Case Study: Natural language processing for call centers


AI-driven NLP models implemented in some areas of an organization's workflow can significantly improve the quality of time spent writing and reading emails and reduce response time, which leads to higher productivity, advancing customer service, and saving the company's funds. 

Business case:  

Call centers have to deal with much text and audio information and they spend plenty of time processing it. By integrating machine learning and NLP into the procedure you can predict customer behavior and find sales potential while saving time and costs. 

Use case 1: E-mails’ classification and prioritization. 

Background: The company receives around 30 000 e-mails monthly. It needs a staff of 10-11 employees working a full-time schedule in the call center to process these e-mails. 

Goal: To automate the indexation process as much as possible using machine learning and natural language processing/text mining methods and consequently free up capacities of the call center operators for more value-added activities. 

Approach: The solution is based on the existing platform, which contains a  

  • backend for natural language processing, including lemmatization, entity detection, topic detection, category classification and 
  • web-based front-end for setting up the workflows for manual classifying of missing or incorrect arrangement and visualizing. 

We also tailored the solution specifically to the needs of the given problem and customized the interface for integration with the call center system.

Use case 2: Understanding the content of the calls and finding sales signals. 

Background: The company aims to understand the content of the inbound and outbound calls and explain what concretely makes the clients come to a branch office and buy their product. Additionally, it is possible to enhance the call scripts using the collected information.

Goal: To identify what factors during the outbound call affect the client’s actions and make him arrive, not arrive at the branch office, and purchase or not purchase the product. 

Approach: Processing transcripts of the calls, information about arrival and purchase together with customer context: what products he bought, demographics of the client, etc. 

Stored information allowed us to build a model based on text flags such as frequency, term frequency-inverse document frequency, word2vec algorithm, n-grams. The model also uses customer data to predict the probability of purchasing a product based on a call inviting to a branch office.


  • Improved KPIs. 
  • Call transcripts were received thanks to the automatization of the classification and information extraction process.
  • Decreased capacities of the call center operators needed for these calls. 
  • Increased success rate of calls inviting to the branch office. 

Wish to learn more about us and our products?