Customer satisfaction prediction

Written conversations with customers are a huge source of data that is often not used due to the complexity of their processing. Deep Talk allows companies, customer success, customer experience, and AI teams to transform this unstructured data into valuable data for decision-making.

A large bank decided at the beginning of the COVID-19 pandemic to move a large part of the conversations from the telephone to chat.
It achieved more than 5,000 conversations between customers and its contact center daily in a short period. To improve service and quality measurement, it added a question to the customer every time a conversation was closed: “Were you satisfied with the service? And the customer could answer on a scale of 1 to 10.
Only 14% of customers answered the question, yet 14% meant hundreds of responses every day, and in terms of data science, we added hundreds of labels to the conversations every day.

With all these tags and through an advanced predictive model using deep learning, in a short time, we were able to predict customer satisfaction with 86% of the services that did not receive feedback from customers, and based on this prediction; the Bank was able to take actions to build loyalty with dissatisfied customers, better reporting and special offers to customer segments.

Written conversations with customers are a huge source of data that is often not used due to the complexity of their processing. Deep Talk allows companies, customer success, customer experience, and AI teams to transform this unstructured data into valuable data for decision making.