Optimized customer service through NLP

• Automated classification of all incoming customer e-mails

• Facilitation of prompt, efficient customer feedback dispatching

• Identification of complaints hot spots and increase of throughput by 30 per cent

Zühlke supports a Swiss transport company with an efficient solution to perform qualitative and quantitative analyses on large volumes of customer e-mails.

The need to tackle the rising flood of e-mails

Today, more than 100 agents in a customer service centre respond to customer e-mails such as feedbacks or requests for support. Every year the number of requests is increasing. The company is looking for solutions to efficiently optimize their inquiry and improvement management processes using machine learning.

From scratch to a productive solution in six months

For this purpose, the team creates an architecture for the data pipeline in the cloud using Amazon Web Services (AWS). This architecture consists of an anonymized replica of the database, a text mining component to predict the categories of the e-mails, a search engine and a dashboard for visualization. Corporate Identity and Design, security, and data protection (GDPR Compliance) are set up and guaranteed for all relevant components. In only six months, Zühlke’s interdisciplinary team of data scientists, software engineers and consultants develop a productive solution in close collaboration with the business experts.

Valuable insights and further optimisation potential

The solution includes a self-service dashboard for the business experts, enabling them to extract valuable insights for their division in self-service. As a basis for further development, Zühlke establishes a modern, scalable cloud infrastructure automatizing the data processing end-to-end. Furthermore, Zühlke enables the client´s team of business experts and data scientists to elaborate further process optimization based on machine learning.