The Success of Marketing Management Support Systems

This paper provides an introduction to this Special Issue by a) providing a framework for evaluating the potential and actual success of marketing management support systems (MMSS), and b) briefly discussing how each paper in this Special Issue addresses the general topic of managerial decision making. The paper concludes by outlining some key questions that still need to be addressed. (Measures of Success; Decision Aids; Managerial Decision Making) Introduction This issue of Marketing Science is about managerial decisionmaking; howmarketingmanagers nowgo about making decisions and the description of a number of new decision aids that support marketing managers in the preparation, execution, and evaluation of marketing activities. The purpose of this opening article is to provide a framework for evaluating these decision aids and providing a better understanding of the interaction of these aids with the marketing manager. We then use this framework to briefly discuss how each of the papers in this issue addresses the general topic of managerial decision making and lay out possible future research agendas. Decision aids have been a central activity of marketing scientists for over 30 years. Initial efforts centered on building complex models that often looked for an optimal solution. Prototypical models of this type were MEDIAC (Little and Lodish 1969) and SPRINTER (Urban 1970). In 1970, John Little introduced the concept of a simple, but robust marketing model that usually required judgmental input from the manager. He coined the term decision calculus models for such an approach. Subsequently, we find the introduction of marketing information systems (Kotler 1966; Amstutz 1969),marketing decision support systems (Little 1979) such as ASSESSOR (Silk and Urban 1978), marketing expert systems such as ADCAD (Burke et al. 1990), and most recently, marketing case-based reasoning systems such as ADDUCE (Burke 1991). We use the term marketing management support systems (MMSS) to refer to this whole set of tools. Many of these tools have now become available for direct use in real world situations (Lilien and Rangaswamy 1998). Given this long history of development, it is somewhat surprising that there are few studies that investigate the impact of these models on the decision making of the manager and firm. One notable exception is the work of Fudge and Lodish (1977) who used a field study to document the impact of the CALLPLAN model (Lodish 1971). They did this by comparing sales generated by one set of sales people using CALLPLAN to another group using their habitual planning methods. They found that use of the model led to increased sales. Other examples of field studies can be found in Lodish et al. (1988) and Gensch et al. (1990). Interestingly, management did not completely carry out the major reallocation of resources to products and markets as was recommended by the model proposed by Lodish et al. (1988) even though this study showed positive impact. In contrast, the Gensch et al. model (1990) was implemented company-wide after the field study showed improved firm performance due to the THE SUCCESS OF MARKETING MANAGEMENT SUPPORT SYSTEMS Marketing Science/Vol. 18, No. 3, 1999 197 model. According to the authors, one of the key factors for the success of the MMSS was the direct involvement of the company’s CEO in the development and implementation of the decision aid. The above referenced studies deal with the use of MMSS for real-life decision making in companies. However, most of the empirical studies designed to test the efficacy of the MMSS have been conducted in laboratory settings. Chakravarti et al. (1979) carried out an experiment using practicing managers as subjects. They measured the effect of using the ADBUDG model (Little 1970) for supporting advertising decisions and found that use of the MMSS led to poorer decisions in terms of operating profit and prediction accuracy of market shares. McIntyre (1982) carried out a similar experiment, using the same type of MMSS, but found a positive effect of the use of the MMSS on profits. One of the major differences between the two experiments was that McIntyre used a setting where the underlying response function had no lagged effects, while in the Chakravarti et al. study the underlying response function was more complex, i.e., had a lagged term. It appears that managers were not able to estimate the true response in the latter case and this led to the poorer model results (Chakravarti et al. 1981). Zinkhan et al. (1987) studied the effects of several decision-maker characteristics on the success of MMSS measured by use and satisfaction. They found that cognitive differentiation (a cognitive style variable) and prior involvement with decision support systems (an experience variable) were positively correlated to the use of an MMSS. Van Bruggen et al. (1998) carried out an experiment in the MARKSTRAT (Larréché and Gatignon 1990) environment. They found a positive effect of the use of MMSS on market share and profit. However, subjects using theMMSS did not report having more decision confidence than those not using the MMSS. Likewise, McIntyre (1982) found no relationship between objective results of the use of an MMSS and subjective variables as reported by the managers. In a recent study, Hansen and Staelin (1999) found that managers’ confidence in their decision concerning the selection of an option among risky alternatives has more to do with their ability to take into account all the relevant factors affecting the choice than the decision rule they use based on these factors. We take these studies to indicate that managers’ confidence in the decision taken has little relationship to veracity of the decision. Hoch and Schkade (1996) studied the effect of the decision environment on the impact of MMSS. In their study subjects had to predict future credit ratings of applicants based on four financial characteristics of the applicant. In a predictable environment, historical cases and a pattern matching strategy turned out to offer adequate support to decision makers. However, in less-predictable (dynamic) environments, linear models were more effective decision aids. This finding indicates that the degree to which a decision support tool is effective may depend on the decision environment. We draw several conclusions from these studies. First, there is substantial proof that MMSS can increase firm profit and other measures of performance. However, this success does not appear to be universal, but instead depends on the specific characteristics of the situation in which the system is used and specific success measure one is looking at. Second, there is still a need for further research that provides better insights in the conditions under which MMSS are successful. From the studies reviewed above several antecedents of MMSS success emerge such as support from top management, cognitive style and experience of the MMSS user, and fit of the MMSS with the decision environment. However, the interaction of these factors and the effects of other factors are still not well understood. Third, studies of this type used many different measures for the success. Examples include the extent