Special Session Summary Personalization and Customization: Implications For Consumer Decision Making and Behavior

John Godek, University of Michigan
[ to cite ]:
John Godek (2002) ,"Special Session Summary Personalization and Customization: Implications For Consumer Decision Making and Behavior", in NA - Advances in Consumer Research Volume 29, eds. Susan M. Broniarczyk and Kent Nakamoto, Valdosta, GA : Association for Consumer Research, Pages: 155-157.

Advances in Consumer Research Volume 29, 2002     Pages 155-157



John Godek, University of Michigan

Burger King claims, "When you have it your way, it just tastes better." Gateway suggests that consumers "Give us a call and we’ll build one that’s right for you." From fast food to fast computers, firms are seeking new ways to meet the needs of individual consumers rather than focus on larger, homogeneous market segments (Sheth, Sisodia, and Sharma 2000). In the past, companies sought to appeal to a broad range of consumer tastes by offering a wide variety of items, leading to the creation of "category-killers" such as Toys 'R’ Us, Best Buy, and Circuit City (Huffman and Kahn 1998). Recent advances in communications and manufacturing technology, however, have allowed firms to tailor their product offerings to individual consumers using both customization, where consumers have the ability to alter or even create products that contain precisely those attributes that the individual consumer specifies, and personalization, where firms directly elicit or indirectly infer consumer preferences and then identify which of the firm’s existing products provides the best fit for these preferences. Despite the significance of this topic to both marketing managers and marketing scholars, few forays into this area of research have examined the effect of such strategies on consumers and their decision processes.

The papers in this special session examined several aspects of customization and personalization and identified the implications for consumer decision making and behavior. The first paper demonstrated how personalizing product offerings by ordering options according to consumers’ preferences can influence perceptions of the retailer as well as the merchandise presented. The second paper showed that although consumers in general prefer not to be constrained in their customization decisions, certain types of constraints actually increase satisfaction. The final paper in the session identified conceptual differences between customization and personalization in terms of the amount of perceived control each affords the consumer, and further demonstrated how differences in the ability to match products to preferences influenced the effect of customization and personalization on the evaluations of chosen products.

Our discussant, Drazen Prelec, suggested how each of these papers brought to light several aspects of product selection decisions that go beyond the benefits directly associated with the product itself, but nonetheless influence consumers’ choices and satisfaction with them. Examples of these included consumers’ ability to justify their choice and/or assuage guilt that might be associated with selecting certain products, and how consumers’ perceptions of control during the product selection process influenced their judgment and decision processes.



Kristin Diehl and Gal Zauberman

Developments in interactive technology, customer databases and smart recommendation agents allow companies to personalize product offerings to each customer. One way in which companies can personalize their offerings is by ranking options according to the customer’s preferences. Consumers who receive an ordered list of recommendations based on their personal needs don’t have to sort through product offerings that are of little relevance to them. Hence, they can find products that better match their preferences (HSubl and Trifts 2000). Little attention, however, has been devoted to the inferences that consumers make from such a personalized list of recommended items. Ordering options according their utility for a specific customer creates a pattern of items over time; that is, consumers see a sequence of items, one following another. We propose that this pattern will have implications for consumers’ quality and variety evaluations of the merchandise and the retailer and for consumer search. The goal of this paper is to further our understanding of the impact of these patterns on consumers’ inferences.

Research on intertemporal choice shows the importance of temporal patterns on overall evaluations of extended experiences. Preferences toward sequences of outcomes cannot simply be deduced from preferences toward their component parts. Instead, this research indicates a general preference for improving sequences (e.g., Loewenstein and Prelec 1993). Thus, research on intertemporal choice suggests that the ordering of recommendation lists might impact consumers’ overall evaluations. Although research on intertemporal choice holds constant the items that compose a sequence and varies only its ordering (i.e., pattern), in our environment consumers actively engage in search, thus shaping the sequence of items they are exposed to. By determining what set of products consumers get exposed to, and in what order, search behavior affects the patterns that are perceived by the searcher and also the average level of quality consumers are exposed to.

