Market structure analysis is to partition the set of brands into subsets (submarkets) such that brands within a particular submarket compete more "closely" with each other than with brands belonging to a different submarket.
One of shortcomings of the existing behavioral approaches is using brand switching data. For product market in whitch variety seeking is an important determinant of consumer behavior or learning effect is large, brand switching data can't measure correctly "substitution" between brands. It misleads complements to substitutes.
This paper use compromise switching data whitch is'nt affected variety seeking and learning effect. It lead more correct market structure. Through empirical analysis of cabonic drinks we confirm the fittness of compromise switching data. This research mearningful both brand manager and retail manager.
Overlapping cluster analysis and comparative analysis of product market whitch variety seeking is small are suggested for further research.