The Design And Implementation Of Chinese Mobile Group Customers’ Warning Loop Management System

Currently, China’s communication industry is under a competition pattern oftripartite confrontation, and the newly-added market share is gradually reduced. Thecost spent by the company to attract a new customer is five to ten times of the cost spentto maintain an existing customer, which results in a situation that the three big operatorsChina Mobile, China Unicom, China Telecom gradually focuses their main battlefieldon abnormality maintenance and retention of their stock market. Because the customermarket of the group accounts for more than50%medium and high-end customers instock market, the importance is increasingly obvious and the competition also becomesfiercer. Due to the shock caused by its opponents in customer market, China Mobileconfronts more and more pressure in its retention work. The original early warningmethod and system can’t provide an effective guarantee for the retention work. It isnecessary to build an accurate customer abnormality management system to foreseeabnormal group customers and then adopt targeted retention strategy to effectivelyretain group customer and enhance customer’s viscosity and loyalty index.Therefore, with reference to domestic and foreign communication operators’experience, this paper combines with actual demand of China Mobile customer marketto build customer abnormality closed-loop management system of China Mobile for thepurpose of solving the problems existing in retention work. This paper mainly takescustomer life cycle theory as guidance, and comprehensively makes use of decision treealgorithm, survival analysis algorithm, logistic regression algorithm and the “improvedlogistic regression algorithm” proposed via research in this paper to build an earlywarning model for customer’s abnormality and catch customer’s abnormality tendencyin advance. Moreover, based on customer’s behavior property and research result, thispaper makes use of two-step clustering algorithm to build customer segmentation modeland subdivide group customers into five groups so as to realize the positioning ofcustomer’s abnormality reason. Through comprehensive application combined withearly warning model and segmentation model, it is able to build more accurateearly-warning model for group customer’s abnormality. By use of early-warninginformation and abnormality reason provided by early-warning model for groupcustomer’ abnormality, this system generates abnormality early-warning work order information, which will be sent to managers and customer manager later. Then, thecustomer retention scheme is formulated according to the early-warning work orderinformation to carry out maintenance work, feedback to maintenance effect, andevaluation on maintenance effect.The design of three layer architecture of B/S is adopted in this system，and thedevelopment will be in strict accordance with development model of the softwareengineering for analysis, system design and development. In the first layer, presentationlayer, system user interface uses Web browser to ensure that different users withdifferent access methods in different locations to access and manipulate. In the secondlayer, function layer, the system uses B/S structure to external systems to provide Webservices, uses ASP.NET, JSP and C#language for development of Web applications tomeet the needs of different users’ access. In the third layer, data layer, the system usesSQL Server2005as the database server, and uses ADO.NET as data accessinfrastructure. This system can run on the Windows2003/XP/7operating system.