AI · Mixed-method empirical research (scenario-based quantitative studies plus exploratory and qualitative work)
Chatbots’ effectiveness in service recovery
International Journal of Information Management · 2023
Summary
Perceived safety, empathy and anthropomorphism affect different components of chatbot trustworthiness; trustworthiness increases willingness to forgive after failure and reduces negative word of mouth. The paper highlights privacy concerns as a countervailing issue and finds chatbot traits work through trust-related mechanisms rather than as undifferentiated design features. For a founder, use this as a question to test in your own context, not a universal playbook.
For your marketing
Possible founder application
Treat this as evidence for the specific customer behavior and adoption question studied. Compare the setting and sample to your company before changing a growth decision; use a local experiment where the paper does not match your context.
Reported result
What the study found
Perceived safety, empathy and anthropomorphism affect different components of chatbot trustworthiness; trustworthiness increases willingness to forgive after failure and reduces negative word of mouth. The paper highlights privacy concerns as a countervailing issue and finds chatbot traits work through trust-related mechanisms rather than as undifferentiated design features.
Research question
What the authors investigated
Which chatbot traits make chatbot-led service recovery effective, and through what mechanisms do they reduce negative consumer responses after service failure?
Design
The paper reports exploratory work used to select the online-retailing/delayed-order context, a scenario-based survey study and a second/robustness study, followed by a qualitative study. The quantitative model tests mediation through chatbot trustworthiness and downstream forgiveness/negative word of mouth.
Sample and setting
The main final survey sample was 586 UK consumers (304 women; median age 31.02; median income £32,000) responding to an online-retailing delayed-order recovery scenario. The exploratory work included 28 postgraduate students; the paper also reports a separate qualitative study (the publisher page contains its method/results).
Scope and transfer
Limitations to keep in view
The authors did not test contingency conditions that could attenuate the effects, call for cross-cultural comparisons, did not resolve recovery when the AI itself caused the failure, and used a service-failure scenario/context that may not generalize to other chatbot types or service settings.
Research sources
Publisher-hosted ScienceDirect full text exposes the DOI, authors, method sections, sample table, results, conclusion and limitations; this is direct evidence rather than an abstract-only record. Correction notices were not separately verified in this collection; consult the current publisher record before applying the findings.