Media selection model and its formulation by markovian decision process

Abstract

There are many steps of advertising decisions such as determination of advertising goals, determination of appeals, selection of media vehicles or specification of detailed copy and format. Among those steps of advertising decisions, this thesis develops a media selection model whose purpose is to determine the optimum number of insertions of advertisements in each media vehicle under consideration when the avaiable budget is given for each period. Markovian decision process is applied in developing the model and the performance criterion of this model is the average anticipated sales per unit period in the steady state. The three requisites for markovian decision process are obtained as follows; 1. Markov transition matrices. The markovian property of exposure level is dicussed and the exposure level is discretized in order to determine the transition probability materices. 2. Reward vectors The response function is used for the determination of reward vectors. 3. Alternative decisions The volume cumulative discounting system is discussed in detail. Under the discounting system alternative decisions are filtered by the budget constraint. The solution of the process is obtained by converting the process into the linear programming problem as Wolfe and Dantzig(11) did.