We argue that the patterns created by product rankings and active search behavior have important implications for customer perceptions and evaluation of both products and retailers. We are especially interested in their effects on perceptions of product quality and variety and on the selection offered by a retailer, strength of preferences for the chosen product, and decision difficulty.

In order to partial out the effect of pattern independent of search behavior we distinguish two conditions. In one condition, search is held constant and participants sequentially view all options (forced search). In the second condition, differential search and stopping is allowed (free search). We used a principal-agent experimental task in which participants searched for and selected an electronic birthday card. The recipient of that card was described on several dimensions (i.e. relationship to the sender, age, gender, likes and dislikes, etc.). Participants saw one card at a time, with cards following each other in an ascending or a descending ordering. Ordering was based on the principle’s utility. Half the subjects were required to go through the entire set of cards offered by the retailer (forced search), whereas the other participants were free to stop their search when they felt that they had seen enough (free search). Following their search, participants answered questions about the retailer and his selection before choosing one of the cards they had seen by clicking on a thumbnail image of their choice. After that, participants answered several questions about the chosen card and their overall experience. This procedure was repeated for a second retailer that presented cards in the reverse ordering from the first retailer (i.e., ascending if trial 1 was descending and vice versa). Order of retailers and patterns was counterbalanced between subjects.

Our data show that the effect of patterns on search was in the predicted direction, such that decreasing patterns lead to less search. For increasing patterns, search was less than in the forced search condition, indicating that participants stopped when they found a card that was good enough, rather than continuing to search for the best card possible. We find significant effects of pattern on perceptions of quality and of variety of the selection. Increasing patterns lead to perceptions of higher quality o the selection. We also find a significant interaction of pattern and search on perceptions of selection variety. Perceived variety of the selection was not affected by pattern under free search; however, under forced search, searchers perceived more variety in the product selection when ordering was increasing. We also find that decreasing patterns leads to higher quality of the chosen option. This main effect was qualified by an interaction of pattern with search. Under forced search, pattern did not affect choice quality. In free search, however, a decreasing pattern lead to significantly better choices than an increasing pattern, mainly because searchers used a satisficing approach and stopped before having seen the entire increasing sequence. Within the decreasing pattern, allowing searchers to stop the search lead to higher quality choices, perhaps due to overload effects in the forced search condition.

In sum, we find that personalizing product offerings by ordering options according to their fit for the customer has important implications for customers’ perceptions of the retailer as well as the merchandise presented. We also illustrate the interactive effects of search and patterns of items over time.



Rebecca W. Hamilton and Bharadhwaj Sivakumaran

There are several product categories in which consumers customize products by selecting various features and options (Park, Jun and MacInnis 2000). For example, consumers purchasing a car often have the ability to choose options ranging from side air bags to leather seats to a premium sound system. Consumers’ ability to customize products in such a manner is rapidly increasing, and this freedom of choice is usually seen as a benefit for consumers. However, we suggest that there are cases in which consumers would rather not choose, cases where both consumers and firms would be better off if firms constrained choice by bundling certain options together.

There are several reasons why sellers might decide to offer bundles of options for a product rather than individual options. Consumers may have heterogeneous reservation prices for options (Yadav 1994, 1995), making it more profitable to offer option packages than individual options. Bundling options may lower production and inventory costs due to increased standardization. We focus on a third rationale, that of increasing consumer satisfaction. Previous research has suggested several reasons why consumers might be more satisfied choosing bundles than individual options: for example, consumers might be more satisfied when the choice process is less complex (Huffman and Kahn 1998; Iyengar and Lepper 2000), or when costs are aggregated rather than segregated (Thaler 1985). In addition, we propose that bundles may increase consumer satisfaction because they allow choices to be justified more easily (Tetlock 1985).

