Customer relationship management (CRM) at supermarkets is willing to interact with customers appropriately with the aim of making strong relationship and resultantly gaining maximum profits. Customers consist of various groups of people and have different needs, styles and expectations. Marketing management of a supermarket segments customers to respond their different demands correctly. Another important concern of supermarket managers is to detect profitable customers. These customers supply main profit of company and saving them guarantees existence of the supermarket. This research presents a model completing CRM process from understanding customers to assigning marketing strategies. Profitable customers will be distinced as a result of correct understanding of all customers. Present research is comprised of two phases. At phase one, dataset with recency, frequency and monetary (RFM) measures is constructed and clustered using K-means algorithm. Six segments of customers are detected based on the results of clustering. All segments are comprehensively analyzed and marketing strategies for them are described in phase two. Transactions of every segment of customers are separated and association rules are extracted using market basket analysis and Apriori algorithm. Consequent and also antecedent product items are proposed to customers who purchased antecedent product iems. So, dedicated marketing proposals are developed for some special customers.