Modeling the Coupon Redemption Decision

ABSTRACT - This article develops a model of the consumer's decision to redeem a coupon on a purchase occasion. The redemption decision is a function of four types of variables - coupon characteristics, characteristics of the purchase, brand loyalty, and concurrent promotional conditions. The underlying hypothesis is that consumers attempt to balance their desire for economy with the minimization of shopping time and effort. As coupons involve both costs and benefits, they will be used only when the incremental perceived effort to redeem is relatively low ar.d perceived value is relatively high. A simplified version of this model is tested with scanner panel data for two product categories.

Citation:

Caroline M. Henderson (1985) ,"Modeling the Coupon Redemption Decision", in NA - Advances in Consumer Research Volume 12, eds. Elizabeth C. Hirschman and Moris B. Holbrook, Provo, UT : Association for Consumer Research, Pages: 138-143.

Advances in Consumer Research Volume 12, 1985      Pages 138-143

MODELING THE COUPON REDEMPTION DECISION

Caroline M. Henderson, Dartmouth College

ABSTRACT -

This article develops a model of the consumer's decision to redeem a coupon on a purchase occasion. The redemption decision is a function of four types of variables - coupon characteristics, characteristics of the purchase, brand loyalty, and concurrent promotional conditions. The underlying hypothesis is that consumers attempt to balance their desire for economy with the minimization of shopping time and effort. As coupons involve both costs and benefits, they will be used only when the incremental perceived effort to redeem is relatively low ar.d perceived value is relatively high. A simplified version of this model is tested with scanner panel data for two product categories.

INTRODUCTION

With the current annual distribution of coupons surpassing 120 billion (U.S. News and World Report 1984), managers recognize the need to understand consumer response to coupons. In the literature, consumer response has been dealt with in a variety of ways: prediction of redemption rates (Reibstein and Traver 1982, Ward and Davis 1978); the effect of couponing on incremental sales (Hee 1981), on purchase acceleration (Neslin, Henderson and Quelch 1984) and on brand switching (Dodson, Tybout and Sternthal 1978); coupon profitability (Neslin and Shoemaker 1983, Klein 1981, Irons, Little and Klein 1983); and the identification of coupon users (Teel, Williams and Bearden 1980). Focusing on the behavioral issues in consumer response, previous work has conceptualized why consumers use coupons (Schindler 1983, Beem and Shaffer 1981) and how coupons achieve results (Ward and Davis 1978a, Raju and Hastak 1979, Rothchild and Gaidis 1981, and Strang and Gardner 1983). In all of this work, there is little indication of how consumers actually integrate coupons into shopping patterns -the process by which the 1400 coupons received annually by each household are converted into 63 redemptions (average redemption rate is 4.52, Nielsen Clearing House 1982). Such basic data as the "where, when and how" of coupon use in shopping can be key in designing coupon programs that achieve their intended results.

THE COUPON REDEMPTION DILEMMA

Paradoxically, the last decade has seen two conflicting trends: the rise of convenience orientation, or "time-buying" (Berry 1979), versus an increase in motivation to economize (Progressive Grocer 1983). Interest in economy is seen in self-reported data (Opinion Research 1982), in the spread of coupon usage to a larger percentage of the population (Bowman 1980) and, until recently, in an increase in the average number of coupons redeemed (PMAA 1982). Yet a consumer who uses coupons should incur more total shopping time than one who does not. The major cost elements of coupon usage include the time and effort to clip coupons, to maintain a coupon collection, to use this collection in planning shopping and in in-store decisions, and to wait an additional few moments for check-out. Balancing these costs, coupons offer a price reduction.

To some extent there are separate segments involved in the time-buying and economizing trends. There is a slight tendency for nonworking consumers to become more involved with coupons (Strang 1981) and for workers to be more concerned with time-buying (Anderson 1972). However, these statistical correlations are very weak; many consumers must be balancing dual interests. Evidence for the wide appeal of coupons is seen in the shopping strategy typologies developed by Guiltinan and Monroe (1979) in which 80% of the subjects have positive attitudes toward couponing while differing on shopping interest, degree of pre-planning, interest in store specials, brand loyalty, and other factors.

