Case study Kooperativa


Using advanced natural language processing (NLP) helps to understand customer interactions and automatically route cases to the right specialists, detect and monitor important topics customers care about, and understand what drives customer satisfaction and predict churn risk.

Business case  

The insurance company Kooperativa transcribes calls from within the call center to determine the most common customers’ complaints. It aims to distinguish which topics are prevailing and, therefore, identify the problems faced by their clients. The company was also interested in the churn analysis to reduce the loss rate of their customers.


Our solution, based on machine learning and natural language processing consisted of evaluating the net promoter score (NPS) forms and churn analysis. We analyzed the duration of the call, number of calls from the same customer, intonation, pitch and speed of the client’s speech. We proved the hypothesis that it is possible to predict from the call whether the customer is dissatisfied, and that it could lead to churn.

During the three months of the project duration, we used NLP practices, working with audio and text evaluations, machine learning for categorization of NPS forms, AdPicker for improving AdForm advertisements, and helped to optimize the cost of online acquisition. 


  • Creation of personal profile based on a combination of Google Analytics and AdForm. This resulted in an increased web traffic per person and cost reduction of one visit by 40 percent.  
  • Categorizing of 80 percent of the NPS forms.   
  • Prediction of churn by analyzing customers’ calls.  

Our vision  

  • To implement the identification of customer dissatisfaction quickly.  
  • To connect data as much as possible for a better customer experience. 
  • To target advertising more precisely on relevant customers.  



Categorization of NPS forms by topic based on free text fields. The solution significantly facilitated manual sorting, as up to 80 % of forms can be categorized automatically. Other outputs included keyword detection and text sentiment determination.

Wish to learn more about us and our products?