The influence of a personal conversational recommendation on the user experience in chatbots
Chatbots are increasingly used in our daily lives, such as a virtual assistant, in e-commerce or in customer service. Although these chatbots can get the job done, customers often feel dissatis ed when, for example, the chatbot speaks out of context or does not remember information given earlier by the user. Also forms or advanced searches on the web are often replaced by chatbots. The question is raised whether this is actually an improvement. In this research is investigated how a personal conversational recommendation in uences the user experience of a chatbot whose purpose is to give movie recommendations. A personal conversational recommendation (PCR) is a combination of a setting in the dialogue that enabled a more form-free chat and the personalisation of a chatbot, such that it looks and feels like an actual form. Hypotheses about the user experience were testi ed using a fully implemented chatbot that is able to give a PCR versus a chatbot that does not have this attribute with user testing. Participants had to answer a questionnaire with questions that are mainly conform with the Technology Acceptance Model (TAM). The results indicate that both chatbots show similar user experience characteristics: the perceived ease of use, perceived usefulness, social presence and enjoyment were not signi cantly di erent. On this basis, it seems that people do not care about how the chatbot gives a movie recommendation. As long as the chatbot actually gives a recommendation, people are satis ed, because they achieved the intended goal, which is related with self-e ciacy in chatbots. Further research is needed to identify how people react to an actual form instead of a chatbot that converses in a form-like unpersonalized manner.
Faculteit der Sociale Wetenschappen