Some observers of the coupon dilemma conclude that consumers have been entrapped and are trading their labor for negligible benefits-a dead-weight economic loss (Uhl 1982). Yet another interpretation is possible.

A plausible hypothesis is that consumers allocate their resources of time, energy and money (Downs 1961) through following logical strategies in which coupon use is made "convenient"-either through selecting redemption occasions which incur minimal time and energy costs or through selecting redemptions which offer abnormal savings to pay for these costs. Evidence for selectivity is that consumers may specialize in using coupons only in particular product categories (Henderson 1983).

Within a category, consumers may adopt further approaches to maximize coupon savings while minimizing time and effort. In this way, consumers who use coupons may actually spend no additional time shopping than those who do not. In contrast to shopping behavior research which measures total shopping time (for example, Arndt and Gronmo 1977), this research focuses on the coupon redemption occasion as the unit of analysis. The implicit time-money-effort trade-off of such occasions car be examined.

MODELING COUPON REDEMPTION

The literature on coupon redemption has focused on predicting aggregate redemption rates-an important calculation in budgeting and evaluating coupon programs-(Reibstein and Traver 1982, Ward and Davis 1978b). These studies have shown that redemption rates vary by brand, category, media used and other factors. The relevant factors in determining overall redemption rates are shown in Table 1, categorized into four groups: coupon characteristics, brand factors, purchase characteristics, and other promotional activities.

In modeling coupon redemption on a disaggregated basis, both the dependent and independent variables need to be adapted. The dependent variable-coupon redemption-actually includes a three stage decision process-collection, intention, and redemption. In the collection stage the consumer makes the decision to clip an available coupon and to save it for future use. The intention stage takes place before a store visit and involves either the explicit decision to use a particular coupon or the preparatory work (taking the coupon to the store) that will make the later redemption possible. The redemption stage is the submission of the coupon at check out. While the redemption decision is intertwined with brand choice, the two decisions can be modelled separately.

TABLE 1

INFLUENCES ON THE COUPON REDEMPTION DECISION

The factors that affect these three stages can be adapted from the aggregate case. Since half of all redeemed coupons are kept at least four weeks (Blair 1982, Nielsen Clearing House 1982), the intention and redemption stages must be strongly related to factors that occur beyond the time of the coupon drop itself. For this reason, Table 1 expands on the influences appropriate for the aggregate case and includes more characteristics of the purchase. These factors are expected to affect coupon redemption in the following ways.

At the collection stage, there are two primary influences: consumers report that face value and brand are important considerations in choosing coupons (Nielsen Clearing House 1977). Consumers should be more inclined to notice coupons for brands which they normally buy and more inclined to decide that such coupons are worth collecting. The characteristics of the coupon itself influence this process. The media used to distribute the coupon will affect the collection decision through spotlighting the coupon. The value offered and the quality of the execution of the coupon make the coupon more or less attractive to collect. Thus, the factors used in the aggregate models are most appropriate at the collection stage as they influence how many coupon collections the new coupon will join. A combination of effort and value effects are evident in these variables. High value coupons clearly offer superior economic incentives. Perceived value may also increase if the coupon is for a preferred brand. Consumers will incur low collection costs if coupons are made generally more available.

At the intention stage consumers will be affected by two additional types of influences -the characteristics of the purchase and the presence of other promotions. If consumers have the time and inclination to plan their shopping they may be more inclined to use coupons. There is some evidence that coupon use and pre-planning (use of a list, reading newspaper ads - Progressive Grocer 1975) are interconnected. Under hurried conditions, coupons will be less likely to be taken to the store. If the consumer has a greater need to purchase in a particular product category a long time has elapsed since the last purchase was made or a greater than average quantity is needed -then the product category may be more salient. Under this condition, the perceived value of the coupon may rise and redemption intentions may be more likelY.

