Nowadays, chatbots have been widely used as one of the companies' operational CRM solutions, due to their efficiency in handling many customer requests simultaneously. Based on previous studies, users can have different preferences regarding how to interact with a chatbot. By conducting a literature study from more than 36 relevant works of literature, this research explores four chatbot attributes that have trade-offs on user preferences: chatbot appearance, conversational style, chatbot proactive behavior, and user input control. The first attribute relates to how a chatbot is shown to the user. Alternatives to these visual social cues include; brand/company logo, avatar, and human photograph. The conversational style talks about verbal social cues, which is how information is conveyed by a chatbot using certain language variations. This attribute is based on two main dimensions in social judgment studies, which are warmth and competence. The third attribute relates to how a chatbot behaves, whether proactive or reactive. Finally, the last attribute talks about which method that users like to enter messages, whether by using quick replies or a free-text field. Then, we propose that heterogeneity in chatbot user preference is related to user characteristics, such as demography, experience in using chatbots, and personality.