to which the MMSS was actually used by decision makers, the effect of an MMSS on market share, profit, forecast accuracy, decision confidence, and the acceptance of the system’s recommendations by management. In further work it is important to distinguish between different success measures, to examine their mutual relationships, and to be clear about which dependent variable(s) to include in empirical studies on the effects of an MMSS. In the next section we present a comprehensive framework of the factors that determine the success of an MMSS. WIERENGA, VAN BRUGGEN, AND STAELIN The Success of Marketing Management Support Systems 198 Marketing Science/Vol. 18, No. 3, 1999 A Framework for the Success of Marketing Management Support Systems We highlight five factors that determine the success of a marketing management support system. These are: (1) the demand for decision support (2) the supply of decision support (the decision support offered by the MMSS), (3) the match between demand and supply, (4) the design characteristics of the MMSS, and (5) the characteristics of the implementation process of the MMSS. Together with (6), the dependent variable success of the MMSS, these factors constitute the main building blocks of the framework presented in Figure 1. We posit that the match between the demand side (the decision processes to be supported) and the supply side (the functionality of the management support systems employed) is the primary driver for the potential success of an MMSS. The extent to which this potential success will be actually realized depends on the design characteristics of the MMSS and the characteristics of its implementation process (Davis 1989, Alavi and Joachimsthaler 1992). Most of the factors in the framework are self-explanatory and we will highlight only a few elements here. We start with the context of problem-solving activities of (marketing) decision makers, i.e., the demandside of decision support (Figure 1, Box 1). The early writings in the decision support systems/information systems (DSS/IS) literature (Mason and Mitroff 1973, Mock 1973, Chervany et al. 1972; Lucas 1973) mentioned three basic factors that characterize the decision situation. These are (i) the problem that has to be solved, (ii) the environment in which the problem is solved, and (iii) the decision maker who has to solve the problem. The problem being solved can be characterized by its degree of structuredness. Marketing problems vary enormously along this characteristic. Thus, sales-force allocation and media planning are examples of relatively structured problems, while designing a marketing communication or developing a marketing strategy are examples of less-structured problems. The decision-environment can be characterized by the level of market dynamics. When firms are operating in stable markets it is relatively easy to build mathematical models and perform some form of optimization. However, in turbulent markets decision makers are hardpressed just to understand and interpret what’s going on (Bucklin et al. 1998). Consequently, MMSS need to be adapted to reflect these less-structured conditions. Table 1 provides our characterization of the papers in this issue along these two dimensions. Interestingly, we find no papers addressing issues where there is low problem structure and the environment is turbulent. This is not surprising, since this situation is probably the most difficult to model. Still, it also points to the need for others to develop methods for providing help to managers in such situations. A third factor that characterizes the decision situation is the decision maker’s cognitive style, i.e., the process through which a (marketing) decision maker perceives and processes information. One common classification of this cognitive style is analytical decision making versus nonanalytical or heuristic decision making. It seems that an analytical cognitive style facilitates the use of MMSS (Larréché 1979, Zinkhan et al. 1987, Van Bruggen et al. 1998). However, Benbasat and Dexter (1982, 1985) found that especially lowanalytical decision makers have the most to gain from decision support aids if they actually use them. Van Bruggen et al. (1998) also observed that an MMSS can reduce the difference between highand lowanalyticals. The paper by Brown in this issue provides some insights into this difference in cognitive style by showing how analytically trained advertising personnel use different decision rules to evaluate potential ads than those used by the creative staff. The counterpart of the demand side is the supply side, i.e., the type of the decision support offered by the MMSS (Figure 1, Box 2). An MMSS can support a decision maker in different ways. It can help the manager carry out the actual calculations (e.g., find the “optimal” value), it can support the analysis and diagnosis of a specific situation, or it can come up with suggestions for users that stimulate the generation of (new) solutions. Perhaps most importantly, it can help frame the important issues and uncertainties associated with the problem at hand and in the process help the decision maker come to an acceptable decision. It is this feature of getting managers to think about the problem WIERENGA, VAN BRUGGEN, AND STAELIN The Success of Marketing Management Support Systems Marketing Science/Vol. 18, No. 3, 1999 199 Figure 1 Integrative Framework of the Factors that Determine the Success of a Marketing Management Support System in a structured way and “quantifying” their beliefs that is often pointed to as the major benefit of MMSS. However, it is still important to determine if this structure leads to better firm performance. MMSS can be classified as either being data-driven or knowledge-driven. Interestingly, only one paper in this issue, the paper by Goldenberg et al., is a prototypical example of a knowledge-driven system. In fact, no other paper submitted for review for this special issue fits into this classification. All of the other MMSS papers use existing databases, often coming from scanner data. Apparently, the developments in MMSS, so far, have been dominated by researchers with a modelbuilding background who prefer data-driven approaches and who are attracted to available data. With this said, we note the diffusion of the achievements in cognitive science and artificial intelligence into the field of consumer decision making (e.g., Bettman 1979; Alba and Hutchinson 1987). We forecast that these advances will soon find their way into the study of managerial decision making and marketing decision support systems. The success of MMSS depends on the match between demand and supply of the decision support (Figure 1, Box 3). Although such a match should be “obvious”, it is useful to classify the types of problemsolvingmodels found inMMSS and the conditions that favor the use of each type of mode. Table 2 lists a possible partitioning of marketing problem-solving modes in four categories: optimizing, reasoning, analogizing, and creating, along with the main characteristics favoring each mode (Wierenga WIERENGA, VAN BRUGGEN, AND STAELIN The Success of Marketing Management Support Systems 200 Marketing Science/Vol. 18, No. 3, 1999 Table 1 Demand Characteristics for Papers in this Issue