The presence of other types of promotions can also affect this stage as they act to remind the consumer to use the coupon and to offer an ever. better deal or. a particular brand. For example, if a particular brand is featured in retail advertising, the advertising may remind the consumer to use a saved coupon. Alternatively, the consumer may be looking for multiple promotions to get the best overall deal. While it is doubtful if consumers car. actually perform the calculations necessary to carry through on such strategies (Capon and Kuhn 1982), the simultaneous promotions may trigger the perception of in.creased savings. In reminding the consumer to use the coupon, such ads also lower the consumers effort -or thinking costs -by presenting an acceptable and easy choice.

At the redemption stage, the consumer is in the store and makes the final decision to redeem a coupon for a particular brand. All four types of influence are assumed to be operative at this point. A consumer with more time, with greater need for the product, who has recently received or beer. reminded cf a coupon, and who sees that a brand is also being featured should be more likely redeem than if these conditions are not in effect.

TESTING THE MODEL

In this research, a simplified form of this model is tested using only the final stage -redemption- and a limited set of redemption predictors.

Data

Data were available from the SAMI "northeast scanner panel" and covered consumer purchases from April to October 1981. The scanner market includes 15 .stores (3 chair.s) and tracks panel purchases and coupon redemptions by store. Two product categories were available for analysis: a paper product ("paper") and a coffee product ("coffee"). A total of 2023 panelists made 18294 paper purchases and 1232 panelists made 4579 coffee purchases during this time period. A data base was constructed by merging coupon redemption records with panel purchases and with local newspaper advertising data. Each observation consisted of the record of one purchase occasion-brand choice, coupon redemption, price and promotion data, store and date. All consumers who had nu coupon redemptions ." the two product categories were discarded, leaving a total of 2739 coffee purchases and 8312 paper purchases made by consumers who were at least casual coupon users. Of these, 1551 paper purchases and 1322 coffee purchases involved a coupon redemption.

Model

As coupon redemption is a binary choice, it is appropriate to model redemption using logistic regression (Fienberg 1979):

EQUATION (1)

Where:

P_i probability that a coupon is redeemed on occasion i (y = 1)

1-P_i = probability that a coupon is not redeemed on occasion i (y = 0)

X_i = attribute of purchase occasion i

The observed dependent variable assumes values of 0 -no coupon redeemed-or 1--one or more coupons redeemed on the purchase. Using maximum likelihood estimation, the model finds parameters that yield predicted probabilities of coupon redemption for a particular set of independent variables. The FUNCAT procedure in the SAS statistical package (SAS Institute 1982) was used for estimation. The significance of the overall regression was tested with a chi-square test using the log likelihood ratio. The ability of the model to provide information beyond an equally likely null model was analyzed with U2 (Hauser 1978). Each coefficient was tested with a t-test and each category analyzed in split-halves to test the reliability of the model.

This model is cross-sectional as all purchases of potential coupon users are analyzed together. Based or the widespread nature of coupon use, the model makes the assumption that coupon. redemption is a characteristic of purchase situations rather than solely a characteristic of individual consumers. However, this homogeneity assumption is tested and variations in redemption behavior are analyzed for specific consumer segments. The model also assumes that coupon redemption is a zero-order process. Any time series effects of sequential redemptions by individual consumers are not modelled. Each consumer's history is known to some extent and is used to calculate a brand loyalty variable, the time since last purchase and the average quantity purchased. The model also assumes that coupon distribution has reached the point where coupons are always available to consumers who wish to use them.

TABLE 2

EMPIRICAL TEST OF THE COUPON REDEMPTION DECISION

Variables

The variables used in the test (Table 2) are defined as follows. Elapsed time measures the time difference time in days since the last purchase of the product category, standardized for each consumer. Quantity is the standardized difference from the average purchase quantity. While these variables are used in this analysis as surrogate measures for the perceived value of using coupons in a purchase situation, the data set does not control causality. Quantity, in particular, may be a result of the redemption. rather than its cause.

