumer is likely to have one store as a base-line store which is best able to fulfill the consumer’s need in terms of providing highest utility according to the individual’s utility function. The second store is entered into the choice set when the baseline store can not provide sufficient value across different choice situations, and the third is entered when the two primary stores can not provide sufficient value, and so forth. Thus, we assume that the cosumer prefers to have fewer alternatives, all other things being equal. It follows that the higher value the consumer experiences with his or her baseline store, the less likely it is that other alternatives will be added. Other alternatives are added as a search mechanism because some products can not be evaluated without consumption (Nelson 1970). We suggest that overall satisfaction with a store is a good indicator of the value produced. This is in line with Oliver’s (1996, p.13) definition of satisfaction as a fulfillment response. Thus, the higher the satisfaction with the base line store, the fewer alternative stores the consumers will have in their choice set.

H1: Satisfaction with the base line alternative will have a negative effect on size of choice set.

Even if the base-line alternative or store is fulfilling the consumer’s need and thus satisfying the consumer, the consumer may alternate his or her buying objectives across purchases in a way that precludes the base-line store as a feasible alternative. When the utility function varies across decisions, it is reasonable to expect that consumers may employ more than one supplier or brand. In grocery shopping, some situations will have high time-constraints and thus stores with high convenience will have high utility (all other things being equal). In other situations where the consumer for example is seeking novelty or excitement, stores with wide or exotic assortments will have high utility. This is in keeping with Meyer’s (1979) suggestion that different brands have different relative utility on different purchase occasions. Thus, when the consumer is alternating his or her utility function across decisions, we expect the consumer to form relationships with more than one supplier or brand.

H2: Alternating utility-functions across repetitive decisions will increase the size of the choice set.

Based on the discussion above we assume that consumers have fairly stable choice-sets in high-frequent decision categories like for example grocery shopping. Given that the consumer has a choice set larger than one, we next ask what factors determine their choice behavior within the "fixed" set. We first simplify the discussion by assuming that the consumer has only two alternatives, and that these alternatives provide equivalent utility and that the utility function is stable across decisions. It is then reasonable to assume that each store should receive about equal share of the consumers total purchases, that is 50% each. Now, if one of the alternatives can provide a higher utility, what share can we then expect for this alternative? The assumption usually made in choice-models (e.g. McFadden 1974; Bucklin, Rusell and Srinivasan 1998) is that the choice is a probability function based on expected utility of the focal alternative divided by the sum of expected utilities of all alternatives. An equivalent model is to express the alternatives as deviations in utility from a base-line alternative. Thus, if two alternatives have equal utility, this relative measure will be zero. Thus, we will employ relative satisfaction as an expression of the difference in utility derived from the base-line alternative and any of the other alternatives.

H3: Relative satisfaction within a set of stable-choice alternatives has a positive effect on probability of choice.

When consumers vary their utility function across purchase decisions, the probability of choosing a given alternative will also change. We assume that the consumer will then have to make a tradeoff between marginal utility (elative satisfaction) and buying objective. In our example with grocery shopping, a consumer may have a situation with a strong desire for seafood but his favorite store does not have a good variety of seafood. Now, the consumer must balance out his desire for seafood and his preference for his favorite store, and thus his relative satisfaction with the favorite store will have a less impact on his decision. Thus, when consumers vary their utility-function across purchases, the positive effect of relative satisfaction, is expected to be less. At an extreme, the choice objectives may vary so much that the decision really is a choice between categories. Thus, in such extreme situations, relative satisfaction within the choice-set (of two alternatives) will have no effect on share of purchases.

H4: The positive effect of relative satisfaction on probability of choice is moderated by the heterogeneity of utility functions across purchase situations.

Because of the general dynamics in the market and the consumers’ utility functions, the composition of the choice set over time will vary. Some alternatives will be entered into the basket of alternatives, whereas others will be removed. The propensity to leave a relationship has received considerable attention in the literature, and lack of satisfaction has been identified as a major explanation (e.g. Hirschman 1970; Kelly and Thibout 1978). Thus, when the consumer is dissatisfied with one of the stores in his or her choice set, this store is likely to be excluded from the choice-set. A likely scenario is that consumers gradually decrease their relationship with a store they are not satisfied with.

H5: Dissatisfaction with an alternative is increasing the motivation to exit the relationship.

METHOD

Ten grocery stores from two national chains (focal stores) were chosen for the study. The subjects were randomly selected from within a distance of about 2 miles around each store, and when approached they were asked if they counted themselves as customer of the focal store. Only those who considered the focal store as a store they regularly shopped in (at least once a month) were included in the study if they volunteered to participate (n=902). The variables used for the study are reported in Table 1.

Relative satisfaction was assessed by subtracting satisfaction with the most frequently employed alternative from satisfaction with the focal alternative. Of the total sample, 109 respondents (12.1%) had only one alternative in their choice-set. These were excluded from further analysis. Descriptive statistics from the 793 remaining subjects are reported in Table 2.

