Multiple criteria decision support system for customer segmentation using a sorting outranking method

For companies, customer segmentation plays a key role in improving supply chain management by implementing appropriate marketing strategies. The objectives of this research are to design and validate a multicriteria model to support decision making for customer segmentation in a business to business context. First, the model based on the transactional customer behaviour is extended by a hierarchy with three main criteria: Recency, Frequency and Monetary (RFM), customer collaboration and growth rates. Customer collaboration includes quota compliance, variety of products and customer commitment to sustainability (reverse logistics and shared information). Second, the Global Local Net Flow Sorting (GLNF sorting) algorithm is implemented and validated using real company data to classify 8,157 customers of a multinational healthcare company. Third, the SILS quality indicator has been implemented and validated to assess the quality of preference-ordered customer groups and its parameters have been adapted for contexts with thousands of alternatives. The results are also compared with an alternative model based on data mining (K-means). The multicriteria system proposed allows to segment thousands of customers in ordered categories by preferences according to company strategies. The segments generated are more homogeneous, robust and understandable by managers than those from alternative methods. These advantages represent a relevant contribution to automating supply chain management while providing detailed analysis tools for decision making.

Automated summary

This research designs and validates a multicriteria model to support decision making for customer segmentation in a business to business context. The model extends the transactional customer behavior by considering factors such as customer collaboration and growth rates. The proposed multicriteria system can segment thousands of customers into ordered categories by preferences, providing more homogeneous, robust, and understandable segments compared to alternative methods. This contributes significantly to automating supply chain management and offering detailed analysis tools for decision making.