The store of purchase is operationalized by two dummy variables -store 1 and store 2--for two of the three grocery chairs in. the area. Both of these variables are hypothesized to be positive predictors of redemption. Store 1 uses a warehouse format and store 2 is the leading chain with an active promotional posture. By contrast, the excluded store is smaller and less promotionally active. A consumer selecting the excluded chain may be less concerned with value ar.d less likely to be on a major shopping trip. The time of purchase is operationalized with two dummy variables representing purchases made at the beginning of the week (Mondays and Tuesdays) and end of the week (Saturday). The beginning of the week measure is used to capture the effect of Sunday FSI couponing and best food day advertising (Tuesdays in this market). Consumers should be more inclined to use coupons while promotional messages are still fresh in their minds and before they incur costs of maintaining a coupon. collection. Alternatively, the end of the week variable is used to capture the effect of increased free time.

Brand loyalty is measured by proportion of purchase. There are four measures of promotion in the data base. Manufacturer advertising is a newspaper ad, containing a coupon, which has beer. placed by the manufacturer. This variable acts as both a distribution device for coupons and as a reminder. Retail ad is a newspaper ad -often featuring a reduced price - which is placed by the store. Both of these variables are coded for the week of purchase. Price change refers to any change in price - either regular shelf price or a promotional reduction -on the purchased brand. The variable deal pack indicates that the brand used some sort of special packaging -cents-off, extra product or a reusable container. There were no deal packs used in the paper category during this time period.

There are two measures of coupon characteristics in these data. Both take account of competitive couponing conditions. Coupon value is to the difference between the average coupon redeemed on the brand purchased and the overall average coupon in the product category for the week. The availability of high value coupons in the marketplace should stimulate redemption. Coupon share is a measure of the relative availability of the brand's coupons during the week of purchase. This variable is a surrogate measure of true availability as it is based on redemption rather than distribution share.

RESULTS

The model results for each product category are shown in Tables 3 and 4. The predictors for coupon redemption are very similar in the two categories. In both cases purchase quantity, the store dummies, brand loyalty, manufacturer ad and the coupon characteristics are correlated with redemption. However, there is some instability in the results. Although two-thirds of the coefficients are within one standard error between split-half samples, the effects do not have a uniform significance level between halves. Overall, the model for both product categories is statistically significant in predicting an individual redemption. However, the model for coffee provides only a small (.06) amount of information, measured by U , compared to a null model. The model for paper provides a substantial amount of information (.44).

The results for elapsed time, while statistically insignificant, are reversed in sign. As this test is not a true experiment it is difficult to determine the causal direction in these data. Coupons may stimulate consumers to buy sooner than expected or longer elapsed time may lead consumers to use coupons. The quantity variable may be associated with redemption through purchase requirements.

The store dummy variables validate the hypothesis as purchases that occur in certain stores are more likely to involve coupons. This result needs confirmation, however, as those who redeem coupons more frequently may tend to shop in certain stores. This hypothesis is addressed by the segment analysis. The day of week dummies both have the correct sign but are not significant. Coupon usage appears to be spread evenly throughout the week. Day of the week may not fully capture the nature of the shopping trip and the degree of difficulty involved in redeeming a coupon.

TABLE 3

MODEL RESULTS BY SPLIT HALVES - PAPER

The brand loyalty variable is uniformly significant in these data but has the reverse sign. A brand less frequently purchased is a good predictor of coupon redemption. This result may be explained by the nature of the dependent variable in this test. Consumers may base their collection and redemption interior decisions on brand preference while supplementing their collections with coupons for less preferred brands. As these brands are primarily purchased for promotional reasons alone, coupons may be used on a higher percentage of these purchases.

