Abstract: The fact that conversational commerce has been one of the most favorable ways for companies to reach customers has generated an increasing number of use cases for chatbots. Making a comparison between human-chatbot communication and human-human communication can shed light on the behavioral rules which determine how humans interact with intelligent agents and provide design and application guidelines for chatbots. This study compares human-chatbot conversation with human-human conversation in the context of online customer service to figure out whether two-sided arguments have different effects on users’ purchase intentions and experiences. A laboratory experiment is conducted with a 2 (user’s perception of the customer service agent: chatbot vs. human) 2 (conversation design: two-sided argument vs. one-sided argument). The experiment applies Wizard of Oz method, with humans performing service tasks while giving the impression to that users they are chatting with chatbots. The results show that in human-human communication, two-sided arguments have a negative impact on users’ purchase intention for the product, while in human-chatbot communication, the effect is non-significant. When users believe they are interacting with a chatbot, the positive effects of two-sided arguments on user perceptions of source credibility and pleasure are amplified.

Key words: Chatbots, Human-computer interaction, Conversational commerce, Two-sided arguments, Wizard of Oz method