Satisfaction (SATIS) is negatively skewed (biased towards higher values), which is the normal observation for satisfaction measures. Size of choice-set (SIZE) has an average of 2.68 alternatives, where 43.8% of the subjects have 2 alternatives, 46.5% have three alternatives and thus only 9.7% have four or more alternatives. Satisfaction with the most preferred alternative (SATALT1) is higher than the satisfaction with the focal alternative, and thus relative satisfaction (RELSAT) has a negative mean. A large fraction of the subjects, 36.7%, say they vary their utility function across decisions (HETER). We observe that most subjects (87.6%) in our sample say they will continue to select the focal store about equally frequently, whereas a small fraction of 3.4% say they will most likely choose this store less in the future (EXIT).

RESULTS

Satisfaction with the focal store (STIS) and heterogeneity of the choice situations (HETER) were entered as independent variables, and size of choice-set (SIZE) was entered as dependent variable in an ordinary least-square multiple regression model. The analysis shows that satisfaction has a negative and significant effect on size of choice-set (beta=-0.115; p=0.001), and thus H1 is supported. Heterogeneity of utility functions across decisions did not have a significant effect, although the coefficient was in the expected positive direction (beta=0.08; p=0.122), and thus H2 was not supported. The size of choice-sets is thus not influenced by heterogeneity of decision situations.

Relative satisfaction (RELSAT) was expected to have a positive effect on share of the focal store (H3). This effect is expected to be moderated by heterogeniety of the utility functions across decisions (H4). We computed an interaction variable of relative satisfaction and heterogeneity (RELSAT*HETER). Number of alternatives in the choice set (SIZE) and level of satisfaction with the focal store (SATIS) were entered as a covariates. The dependent variable was share of total shopping allocated to the focal store. The variables were entered into an ordinary least-square model, and the results are given in Table 3.

Relative satisfaction has a positive and significant effect on share of total purchases (b=2.17; p=0.000). The effect of relative satisfaction is moderated by heterogeneity of utility functions as the interaction effect is significant, and negative as expected (b=-1.67; p=0.028). Thus both H3 and H4 are supported. Both covariates are significant and in the expected direction.

The propensity to reduce shopping at a store was expected to be driven by dissatisfaction. Analysis of variance (ANOVA) was used to analyze the level of satisfaction across the three categories, that is the propensity to increase share, the propensity to remain constant, and the propensity to reduce share (EXIT). The means and standard deviations in the three categories are reported in Table 4. The overall model assumed a linear (negative) relationship across the three categories, and is not significant (F=1.956;p=0.142). The difference in satisfaction level between those consumers that will increase or remain constant, was not significant. However, the difference between the propensity to exit and each of the other categories are negative and significant. The difference between remain constant (category 2) and reduce (category 3) has a t-value of 2.00 (p=0.055), and the difference between increase (category 1) and reduce (category 3) has a t-value1.69 (p=0.095). Thus, the motivation to reduce and gradually exit from a relationship is driven by dissatisfaction and H5 is supported.

TABLE 1

VARIABLES INCLUDED IN THE STUDY

TABLE 2

DESCRIPTIVE STATISTICS

TABLE 3

ESTIMATED REGRESSION MODEL OF SHARE (N=716)

TABLE 4

LEVELS OF SATISFACTION ACROSS FUTURE MOTIVATION

DISCUSSION

Research on choice sets so far has focused on selection of a brand from the choice set as a result of maximizing utility which is implicitly assumed to remain constant. While this may be true for less frequently purchased items, it is not the case for frequently- made repeat purchases. In fact, we demonstrate that for frequently-made repeat purchases, the utility function varies across situations and thus provides an explanation for choice being distributed over several alternatives rather than being restricted to one.

Establishing that choice is distributed over several alternatives we show that consumers are using one alternative as their 'preferred choice’ and the share of this alternative of the consumers’ total purchases is moderated by the consumers’ relative satisfaction with this alternative and the variability in their utility function. In addition, the level of dissatisfaction with an alternative in the basket of suppliers determines the exclusion of that supplier from the choice set. Although, these are exploratory findings, they are nevertheless significant in explaining some of the dynamics within a choice set- an area not explored in-depth so far.

These findings also have a number of managerial implications. If retailers were to use the popular definition of loyalty that 'you shall not be served by others than me’, they may erroneously conclude that they do not have loyal customers when in fact they do. Following our way of reasoning, retailers would need to focus on getting their store as part of the customers’ choice set and when inside, increase the share of the purchases.

Store managers need to define loyalty as a relative share of purchases made from a basket of alternatives rather than assuming that a loyal customer is one who makes 100% of the purchases from them. Managers should then focus on variables that increase their share within the basket. One strategy is to increase the utility derived from ones store, either by increasing the benefits or reducing the costs. Level of satisfaction with the focal store is a good measure of utility created. Another strategy could be to differentiate the focal store from competing alternatives in ways that increase the utility derived from the store. Such differentiation would alter the utility function, and the relative satisfaction with competing alternatives have less impact.