Manufacturer ads appear to be a predictor of redemption; at .east some consumers redeem quickly following the newspaper appearance of a coupon-containing ad. Retail ad seems to be generally negative, indicating a tendency to redeem coupons on brands that are not also being featured. Coupons may be redeemed when prices are also reduced and, for coffee, when a deal pack is being used, however these effects are rot statistically significant.

The coupon characteristic variables are generally as hypothesized: consumers redeem when coupons are relatively more available and when they have a higher than average value.

TABLE 4

MODEL RESULTS BY SPLIT HALVES - COFFEE

Consumer Segments

The coupon redemption model was run on selected deal proneness segments developed in an earlier part of this research (Henderson 1983). Two segments are shown in Table 5 for coffee: coupon users who redeem on ar. average of 65% of their coffee purchases and r.on-responsive consumers who redeem or 6%. As these segments differ on a variety of demographic and purchase characteristics, they may be considered fairly homogenous groups with respect to shopping behavior and couPon usage.

The results show that store of purchase is still a factor in determining redemptions. Interestingly, the light and heavy coupon users also show a different orientation toward coupon value. Heavier coupon users appear to be more selective in redeeming higher value coupons.

TABLE 5

REDEMPTION BY CONSUMER SEGMENT

CONCLUSIONS

Coupon redemption can be analyzed on the level of the individual purchase and involves a three stage process. This view can be valuable in designing coupon programs which have an impact on collection, intention and redemption through careful selection of distribution media, face values, and use of newspaper advertising or in-store promotion. While the empirical analysis has shown that the redemption stage is correlated with several variables which are under the manager's control, the degree to which redemption occasions are determined by time ar.d effort minimization needs further analysis. This exploratory analysis needs to be extended with additional variables and with breaking the decision down into its three component parts. Future work might also contrast this approach with shopping strategy models derived from alternative views: for example, that consumers integrate coupons indiscriminantly into their shopping patterns to derive satisfaction from being a "smart shopper" (Schindler 1983) or as part of 3 scripted response (Gardner and Strang 1983).

REFERENCES

Anderson, B. (1972), "Working Women vs. Non-Working Women: A Comparison of Shopping Behavior," in. AMA 1972 Combined Proceedings, Becker and Becker, eds., American Marketing Association, 355- 359.

Arndt, J. and S. Gronmo (1977), "The Time Dimension of Shopping Behavior: Some Empirical Findings," in Advances in Consumer Research, Vol. IV, W. D. Perrault, Jr., ed., Association for Consumer Research, 230-235.

Beem, E. R. and H. J. Shaffer ( 1981), "Triggers to Customer Action - Some Elements in a Theory of Promotional Inducement," Cambridge: Marketing Science Institute.

Berry, L. L. (1979), "The Time-Buying Consumer," Journal of Retailing, 55 4 (Winter), 58-69.

Blair, K. C. (1982), "Coupon Design, Delivery Vehicle, Target Market Affect Conversion. Rate: Research," Marketing News, 1 (May 28).

Bowman, R. D. (1980), Couponing and Rebates: 2: Profit on the Dotted Line, New York: Lebhar-Friedman Books.

Capon, N. and D. Kuhn ( 1982), "Can Consumers Calculate Best Buys?", Journal of Consumer Research, 8 (March), 449-452.

Dodson, J. A., A. M. Tybout, and a. Sternthal ( 1978), "Impact of Deals and Deal Retraction or. Brand Switching," Journal of Marketing Research, XV (February), 72-81.

Downs, A. (1961), "A Theory of Consumer Efficiency," Journal of Retailing, 37 1 (Spring), 6-12.

Fienberg, S. E. ( 1979), The Analysis of Cross-Classified Categorical Data, Cambridge: The MIT Press.

Gardner, M. P. and R. A. Strang ( 1983), "Consumer Response to Promotion: Some New Perspectives," in Advances in. Consumer Research, Vol XI, T. C. Kinnear, ed.. Association for Consumer Research, 420-425.

