Automating customer communications of the Social Insurance Board

The Social Insurance Board, in cooperation with Feelingstream, conducted five pilot projects aimed at analyzing customer calls and web chats in detail and finding patterns in the reasons for customer inquiries. As part of the pilot, the Feelingstream application transcribed Social Security Board calls from May to August 2020 (3 months) in Estonian and Russian. The pilot project involved an information line manager and two service managers.


The application developed by Feelingstream provides an opportunity to get both a statistical overview and a detailed understanding of customers' concerns across several customer communication channels - calls, emails and chat bots. Analysis of customer communications’ is the fastest and most immediate way to find out which parts of the services the customer is having troubles with and in turn that can be changed quickly. The analysis creates insights for managerial decision-making, which in turn leads to better and more efficient public services.


The descriptions of the service are available on the web and the citizens receive the necessary information themselves. However, it often happens that the text on the public web is incomprehensible to the client and too officially written. When a customer turns to a customer service specialist by phone, they can explain nicely and humanly and direct the customer to the right paths. Unfortunately, service managers create texts for websites, but often do not know what the customer needs or what information is confusing on the web. The analysis of the Social Insurance Board's customer calls focused on the nature of this problem.


The introduction of artificial intelligence in customer service usually requires a long preparatory period to establish the necessary detection models. Speech analysis is different because it only takes 1 day from the first call to transcription. This means that on the first day you can see what is in the speeches, on which topics the Agents need training and which speech has the most simultaneous talk.

The people working in the customer service of the Social Insurance Agency are very competent, they can help the customer like IT support specialists (in case of technical difficulties related to ID card), calculate amounts like accountants, act as law interpreters, help to navigate the web, and much more. The main reason for contacting the Social Insurance Agency is calls related to retirement, as the telephone is one of the main channels for receiving advice on pensions. In the course of the pilot project, the idea arose that if the client has a menu item “My benefits and services” in self-service, where there is no information about the pension, then at the moment you could write what you see in the future. This is one example of how joint discussions and cooperation lead to new ideas and solutions that have not yet been addressed.


In the pilot project, SKA chose Feelingstream as a partner, which specializes in analyzing customer contacts in Estonian and Russian. The analysis creates insights into managerial decision-making, which in turn leads to better public services to make them faster and more convenient for citizens to use.