Results of a 2 (two-sided vs. one-sided recommendation message) ✕ 3 (warm vs. competent vs. neutral communication style) between-subject online experiment show that a two-sided recommendation message increases purchase intention, but only for chatbots using a warm or competent communication style. Whereas a warm chatbot leads to higher purchase intentions of a recommendation through promoting its source persuasiveness, a competent chatbot increases recommendation effectiveness by promoting message persuasiveness.

Source

Paper related to this workshop and moc-persuasion.

Can Chatbots Be Persuasive? How to Boost the Effectiveness of Chatbot Recommendations for Increasing Purchase Intention

Melanie Schwede, Nika Mozafari, Niclas von Schnakenburg, Maik Hammerschmidt

The following content (reaction paper) was generated by an LLM.

What is the summary of the paper?

The paper by Adam, Wessel, and Benlian (2021) investigates the use of two-sided recommendation messages as a persuasive communication technique in the context of chatbot shopping. It explores how including both positive and negative features in product recommendations can affect persuasiveness and potentially lead to different behavioral outcomes. The study suggests that more research is needed to understand the effectiveness of various persuasive communication strategies in improving the persuasiveness of chatbots and their recommendations.

What is the research question? (How) was it answered?

The research question (RQ1) is: How does a two-sided (vs. one-sided) recommendation message from the chatbot affect the customer’s purchase intention through the perceived persuasiveness of the recommendation message and source? It was answered by conducting a 2x3 between-subject online experiment with 288 German-speaking participants, which found that two-sided recommendation messages do not increase purchase intention without the humanization of the chatbot. However, when chatbots are humanized, a warm communication style enhances the persuasiveness of the chatbot, and a competent style improves the persuasiveness of the recommendation message, both leading to increased purchase intention.

What research approach did the authors choose? Is it appropriate for achieving the chosen research objective?

The authors chose an experimental research approach using a scenario-based method with questionnaires to measure variables such as persuasiveness and purchase intention on a Likert scale. This approach is appropriate for achieving the research objective as it allows for controlled manipulation of variables (one-sided vs. two-sided recommendation messages) and measurement of their effects on customer behavior, ensuring high internal validity. The use of established scales for measuring constructs further supports the appropriateness of the methodology for the research question.

What are the key findings of the article? What are the implications?

The key findings of the article are that two-sided recommendation messages do not increase purchase intention without the humanization of chatbots, humanized chatbots enhance the effectiveness of such messages, and both message-level and source-level persuasiveness significantly increase purchase intention. The implications are that firms should use chatbots with a warm communication style to strengthen customer relationships and a competent style to enhance message quality, and that human-like design elements are crucial for the effectiveness of persuasive strategies in chatbots. The study contributes to research on conversational commerce and the humanization of chatbots.

What are the limitations of the article? How might future work build on this article?

The limitations of the article include a focus on initial evidence without long-term behavioral change analysis and a lack of consideration for the interaction between message-level and source-level persuasiveness. Future work could build on this article by examining the long-term effects of persuasive communication in chatbot interactions and exploring the interplay between different levels of persuasion to provide more comprehensive insights for firms. Additionally, research could investigate the effectiveness of various persuasive strategies in different chatbot designs and contexts.

@inproceedings{schwede_can_2023,
  title      = {Can Chatbots Be Persuasive? How to Boost the Effectiveness of Chatbot Recommendations for Increasing Purchase Intention},
  url        = {https://papers.ssrn.com/abstract=4305315},
  shorttitle = {Can Chatbots Be Persuasive?},
  booktitle  = {Proceedings of the Hawaii International Conference on System Sciences},
  volume     = {56},
  pages      = {3454--3463},
  author     = {Schwede, Melanie and Mozafari, Nika and von Schnakenburg, Niclas and Hammerschmidt, Maik},
  year       = {2023}
}