Consumer Decision Making in Naturalistic Settings: Salesman-Prospect Interaction

Richard W. Olshavsky, Indiana University
[ to cite ]:
Richard W. Olshavsky (1976) ,"Consumer Decision Making in Naturalistic Settings: Salesman-Prospect Interaction", in NA - Advances in Consumer Research Volume 03, eds. Beverlee B. Anderson, Cincinnati, OH : Association for Consumer Research, Pages: 379-381.

Advances in Consumer Research Volume 3, 1976      Pages 379-381


Richard W. Olshavsky, Indiana University

The potentially great influence personal selling can have upon consumer decision making has been recognized for a very long time by marketers. A great deal of time, money and effort is expended in recruiting, training, evaluating and compensating salesmen. Much has been written about salesmen-prospect interactions. But most of these efforts have taken the form of prescriptive formulas on "How To Become A Successful Salesman." Very few systematic studies have been performed; and most of these have been concerned with the influence of preselected variables such as age, sex or personality on the outcome of the transaction. Only a couple studies have attempted to examine the complex dyadic interaction at the process level. Consequently, in spite of the importance of the topic, today we know very little about consumer decision making in naturalistic settings (see Capon, Holbrook and Hulbert [1975] for a comprehensive and recent review).

There are at least two reasons for the neglect of research on this important area of consumer behavior. The first reason concerns the enormous complexity of the behavior of interest. Two (or more) people interacting over long periods of time (sometimes over multiple encounters) about products or services which are usually technically complex can generate a stream of behavior which is extremely complex. It is a challenging task to develop a methodology which permits reliable and valid measures of the important aspects of this behavior and to analyze the resulting data in meaningful ways.

The second important reason for the neglect of this important area of consumer behavior pertains to the lack of a theory of salesman-prospect decision making. The absence of such a theory makes it difficult to identify the important determinants of the dyadic interaction process and how these determinants interact over time to produce the behavior of interest.

These two important sources of difficulty are actually very closely interrelated. Good theory leads to the identification of relevant determinants of behavior which in turn facilitates the identification of appropriate methodology. Good methodology, on the other hand, produces results which facilitate the development of better theory.


The purpose of this paper is to advocate, in the spirit of a workshop, the general outline of a theory of salesman-prospect interaction and to illustrate the type of methodology that is implied by this new theory.


To explain the behavior of a salesman-prospect dyad, it is posited that it is first necessary to know the decision process of each member of the dyad with respect to the task at hand. This implies that a theory of individual decision making exists which is capable of generating behavior in considerable detail. One theory which meets this requirement is the Newell and Simon information processing theory (IPT) of human cognition (Newell and Simon, 1972; Olshavsky, 1975).

Newell and Simon's IPT accounts for human cognition in terms of an interaction which occurs among a person's goals, internal representation of the task environment and strategy. It is assumed that sequences of observed behavior are generated by the execution of the selected strategy. The strategy itself is nothing more than a particular configuration of elementary information processes which the person is assumed to be capable of interpreting and executing. IPT further postulates that cognitive man is to be viewed as one instance of a general class of information processing systems. The essential features of such a system are: receptors and effectors which transmit symbolic information between the system and the external environment, a central processor which can interpret and execute a limited set of elementary information processes and a memory which is capable of storing symbols and symbol structures over long periods of time.

These basic postulates are turned into testable hypotheses by the application of the theory to a particular problem solver, concept learner or decision maker in a particular task environment. For the particular instance, the problem solver's goals, internal representation of the objective task environment and his strategy are specified in terms of a precise data structure and a precise organization of elementary information processes to be executed. These processing rules and internal representation of the task environment are then expressed in a formal computer language and the logical consequences are derived by performing the indicated operations according to the specified rules. The resulting computer program, therefore, becomes a microtheory of the behavior of interest; at the same time, it is a detailed and testable model of that theory.

