Search

This report covers certain research which has been directed to analyzing and systematizing historically independent market research steps into a rational marketing program for introducing a new product. In the general case, this system is applicable to many situations where dicisions under uncertainty must be made in the light of incomplete and not necessarily comparable data. The system's inter-relationship of tasks and decisions can best be illustrated and formalized by means of a network. The links of the network represent research tasks and the nodes represent evaluations or decisions undertaken in the light of all prior information. These alternatives together with their associated costs and risks are used to determine an optimal course of action for acquiring further information to achieve an optimum goal.

A set of recursive relations called NEWS (for New-Product Early Warning System) is here presented in its own right as a separable part of the DEMON System for marketing new products. NEWS is designed to distinguish the variables and relations that are usually of interest for market-planning by reference to data availability and the decisions that might be made advertising promotions and product properties. Other uses and possible further extensions are examined by reference to planning media schedules and related dynamic applications. (Author)

Media planning has heretofore been restricted, by and large, only to the consideration of average frequency. A recently developed mathematical model explicitly considers both cumulative duplicating audiences over a variety of time periods and the simultaneous selection and scheduling of media. This model thus required the development of methods for handling such nonlinear concave aspects of audience characteristics within a linear programming model and for generating distribution of frequency data within the computational processes of the model. This is done by means of logarithmic transformations and, for the discrete distributions of frequencies, by log-normal approximating devices. Each of these two techniques is important in its own right as a type of method which may be employed in a wide variety of modelling or computational situations. Thus, because neither of these techniques is restricted to the area of media planning per se, it seemed worthwhile to present them on their own terms. (Author)

A goal programming model for selecting media is presented which alters the objective and extends previous media models by accounting for cumulative duplicating audiences over a variety of time periods. This permits detailed control of the distribution of message frequencies directed at each of numerous marketing targets over a sequence of interrelated periods. This is accomplished via a new logarithmic non-reach device and a continuous lognormal generation of the discrete message frequencies. (Author)

DEMON (Decision Mapping via Optimum GO-NO Networks) is a chance-constrained programming model for use in developing a strategy for marketing new products where, inter alia, the pertinent statistical distributions are only partially known and, in fact, may change in response to the strategies employed. The present report depicts DEMON in terms of the decision-and-information networks which are of primary concern for managements confronted with problems in new product marketing. In addition the recursion relations used to establish an optimal marketing plan are discussed in a case history context. (Author), Research conducted in cooperation with Northwestern Univ., Evanston, Ill., rept. no. Systems Research Memo. 138, and with Batten, Barton, Durstein, and Osborn, New York.

Numerous alternatives are available in the kinds of studies which can be elected in the market research which should precede the introduction of a new product. This gives rise to a possible network interpretation and treatment by an associated chance-constrained programming characterization and analysis. The resulting model, called DEMON (Decision Mapping Via Optimum GO-NO Networks, AD-612 187) is replaced by an approach based on extremal equations. A general analytic characterization is achieved and then replaced by more special ones. Means for effecting study decisions and inferences are discussed along with the sensitivity analyses that we associate with GO and NO preemptions, relative to profit and payback constraints. Payback (in a constraint) is here distinguished from its use as a criterion for choice (or objective) , the latter being here oriented toward MEMP (maximize expected maximum profit). (Author), See also AD-414 521, AD-612 187.

This paper develops a new method for the solution of DEMON-type functional equations. It is shown how the latter can be reduced to solution of a separated system of simpler equations which, for discrete distributions, can be solved by linear programming methods. The reduction also permits general characteristics of the solutions to be inferred. Additionally, methods of approximation and bounding are developed and interpreted for the general case. Because of the importance accorded to log-normal distributions for the newproduct marketing problem, further details and developments are undertaken with reference to this class of distributions and then illustrated by reference to the single-parameter case. (Author), Supported in part by a Ford Foundation Grant and Northwestern Univ. under Contract Nonr1228 10, and Projs. NR1228 10 and NR047 021 with U. S. Office of Naval Research.