Guiltinan, J. P. and K. B. Monroe ( 1979), "Identifying and Analyzing Consumer Shopping Strategies," in Advances in Consumer Research, Vol. VII, J. C. Olson, ed., Association for Consumer Research, 745-748 .

Hauser, J. R. (1981), "Testing the Accuracy, Usefulness, and Significance of Probabilistic Choice Models: An Information Theoretic Approach," Operations Research, 25 (May-June), 406-421.

Hee, J. R. (1981), Determining Manufacturers' Couponing Strategies, Unpublished M. S. thesis, Massachusetts Institute of Technology.

Henderson, C. M. (1983), The Effect of Price Promotion on Coupon Redemption and Brand Choice, Unpublished doctoral dissertation, Harvard University.

Irons, K. W., J.D.C. Little, and R. L. Klein (1983), "Determinants of Coupon. Effectiveness," in Advances and Practices of Marketing Science 1983, Proceedings of the 1983 ORSA/TIMS Marketing Science Conference, F. S. Zufryder, ed., University of Southern California, 157-164.

Klein, R. L. (1981), "Using Supermarket Scanner Results to Measure the Effectiveness of Coupon Promotions," Proceedings: Third ORSA/TIMS Market Measurement Conference, New York University.

Neslin, S. A., C. M. Henderson, and J. A. Quelch (1984), "Consumer Promotion and the Acceleration of Product Purchases," working paper 8124, Amos Tuck School.

Neslin, S. A. and R. W. Shoemaker ( 1983), "A Model for Evaluating the Profitability of Coupon Promotions," Marketing Science, 2 (Fall), 361-388.

Nielsen Clearing House (1980), Coupons: The Consumer Speaks out, A. C. Nielsen Company.

Nielsen, Clearing House (1982), Coupon. Distribution Trends and Patterns, A. C. Nielsen Company.

Opinion Research ( 1982), Marketing Index Study on Sales Promotions.

P.M.A.A. (Promotion Marketing Association of America) (1982), Questions and Answers about Couponing.

Progressive Grocer (1983), "Cautious Consumers Cling to Cost Cutting Tactics," May, 77-80.

Progressive Grocer (1975), "Consumer Behavior in. the Supermarket," October, 37-46.

Raju, P. S. and M. Hastak ( 1979) , "Consumer Response to Deals: A Discussion of Theoretical Perspectives," in Advances in Consumer Research, Vol. VII, J. C. Olsen, ed., Association for Consumer Research, 296- 301 .

Reibstein, D. J. and P. A. Traver ( 1982), "Factors Affecting Coupon Redemption Rates," Journal of Marketing, 46 4 (Fall), 102-113.

Rothchild, M. L. ar.d W. C. Gaidis (1981), "Behavioral Learning: It's Relevance to Marketing and Promotion," Journal of Marketing, 45 2 (Spring), 70-78.

SAS Institute, Inc. (1982), SAS User's Guide, Cary, North Carolina.

Schindler, R. M. ( 1983), "How do Cents-Off Coupons Affect Consumer Choice," working paper, Northeastern University.

Strang, R. ( 1981) , How Coupons Influence Consumer Purchases, Nielsen Clearing House.

Teel, J. E., R. H. Williams, and W. O. Bearden ( 1980), "Correlates of Consumer Susceptibility to Coupons in New Grocery Product Introductions," Journal of Advertising, 9 3, 31-46.

Uhl, J. N. ( 1982), "Cents-Off Coupons: Boon or Boondoggel for Consumers?", Journal of Consumer Affairs, 16 1 (Summer), 161-165.

U.S. News and World Report (1983), "Coupon Mania-Impact on Buyers, Sellers", September 19, 44.

Ward, R. W. and J. E. Davis (1978a), "A Pooled Cross-Section Time Series Model of Coupon Promotions, " American Journal of Agricultural Economics, (August), 393-401.

Ward, R. W. (1978b), "Coupon Redemption," Journal of Advertising Research, 18 4 (August), 51-58.

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