Although IPT has been developed and applied primarily to task environments involving single individuals, it is possible to extend this theory to situations where two or more individuals are interacting in the same external environment. Clarkson (1968) for example demonstrated how this theory could be extended to small group decision making by positing that the behavior of the group is the result of the two or more information processing systems acting interdependently over time. In group decision making, two or more individuals are required to work together such that the end result of the joint deliberation is a single decision; each participant's behavior becomes part of the task environment for the other participants. Interpersonal influence in this setting is represented as the effect one or more members of the group have upon the goals, strategies or internal representations of the others. The extent of such influence was taken by Clarkson as a measure of the degree of leadership expressed by each participant. Clarkson was able to show that a valid model of group decision making could be developed by first developing models of each participant and by identifying the type and extent of interpersonal influence which occurred over multiple group decisions.

Dyadic interaction between salesman and consumer differs in some important respects from small group decision making. Participants in a group decision making situation all share essentially the same basic goal. Moreover, they are likely to have similar internal representations of the task environment and similar strategies for attempting to achieve this common goal. In the case of salesman-prospect interactions, however, the goals of the participants will rarely be the same. The salesman's goal will generally be to sell the prospect some product or service from his existing inventory while the consumer's goal will be less definite and more idiosyncratic.

Likewise, while both salesman and prospect interact in the same objective environment, their task environments differ considerably. The salesman's task environment consists of his particular inventory of goods or services along with a particular prospect; the consumer's task environment consists of the salesman and the salesman's inventory of goods or services. In general we may expect the salesman to have a more accurate and complete internal representation of his task environment than the consumer does.

We may also expect major differences in the type of strategies elicited by each participant in an effort to achieve his goals. Whereas the salesman may have a well organized, pre-planned strategy for accomplishing a sale, the consumer's strategy may he poorly organized and in some cases nonexistent. In my study of appliance shoppers, e.g., it was found that several consumers simply turned the choice task over to the salesman (Olshavsky, 1973).

Just as in group decision making, interpersonal influence in salesman-prospect interactions can be represented in terms of the influence each participant has upon the other's goals, internal representations and strategies of each participant and the interpersonal influence which occurs between participants over time in the form of a computer model. Through simulation, the model can then be tested for validity. Validated models can then be used as the basic data from which generalizations about consumer decision making in naturalistic settings can be made. As an additional bonus, generalizations will also be forthcoming about salesman strategies and their effectiveness with different types of consumers. (Valuable insights about other marketing management considerations such as the appropriate depth and breath of inventories may also be gained from such models.)


The research objectives prescribed by IPT are to identify the goals, internal representations and strategies of human problem solvers. In order to expose these hidden mental processes as well as possible, Newell and Simon and others who have adopted the IPT have resorted to a technique called protocol analysis. Protocol analysis involves having the subject under investigation talk aloud as he attempts to perform the specified cognitive task. Close examination of the protocol data leads, by inference, to specific hypotheses about the goals, internal representations and strategies of individual subjects.

A group problem solving situation creates some serious methodological problems for an IPT researcher. Because of the potential impact that the exposed thought process of one participant might have upon the thought process of others in the group, protocol analysis is not a viable methodology unless it is used creatively. Clarkson, e.g., had individual participants talk aloud in a pre-task which was similar to the task they solved as members of a group. Another methodological problem of group or dyadic behavior arises from the very interactive nature of the process itself. This interactive process may result in important and rapid changes in the participants' goals, internal representations and strategies such that continuous monitoring of each participant is required. A final methodological problem that arises in the case of salesman-prospect interaction in naturalistic settings is that a great deal of relevant information can be directly "read" from the external environment by visual inspection. For instance, an appliance shopper can directly observe the styling, color, size and prices of alternatives and a salesman can observe the approximate age, social class and income of the shopper. All of these considerations pose serious methodological problems for the application of protocol analysis to salesman-prospect interactions.