Because people are often reluctant to acknowledge their desires for luxury products or feel guilty about them (Strahilevitz and Myers 1998), part of the value of a bundle to a consumer might be in helping the consumer justify the purchase of "hedonic" (luxury) options when they are combined with "utilitarian" (practical) options. Strahilevitz and Myers (1998) demonstrate that if two different emotions are generated by the components of a product bundleBsuch as guilt generated by purchase of a hedonic product and good feelings generated by a donation to charity used as a purchase incentiveBthen the bundle components may have more value together than apart. Based on such complementarity, we suggest that both consumer satisfaction and sellers’ profits might be increased by bundling together options that individually meet either utilitarian or hedonic needs. It may not be mere coincidence that Volvo’s Cold Weather Package includes both Stability Traction Control (a utilitarian option) and heated front seats (a hedonic option).

A second reason bundles may increase consumer satsfaction is that they allow consumers to justify purchasing a larger number of options. Even if consumers are able to select options either individually or in bundles (e.g., for the Accord and Civic, Honda offers a range of models for which different bundles of options are standard, though most options can also be purchased individually), simply offering the options in a bundled form may have a positive effect on the number of options purchased and the total amount spent. Park et al (2000) show that a subtractive option framing method, in which consumers are presented with a fully loaded model and asked to remove options, increases the number of options selected by consumers relative to an additive framing method, in which consumers are presented with a base model and asked to add options. We propose that if consumers perceive bundles as justifiable reference points, satisfaction and sales may increase when bundles are offered.

Both of the proposed effects should be stronger when consumers feel they must justify their choices to others. If people are reluctant to acknowledge their desire for hedonic options, bundling based on complementarity might be particularly effective when purchases are made jointly, such as when spouses agree on the features of a new car. Bohm and Pfister (1996) demonstrate that contexts that foster justification enhance preferences for utilitarian (instrumental) features relative to hedonic (emotional) features. Similarly, a standard bundle of options may be perceived as a justifiable comparison when others will evaluate a consumer’s choice of options.

We tested these hypotheses experimentally by asking 292 subjects to choose options for a series of three products (a car, a computer and an apartment) using three different procedures. In a pretest (N=33), four of the options selected for each product had been rated as primarily hedonic and four had been rated as primarily utilitarian; these options were paired so that each pair included two options rated significantly differently on the hedonic/utilitarian scale, but for which subjects had indicated an equal willingness to pay. In the bundled condition, subjects were presented with three bundles of four options each; in the customized condition, subjects were presented with eight individual product options; in the unconstrained bundled condition, subjects were presented with three bundles of options and eight individual product options. For example, one third of the subjects selected options for a car in the bundled condition, options for a computer in the customized condition, and options for an apartment in the unconstrained bundled condition. We examined the nature of the options consumers chose, the amount they spent and their satisfaction with the choice process when faced with various constraints on the combinations of options they could choose and various choice contexts (high and low justification conditions).

Our results show that in general, people do not like to be constrained; satisfaction was significantly lower in the bundled condition than in the two unconstrained conditions. However, certain types of constraints increased satisfaction. For example, subjects were significantly more satisfied in the bundled/high justification condition when all of the bundles contained at least two hedonic optionsBin effect forcing them to choose luxury optionsBthan they were when they had a wider range of choices and could choose bundles containing no hedonic options. In addition, making bundles of options available in addition to individual options significantly increased the number of options selected. Though exactly the same options could be chosen in both conditions, subjects selected significantly more options in the unconstrained bundled condition than in the customized condition.



John Godek, J. Frank Yates, and Yeosun Yoon

In the past, firms divided the overall market of consumers ito a number of smaller segments in response to differing preferences (Smith 1956). More recently, firms have shifted strategies from marketing to segments to marketing to individuals (Kahn 1998; Prahalad and Ramaswamy 2000; Sheth et al. 2000). Firms employ two distinct approaches to individual-level marketing. Some, such as Amazon.com (books) and Travelocity (air travel), use personalization, where they attempt to understand consumers’ preferences with respect to a particular domain, and then recommend alternatives (HSubl & Trifts, 2000). Other firms, such as Dell and Gateway, use customization, where flexible and responsive manufacturing systems create products to meet the needs of individual consumers (Pine, 1993). Although merchants have helped consumers sort through alternatives for centuries, and skilled artisans have made build-to-order products available long before mass production was commonplace, the advent of interactive communications, flexible manufacturing, and just-in-time delivery systems have made it economical for many companies to offer different products to individual consumers on an unprecedented scale. These changes in consumer-firm interactions are designed to enhance the likelihood that consumers’ obtain products that match their preferences, however consumers’ perceptions of control are also affected by the differences in how products that are offered.

