OPTIMISING CUSTOMER SERVICE IN E-COMMERCE: A COMPREHENSIVE ANALYSIS OF AI-BASED CHATBOTS IMPACT AND PREDICTIVE MODELLING

Abstract

As technology advances, chatbots are likely to play a larger role in streamlining consumer interactions and building positive relationships between e-commerce platforms and their customers. This study investigates the influence of Artificial Intelligence (AI)-based chatbots on customer support on e-commerce websites. This topic has received relatively limited attention in the Sri Lankan e-commerce context. Recognising the need for more empirical research on the role of AI-driven chatbots in customer care, this study examines their relationship with customer service experiences in the e-commerce industry. A rigorously developed questionnaire was presented to a sample size of 201 respondents to collect primary data for this investigation. The questionnaire was designed to elucidate the perceived role of AI-based chatbot capabilities in customer service on e-commerce platforms. As prediction models, an array of classification technologies, including Naive Bayes, SVM, MLP, J48 and random forest, were used. The sample's demographic distribution showed a Sri Lankan client base, with 48% male and 52% female respondents. Most participants were between the ages of 25 and 33, while respondents aged 52 years represented the smallest age group. Notably, the J48 decision tree algorithm achieved the highest classification accuracy among the evaluated models, at 100%. These findings are significant for refining and enhancing customer service on e-commerce platforms through the strategic integration of AI-based chatbots. Nevertheless, approach the reported predictive efficacy with caution, given the limited sample size and the need for additional validation with larger, more varied datasets.