5 An Application

5.3. A comparison of the two models

Two rather different marketing policies are indicated by the two models. The one-state model indicates that higher market shares will be achieved by making cutting the price of a usually expensive brand (though expensiveness is more important than relative price) while the four-state model indicates that relative price has both high criticality and intolerance for the 30 percent of the market accounted for by the two downmarket CDAP states. The one-state model indicates also that the market values moderately low fashionableness while the four-state model indicates that the largest CDAP state (in terms of the share of volume purchased) values extreme fashionableness very highly while the second largest state values moderate fashionableness and is less extreme in that valuation.

The four-state model also yields more plausible results with regard to specialness. In the one-state model, the special ingredients are ideally high and important while the specialness of the brand is ideally low and not very important (because criticality is low). In the four-state model, the ideal for special ingredients is the same for all CDAP states (0.4) but the importance is significant only for the up-market states. Specialness, on the other hand is important for all four CDAP states though one of the up-market states wants less specialness than do the others.

It is typical of applications of IMIS to such data sets that encouraging the definition of a larger number of CDAP states not only increases accuracy in tracking market shares but also, and more importantly, yields more detailed qualitative information which renders the results more plausible and easily interpreted.

In the application reported here, we can see how important the additional qualitative information is by comparing the two models' respective tables of preferences with the expert's specification of the brand characteristics in Table 3.

Comparing the focus brand characteristics with the ideal characteristic mix of the one-state model, the focus brand would appear to miss the market by being seen as imported, the wrong size and lacking in special ingredients. On the other hand, it got fashionableness about right -- though it is not all that far out from the other brands in that respect. In general, the focus brand's distinctive features were not well suited to the market described by a single CDAP state.

In the four-state model with preferences reported in Table 2, the focus brand's importedness looks offensive and important only to State 2 while both up-market CDAP states would be attracted by its size. Where the brand seems to lose the custom of the up-market CDAP states, accounted for 70 per cent of sales volume between them, is in its lack of fashionableness. The only CDAP state that would values its lack of fashionableness is State 3 which would be put off by the price and expensiveness.

Artificially Intelligent Specification and Analysis of Context-Dependent Attribute Preferences - 03 NOV 97

[Next] [Previous] [Up] [Top] [Contents]

Generated with CERN WebMaker