How should a chatbot speak?
Two experiments examined how humanlike language and appearance affect customer engagement through perceived competence and authenticity. Brand credibility also changed how the language cues mattered.
01 / The question
AI now speaks for organisations in moments that can be routine, frustrating or emotionally charged. My work examines how the design of those exchanges shapes what people feel, how they judge the service, and when human involvement still matters.
02 / The research
Across published studies, I examine the language, format and level of automation that people encounter in service interactions.
Two experiments examined how humanlike language and appearance affect customer engagement through perceived competence and authenticity. Brand credibility also changed how the language cues mattered.
A field study and two experiments found that text-based chatbots produced greater relief than voicebots in the settings examined. The relief pathway was evident for lower-severity failures, with a different pattern for more serious ones.
In a hospitality field experiment, fully automated service reduced customer loyalty relative to human-only service. Partial automation with human support performed better than full automation.
03 / From evidence to design
These studies offer practical questions for teams designing automated service. They are research-informed implications, rather than claims that an organisation has already changed its practice.
A service failure can change what people need from a bot. Consider the seriousness of the situation before choosing a channel or tone.
Ask whether an interaction feels competent, authentic and relieving, alongside whether it delivers a quick answer.
Design a clear path to human support when the task, emotion or customer preference calls for it.
Selected publications
Explore the original research, methods and full findings in these peer-reviewed papers.
How anthropomorphic language and appearance shape competence, authenticity and engagement.
Read publication ↗Comparing text-based chatbots and voicebots after service failures.
Read publication ↗What human-only, partial and full automation mean for customer loyalty in a hospitality setting.
Read publication ↗Research for impact