Market share analysis and prognosis using qualitative reasoning

摘要

Today, extensive quantitative modelling of the performance of marketing activities is possible due to the availability of rich data sets on product sales and related marketing actions. However, there are a number of product categories for which only inaccurate sales figures or data about marketing efforts exist (e.g., because a significant part of the sales is realized in stores without scanners or which do not make the data available to firm outsiders). In such cases, reliable quantitative findings about the impact of marketing-mix variables on sales may not be achievable but qualitative reasoning may, at least, indicate likely implications of past or planned marketing activities in terms of directions of change. Even in markets with good data qualitative reasoning may be a worthwhile first modelling approach when market conditions change so significantly that a previously used model is deemed unsatisfactory and not enough data exist yet for a specification of a revised quantitative model. We show specifically how qualitative reasoning can be used to diagnose and predict market share changes. The diagnostic part represents a combination of qualitative reasoning and extensions of a previously published rule-based expert system for that task while the prognostic part is solely based on qualitative process theory and order of magnitude reasoning.