Given the increasing share of services worldwide, one of the factors affecting customer satisfaction is providing services on time with minimal delay. One of the main concerns of in-person service centers has always been how to allocate tasks, plan and organize, and arrange for handling and routing services to customers. Lack of proper planning in this area will increase traffic load at the network level and increase environmental pollution, noise pollution, waste of time, waste of fuel, and ultimately dissatisfaction of consumers and technicians. In this regard, dividing daily work to provide service as desired and taking into account the opinions of individuals will not be an optimal choice. In this research, with a case study in an after-sales service company in the home appliance industry and using customer demands in Isfahan, using data mining methods, the geographic demand points of customers were clustered with the K-means algorithm and an attempt was made to reduce the search space of the problem by clustering geographical areas. Considering that the routing problem is an NP-Hard problem, the refrigeration simulation algorithm was used to find the path of technicians with possible customer demand while observing the daily work capacity in each cluster. In order to compare the results, the routing problem of technicians was also performed with the same constraints and without applying clustering. Computational results show that in the problem of routing probabilistic demand with respect to the daily work capacity limit for service workers, after clustering with the K-means algorithm, the objective function has significantly improved compared to solving the problem without applying clustering. Routing a service technician using clustering, while being responsive in much less time, has a higher repeatability test, and creates order and increases the sense of responsibility and control over service areas, and plays an effective role in reducing the time to handle consumers and achieving their satisfaction.