Exploring Voice of Customers to Chatbot for Customer Service with Sentiment Analysis

Abstract

Abstract. Chatbots have been widely employed across a wide variety of companies and industries, from small- and medium-sized businesses to large corporations, and from e-commerce to financial institutions. Although chatbots have proven to be far more efficient and quicker than human agents, they do not always provide customers with a satisfactory experience because they lack a personal touch. Customer issues are often left unresolved and many are unsatisfied with chatbot services. This is unfavorable for firms that use chatbots for customer services as this jeopardizes their relationship with valued consumers. Thus, customer input is essential to streamline the product innovation process. This study uses a hybrid method involving lexicon-based TextBlob and logistic regression techniques to identify the sentiments of consumers toward chatbots for customer services based on user-generated content on Twitter. The results show that although people generally have positive encounters with chatbots, the gap between positive and negative sentiments is relatively small. This research provides insights that businesses can use to improve chatbot technology based on the voice of the customer to provide users with higher quality customer services in the future, especially since unsatisfied customers could be a threat to a business’s performance. Keywords: chatbot, customer services, sentiment analysis, social media mining, voice of customers.