Control has several components (Averill, 1973), including behavioral control, the "availability of a response which may directly influence or modify the objective characteristics of an event" (Averill 1973, p.293), and decisional control, the "choice in the selection of outcomes or goal" (Averill 1973, p.289). A higher level of attribute specifiability (as afforded by customization) should lead consumers to perceive a greater amount of behavioral control as their ability to affect the outcome of the process (i.e., attributes of the selected product) is much more salient. Differences in perceived control have been shown to lead decision makers to more positively evaluate outcomes, independent of any actual difference in behavioral or decisional control (Thompson, 1993). Behavioral control has been viewed as a basic motivator of human behavior, with people having a general desire for mastery over their environment (White, 1959). A relevant outcome of increased perceptions of behavioral control in a product selection task is that when consumers perceive that they have more control over a selected product, they are also more likely to engage in dissonance reduction activities such as bolstering of chosen alternatives or denigrating of forgone alternatives, as they have no one else to blame if they perceive that they have made a poor choice (Janis & Mann, 1977). Such bolstering should result in higher evaluations of selected products when consumers perceive they have greater behavioral control.

Conditions also exist, however, where consumers do not wish to exert behavioral control, as when consumers think that others would be better able to identify the product that best fits the consumers’ preferences. In such cases consumers might wish to relinquish control to someone else for selecting a product (Hollander & Rassuli, 1999). Thus although it would be expected that consumers would value the control afforded by customization under most conditions and thus evaluate customized products more highly, condition which exist where consumers would rather defer control to another who they felt could facilitate a better decision than themselves, as might be construed when consumers select from a set of alternatives recommended by a firm via personalization.

In order to test this prediction, we conducted an experimental study with 87 participants who selected a product using a choice task framed as customization, where they were made to believe they were directly specifying the product and thus had more control, or framed as personalization, where participants thought that they were providing information to the firm in order to receive a personalized recommendation, and thus had less behavioral control. Other than the framing, the choice tasks were identical, wih the same information and alternatives presented. In this way we were able to manipulate the choice task such that it affected only perceptions of behavioral control and not perceptions of assortment, which might influence decisional control, and this was confirmed via several pretests.

Besides influencing participants’ perceptions of control, the effects of these selection protocol manipulations might also be expected to interact with participants’ perceptions of the firm implementing the protocol due to the recommendation being made in the personalization condition. In particular, we expected that in instances where the consumer felt the firm could facilitate a better decision, that she would be willing to relinquish control and thus evaluate products selected via personalization higher. Similarly, we expected that the same consumer, if she thought she could better identify the products that matched her preferences than the firm, would evaluate products selected via customization higher. As such, we included a manipulation of product category, where for one product participants perceived (as confirmed by pretest) that the firm could do a much better job at identifying products that matched their preferences (air travel), and for the other product participants felt that they could better identify the product that matched their preferences (pizza). Thus our study was a two (Selection Protocol) by two (Perceived Matching Ability) between subjects design.

Our results indicated that there were no significant main effects for selection protocol or perceived matching ability. However, the interactive effect of the two was significant for evaluations and perceived control. When participants selected a product via customization, they evaluated such a product more highly when it belonged to a category where they felt capable of identifying the product that matched their preferences (pizza in this case). However, when participants selected a product via personalization, they evaluated such products more highly when they belonged to a category where they felt the firm was more capable of identifying the product that matched their preferences (air travel in this instance). In addition, participants perceived they had more control when they customized products belonging to a category in which they felt they could identify the best match with their preferences, while there was no such difference when the selected products via personalization.