The present study has provided some insight into the dynamics of the factors that determine composition of a basket of alternatives, and some of the factors that explain choices within this basket. But these results are not without their shortcomings and give an indication of opportunities for future research. We have used relative satisfaction with the alternatives and heterogeneity of the choice situations to explain some of the dynamics within the choice set. We did not take into consideration the heterogeneity or homogeneity of the consideration set itself. Future research could give attention to how it affects the choice of the size set and the share of total purchases allocated to each alternative.

This study employed cross-sectional data and hence it was not possible to explain the time frame consumers use to enter or/and remove alternatives into or/and from the choice set. Is this movement sudden or gradual? A longitudinal study could better capture this information. In fact, other forms of data collection like experiments or observation of actual behavior might provide a finer understanding of what is taking place.

We did not take into account manipulations brought in by marketers. Future research could investigate how competition, with short term promotions as well as more fundamental changes, affect the dynamics within a choice set. Our study used retail shopping as the setting to establish the proposed hypotheses. Future studies could explore other product or service categories where consumers have frequently-made purchases, for example, business class airline customers. This would help strengthen the results from our study.

REFERENCES

Bronnenberg, Bart J. and Wilfred R. Vanhonacker (1996), 'Limited Choice Sets, Local Price Response, and Implied Measures of Price Competition,’ Journal of Marketing Research, Vol. 33 (May), 163-174.

Bucklin, Randolph E., Gary J. Russell and V. Srinivasan (1998), 'A Relationship Between Market Share Elasticities and Brand Switching Probabilities,’ Journal of Marketing Research, Vol. 35 (February), 99-113.

Dowling, Graham R. and Mark Uncles (1997), 'Do Customer Loyalty Programs Really Work,’ Sloan Management Review, Vol. 38(4), Summer, 71-82.

Hauser, John R. and Birger Wernerfelt (1990), 'An Evaluation Cost Model of Consideration Sets,’ Journal of Consumer Research, Vol.16 (March), 393-408.

Hirschman, Albert O. (1970), Exit, Voice, Loyalty Responses to Declines in Firms, Organizations and States, Cambridge, MA: Harvard University Press.

Hoyer, Wayne D. (1984), 'An Examination of Consumer Decision Making for a Common Repeat Purchase Product,’ Journal of Consumer Research, Vol. 11 (December), 822-829.

Jacoby, Jacob and David B. Kyner (1973), 'Brand Loyalty Versus Repeat Purchase Behavior,’ Journal of Marketing Research, Vol. 10 (February), 1-9.

Kelly, Harold H. and John H. Thibaut (1978), Interpersonal Relations, New York, NY: Wiley.

Lehman, Donald R. and William L. Moore (1980), 'Validity of IDB: An Assessment Using Longitudinal Data,’ Journal of Marketing Research, Vol. 17, 450-459.

Lussier, Dennis A. and Richard Olshavsky (1979), 'Task Complexity and Contingent Processing in Brand Choice,’ Journal of Consumer Research, Vol. 6 (September), 154-165.

McAllister, Leigh (1982), 'A Dynamic Attribute satiation Model of Variety-Seeking Behavior,’ Journal of Consumer Research, Vol. 9 (September), 141-150.

McFadden, Daniel 81974), 'Conditional Logit Analysis of Qualitative Choice Behavior,’ Frontiers in Econometrics, P. Zarembka, ed. New York, NY: Academic Press, 105-142.

Meyer, Robert (1979), 'Theory of Destination Choice Set Formation Under Informational Constraint,’ Transportation Research Record, 750, 6-12.

Nedungadi, Prakash (1990), 'Recall and Consumer Consideration Sets: Influencing Choice Without Altering brand Evaluations,’ Journal of Consumer Research, Vol. 17 (December), 263-276.

Nelson, Phillip (1970), 'Information and Consumer Behavior,’ Journal of Political Economy, 78 (March/April), 311-329.

Oliver, Richard L. (1996), Satisfaction: A Behavioral Perspective on the Consumer, New York, NY: McGraw Hill.

Reichheld, Fredrick F. (1996), 'Learning from Customer Defections,’ Harvard Business Review, March-April, 56-69.

O’Brien, Louise and Charles Jones (1995), 'Do Rewards Really Create Loyalty?,’ Harvard Business Review, Vol. 73, May-June, 75-82.

Roberts, John H. and James M. Lattin (1991), 'Development and Testing a Model of Consideration Set Composition,’ Journal of Marketing Research, Vol. 28 (November), 429-440.

Stigler, George J. (1961), 'The Economics of Information,’ Journal of Political Economy, 69 (June), 213-225.

Swan, J.E. (1969), 'Experimental Analysis of Predecision Information Seeking,’ Journal of Marketing Research, Vol. 6, 62-65.

Uncles, Mark (1994), 'Do You or Your Customers Need a Loyalty Scheme?’ Journal of Targeting, Measurement and Analysis for Marketing, Vol. 2(4), 335-350.

Winter, Franz L. and Rossiter, John R. (1989), 'Pattern-Matching Purchase Behavior and Stochastic Brand Choice: A Low Involvement Model,’ Journal of Economic Psychology; Vol. 10(4), 559- 585.

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