The Potential of Chatbots: Analysis of Chatbot Conversations

Abstract

The idea of utilizing computers for question answering tasks has been around from the early beginning of these systems. First algorithms with the aim to accomplish this were already implemented in the early 1960s. In recent years, chatbots have been gaining enormous popularity in various fields. In the context of business applications, they are considered as useful tools for improving customer relationships. In this paper, chat conversations between customers and the chatbot of a telecommunication company are analysed to find out if these interactions can be used to determine a) users' topics of interests and b) user satisfaction. To reach this goal, chat conversations are interpreted as sequences of events and user inputs are analysed with the help of text mining techniques. The study shows that based on users' written conversational contributions, valuable insights on users' interests and satisfaction can be gained. The majority of users leave the chat conversation after a short period of time if the chatbot was not able to give the desired answer right away. Moreover, a huge number of conversations deal with similar topics. Our results imply that companies offering chatbots must thoroughly analyse the collected data to gain more insights into their customers' needs. Based on our findings, they can improve customers' satisfaction by offering personalized service and implementing real-time feedback.