Conversational agents in the form of chatbots are increasingly becoming a reality in service encounters and messaging applications. Virtual agents are a particularly pressing topic for conversational commerce. This thesis adopted an experimental design with an actual virtual assistant powered by current technology. This study explored how humanlike cues such as the customer relationship influence consumer behaviour in cross- and upsell cases. Moreover, this study investigated proactive ways how to get in contact with customers. A particular focus was given to the attributes like picking the best time, channel and target group for the interaction. This resulted in a design of a prototype, which was further tested in an experiment. When analysing the results of the posttestonly control group design, the study set another focus besides the mentioned attributes. This focus is on the behaviour and differences of the participant age groups. This study followed the design science research approach.