Skip to main content

Case studies

Prioritised customer requests

A text analytics solution using machine learning technology is used to prioritize customer requests at OTP customer service. A “virtual assistant” is working for the customer service, significantly reducing the time spent on a case.

The OTP Group is Hungary's largest financial services provider and a leading player in the Central and Eastern European region through its subsidiaries. It provides high-quality solutions to the financial needs of 13 million customers in 9 countries through more than 1300 branches and state-of-the-art electronic channels.


In a competitive market, the quality, speed, accuracy, compliance and controllability of customer service are increasingly complex:

  • The MNB Recommendation No. 13/2015 (X. 16.) on the complaints handling procedure of financial institutions aims for the fastest possible response to important requests. However, letters received by the customer service are processed without prioritisation, just in the order they are received, which could lead to fines.
  • The response time for requests received via electronic channels is increasing, even though an average of 100 e-mails per administrator per day are processed. The level of service must be raised without increasing costs.


A text analytics solution using machine learning technology is used to prioritise incoming requests at the OTP customer service. Response times were reduced by 20% using the virtual assistant, Emilia, which handles 30% of customer responses autonomously.


Modular deployment of a solution combining machine learning techniques and artificial intelligence algorithms to process written requests in real-time.


  • Consultation, definition of success criteria.
  • Processing past email messages using Clementine's CLEMTEXT solution.
  • Gathering basic information


  • Ensuring parallel operation
  • Redesign of response processes:
    | "Rutin" case management with machine response: automatic operation without operator intervention.
    | Identifying topic categories, assigning them to organisational areas to ensure consistent automatic signing to the relevant departments, administrators.
    | Sorting incoming messages by urgency, identifying customer requests requiring immediate action and creating automatic alerts.
  • Quality assurance, protocol control.