Demo
With the rapid development of technology, all companies that have completed their digital transformation manage and record their transactions with their customers digitally. Satisfying the customers in the customer relationship management database, reducing the customer loss rate, and evaluating opportunities to satisfy dissatisfied customers is critical in sales and customer continuity. By keeping and storing records of purchases, complaints and other customer transactions, businesses can digitally calculate metrics such as customer satisfaction, retention, and churn.
Customer feedback, customer complaints, customer requests, social media messages, messages sent through the company’s website, text conversion of the conversation between the customer and the customer representative through the call center, and the customer’s correspondence with the chatbot or customer representative and many other text format data can be digitized with the help of natural language processing techniques and analyzed with statistical analysis approaches, machine learning and deep learning approaches to obtain more detailed information about the relationship between customers and the company.
Depending on the state of this relationship, there will be opportunities to identify areas of improvement for the business in many aspects, such as warming up the customer, resolving customer complaints, recommending different products to the customer, identifying operational deficiencies, identifying the issues that are intensely problematic and finding solutions. Natural language processing can also be used to understand customer demands and identify customer service issues. For example, the questions that customers frequently ask on a particular topic are analyzed, and the answers to these questions are examined. In this way, a more detailed understanding of what customers’ complaints or problems are can be obtained, and measures can be taken to care of them.
In this study, the Next4biz CRM R&D team worked on text segmentation based on natural language processing and machine learning on actual operational data belonging to customer complaints and customer relations modules kept on Next4biz CRM software. This study aims to obtain attributes that businesses can use in other analyses of their customers by using the available customer relationship text data. In this study, Turkish text data obtained from different data sources of customers were digitized using the TF-IDF method. Then the texts were segmented using the K-Means algorithm. The quality of these segments is evaluated using Elbow, Calinski-Harabasz, and Davies-Bouldin metrics.
