CRM systems have been popular as they enable organizations manage their relationship with their customers and the relevant business processes more effectively. This study focuses on developing an efficient way to manage customers’ needs and complaints using machine learning techniques. We adapted two topic modelling techniques, LDA and GSDMM, to find out the main themes mentioned by customers with negative sentiment. LDA and GSDMM were used in a complementary way considering major limitations and advantages of the both technique. Using LDA, four topics emerged from the data. GSDMM algorithm was used to analyze the data using the same number of topics. The results showed that the most commonly topic discussed by the customers were “unsubscribe” and “login” in our case. Organizations using the same method can gather further information from the analysis and also customers to find out a solution to their current problem. Combining machine learning techniques like in this study can help organizations to realize the problems and develop solutions simultaneously when these analysis are integrated into their system.