In the interest of stimulating research of the type prescribed by IPT, a specific methodology will now be advanced to illustrate how some of these methodological difficulties might be overcome. The specific methodology to be proposed is a variation on the Willett and Pennington (1966) field study. In the Willett and Pennington study, audio recordings were made of actual transactions between salesman and prospect as they took place on the retail floor for several types of appliances. These audio recordings were later supplemented by post purchase interview data. The Willett and Pennington procedure has certain definite advantages. By dealing with actual transactions in actual settings, findings from such studies are likely to have high external validity. Moreover, since the subjects in this study were unaware that they were being recorded, they were not influenced by the experimental procedures.

Prom the point of view of an IPT theorist, however, the Willett and Pennington procedure has certain serious shortcomings. In particular, the audio recordings do not reveal the thought processes as well as protocol data can. The lack of detail in the audio recordings does not permit easy inference concerning the participant's strategy or their internal representation. And it is difficult with audio data alone to detect changes in the important aspects of cognitive behavior and to detect the factors which produce these changes. Audio recordings also fail to reveal the type and amount of information potentially available to participants by direct visual inspection.

The first modification that should be made to the Willett and Pennington procedure therefore is to supplement the audio recordings with video recordings. The second modification is to have each participant separately provide a detailed analysis of the transaction immediately after the transaction is over. This retrospective analysis should use the audio-video recording the way today's sports announcers and athletes use the instant replay for diagnosis of sports events. Such an analysis could provide the necessary additional detail concerning each participant's goals, internal representations and strategies throughout the transaction. The use of the audio-video recording will stimulate detailed analysis and will minimize any possible memory loss or distortion. Each participant's detailed remarks should be recorded for later analysis by the researcher.

These two types of data, the original audio-visual recordings and the audio recordings of the detailed retrospective analysis by the participants, can then be subjected to analysis by the researcher. Original goals, internal representations and strategies can be inferred as well as any changes which occur in them and the determinants of these changes. Hypotheses can then be formalized in the form of a single comprehensive model of both the salesman and the consumer and the interactions which occurred between them. The resulting model can then be tested by making detailed comparisons between the model's output and the actual transactional data. Validated models of several different salesmen and prospects can then serve as data points from which generalizations about consumer and salesman decision making in naturalistic environments can be made.

In order to facilitate the success of this proposed procedure, it is suggested that initial studies be limited in the following ways: only one type of product or service, or very few, should be investigated, only dyads should be studied as opposed to groups of three or more, realistic settings should be used but unnecessary distractions and interruptions should be minimized, e.g., by using an actual retail floor after closing hours and one shot transactions should be investigated as opposed to multiple exposure transactions. After some experience has been gained with the proposed methodology and theory, it will be an easy matter to relax these simplifying restrictions.


In closing, I would like to emphasize that a salesman can and often does exert a profound influence upon consumer decision making for many products and services. Until we come to grips with the complexities of dyadic and small group transactions, we will never have a truly complete understanding of consumer decision making and our ability to make meaningful suggestions for marketing management will be greatly limited. It is my hope that the theoretical and methodological proposals outlined in this paper, brief as they are, will stimulate further empirical research and theoretical development in this important area of consumer behavior.


Capon, Noel, Morris Holbrook and James Hulbert. The Selling Process: A Review of Research. (Mimeograph, submitted for publication to the Journal of Marketing 1975. )

Clarkson, Geoffrey P.E. Decision Making in Small Groups: A Simulation Study. Behavioral Science, Vol. 13, 1968, 288-305.

Newell, Allen and Herbert A. Simon. Human Problem Solving. Englewood Cliffs, N.J.: Prentice-Hall, 1972.

Olshavsky, Richard W. Customer-Salesman Interaction in Appliance Retailing. Journal of Marketing Research, Vol. X (May 1973), 208-12.

Olshavsky, Richard W. Implications of an Information Processing Theory of Consumer Behavior. Proceedings, American Marketing Association, Fall conference (1975), forthcoming.

Willett, Ronald P. and Allen Pennington. Customer and Salesman: The Anatomy of Choice and Influence in a Retailing Setting. Proceedings, American Marketing Association, 1966, 598-616.