A mathematical programming model for market share prediction

The primary objective of the multiattribute models in the area of marketing is to discover the way the consumers make their choice among available alternative products. Discovering the process of choice of the customers serves as a basis for many marketing decisions such as new product planning, advertising message selection, product positioning, pricing, concept testing, sales forecasting, benefit segmentation, personal selling strategy formulation, etc. The multiattribute models proposed for this purpose can be grouped into classes: (i)models based on mathematical programming techniques, and (ii)models based on statistical techniques. The recent literature surveys indicate that the results obtained from linear programming models show greater stability and better predictive ability. In this study therefore the authors will concentrate only on the mathematical programming methods of estimation in general and on a new market share prediction model in particular. The main feature of the proposed model is that it recognizes the dependence of attribute weights on attribute scores through a step function. The predictive ability of the model is discussed via a hypothetical numerical example.