Not yet human-reviewed
This source record was written automatically by the research pipeline and has not yet been reviewed by the maintainer or integrated into a topic page. Treat it as a provisional summary and verify claims against the original source.
Engaging Customers with AI in Online Chats: Evidence from a Randomized Field Experiment
Citation. Zhang, S., & Narayandas, D. (2026). Engaging customers with AI in online chats: Evidence from a randomized field experiment. Management Science, 72(1), 73–95. https://doi.org/10.1287/mnsc.2022.03920
Summary. In a randomized field experiment at a meal-delivery company, giving customer-service agents AI-generated suggestions improved response speed, depth of engagement, and customer sentiment. Benefits were strongest for less-experienced agents and varied by conversation type; after chatbot comprehension failures, unusually rapid AI-assisted replies could reduce sentiment by making customers think they were still interacting with a chatbot.
Key claims
- AI-assisted agents responded faster, engaged customers more deeply, and achieved greater improvements in customer sentiment than agents without AI suggestions.
- Benefits were strongest for less-experienced agents, and effects differed by conversation type: AI improved cancellation interactions but was least effective for repeat complaints involving systemic issues.
- After a chatbot comprehension failure, transfer to an AI-assisted human agent could lower customer sentiment because rapid responses led customers to infer they were still speaking with a chatbot.
Evidence & limitations
- The study uses a randomized field experiment in one meal-delivery company’s online customer-service interactions; results identify causal effects in that setting but may not generalize to other service designs, AI systems, or organizational contexts.
- The retained Jina artifact contains the publisher landing page and abstract rather than the full article text; this record limits claims to the abstract and Crossref metadata.