Abstarct: Since in the competitive world of today, the business environment has changed from product-oriented into customer-oriented, and understanding customer behavior has become more important. One of the major challenges of the customer-oriented organizations is customer recognition, and distinction between different groups of customers and classifying them. Thus, by classifying customers baxsed on effective parameters and measuring the value of each class, marketing strategies and customer relationship management can be considered, and in the higher levels, the organization's large-scale strategies are tailored to the characteristics of different groups of customers. Therefore, the main goal of this research is classifying customers and understanding the behavioral rules of each class. In this regard, first, a new approach is proposed for classifying customer according to previous researches on customer classification, which is called ABC classification. Then, a frxamework is prepared to provide an appropriate solution for maintaining each class of customers that is being implemented on Samsung mobile phone data baxsed on axiom design technique. For this purpose, 5579 questionnaires were randomly filled in by Samsung's active agent customers during the three-month period from 96.1.2 to 96.1.5. This information was included as Rosetta software inputs and the behavioral rules of each customer group were extracted baxsed on Rough- set theory by the help of this software. Using the proposed classification method of this research, measures the value of each group of customers according to the services / products, and thus can provide appropriate marketing strategies for each group of customers.