Research brief
Empathic chatbots: a double-edged sword in customer experiences
Antoine Juquelier, Ingrid Poncin & Simon Hazée · Journal of Business Research (2025)
In one sentence
Chatbots that show empathy make customers more satisfied, they feel more human and their answers seem more useful, but the same empathy hurts satisfaction when customers are under time pressure.
Why this matters
Customers increasingly turn to chatbots to rebook a cancelled flight, order a coffee on the run or schedule a medical appointment. Advances in AI now let chatbots detect a customer’s emotions and respond with care, and many companies are betting that this “artificial empathy” will close the gap between bots and human agents. Yet earlier studies disagreed on whether empathic chatbots actually improve customer evaluations, leaving managers unsure whether the investment pays off.
What we did
We ran three experiments with about 600 UK and US participants. Each watched a video of a customer-service chat and imagined being the customer. The customer’s messages and the outcome were identical; only the chatbot’s tone changed: empathic or purely mechanical.
Study 1
Travel insurance
Does empathy raise satisfaction?
Study 2
Van rental
Why does it work?
Study 3
Accommodation cancellation
When does it backfire?
What we found
Empathy pays off. Customers were more satisfied with an empathic chatbot than with a mechanical one, even after accounting for their age, gender, chatbot experience and preference for human contact.
It works through two routes. Empathic chatbots feel more human (social presence), but they also make the information itself seem more accurate, relevant and complete, even though both chatbots pointed customers to the same resources. Interestingly, perceived information quality explained more of the satisfaction gain than the human touch did.
Under time pressure, empathy backfires. When customers had only five minutes left to cancel and get a full refund, satisfaction with the empathic chatbot dropped, while the mechanical chatbot was unaffected. In a hurry, customers want clear, task-focused answers; warm language becomes a distraction and even lowers how useful the information seems.
Key takeaways for managers
Add empathy, even to routine chatbots. Short, transactional bots are usually built to be purely functional; empathy generally improves the experience at little cost.
Treat empathy as part of information design, not decoration. Acknowledging the customer’s situation makes the same answer feel more relevant and complete.
Let the chatbot read the clock. Use no-frills scripts for requests typically made in a rush (deadlines, cancellations, urgent rebookings), or detect urgency early in the conversation and dial empathy down.
Don’t assume one size fits all. The same empathic behaviour can delight one customer and irritate another, depending on their situation.
How to design an empathic chatbot
Our empathic chatbot combined three ingredients, mirroring the dimensions of empathy long documented in psychology research: understanding the other person’s situation (the cognitive side) and sharing and caring about their feelings (the affective side):
Perspective-taking
Acknowledge the customer’s situation; personalise the answer.
Mechanical: “Please find the link about the refund policy below”
Empathic: “I understand your concern about the refund”
Empathic concern
Show care and reassure.
Mechanical: “Write your questions here and I’ll look for answers”
Empathic: “Don’t worry, I’m sure I can help you”
Emotional contagion
Mirror the customer’s feelings, for instance with emoticons and emotional words.
Empathic: “You are right; coverage can be confusing”
Keep in mind
These were scenario-based experiments with video conversations, not live chats. In the third study, empathy did not raise satisfaction even without time pressure, possibly because customers expressed fewer emotions there. We tested the three empathy ingredients together, so their separate contributions remain open; as does how empathic bots fare in highly emotional situations such as medical or breakdown assistance.
Citation: Juquelier, A., Poncin, I., & Hazée, S. (2025). Empathic chatbots: A double-edged sword in customer experiences. Journal of Business Research, 188, 115074.
