68/100

Moderate evidence

Customers still wanted a human after a chatbot fixed their complaint, a 200-person study suggests

Study title: Chatbot or Human? Rethinking Complaint Handling in Customer Service

Evidence type
Controlled or validated comparison
Sample
n = 735
Sources
1 sourcePeer-reviewed
Published
2 min read

Study title: Chatbot or Human? Rethinking Complaint Handling in Customer Service

Four pre-registered experiments tested when consumers accept chatbots for emotionally charged complaints. In the first, 200 university participants preferred human agents to resolve complaints, even after a chatbot had already resolved their issue.

Read the study abstract

ABSTRACT AI chatbots face major challenges in managing emotionally charged customer complaints. While they offer scalability, speed, and unlimited availability, they are perceived to be impersonal, particularly in service recovery situations that require empathy. This research examines when and how consumers accept chatbots in complaint‐handling contexts and how specific design attributes influence that acceptance. Across four pre‐registered studies conducted with student, professional, and online (Prolific) con…

The problem

In a study of 200 university participants, consumers preferred human agents to resolve complaints even after a chatbot had successfully resolved their issue.

In brief

Later studies point to ways of improving acceptance. Studies with 75 professionals and 300 online consumers found that faster response time significantly increased chatbot acceptance, and a fourth study of 160 university students found that early, user-initiated handovers to a human were preferred over delayed ones or those triggered by frustration.

The authors conclude that efficient handling can coexist with relational disengagement, and they offer guidance for designing hybrid service systems that combine chatbots and human agents.

Main points

  1. Study 1 involved 200 university participants and suggests a preference for human agents even after successful chatbot resolution.
  2. Studies 2 and 3 used 75 professionals and 300 online consumers (Prolific) and found faster response time raised chatbot acceptance.
  3. Study 4 had 160 university students and examined chatbot-to-human handovers.
  4. Early, user-initiated transfers were preferred over delayed transfers or those triggered by frustration.
  5. The research used between-subjects and discrete-choice experiments testing response speed, autonomy and escalation dynamics.
  6. The authors say procedural efficiency can coexist with relational disengagement in chatbot-handled complaints.

The full story

The research was published in the Journal of Consumer Behaviour in June 2026. It spans four pre-registered studies using student, professional and online consumer samples, combining between-subjects designs with discrete-choice experiments.

Beyond chatbot versus human preference, the experiments tested chatbot attributes including response speed, autonomy and escalation dynamics. Speed emerged as a significant factor in whether consumers accepted a chatbot.

In the handover study, participants favoured transfers to a human that they initiated early over delayed handovers or ones triggered by their own frustration. The authors say this points to hybrid service designs and extends justice and coping theory beyond human-centred assumptions.

Sources

1 on file
  1. Peer-reviewed2026-06-12

Talk it over

Four questions to start on
Who carries this cost, and what would have to change to prevent it?

Start with a question

Customers still wanted a human after a chatbot fixed their complaint, a 200-person study suggestshttps://truvace.com/item/chatbot-or-human-rethinking-complaint-handling-in-customer-service

  1. Who carries this cost, and what would have to change to prevent it?

    The problem In a study of 200 university participants, consumers preferred human agents to resolve complaints even after a chatbot had successfully resolved their issue.

    Answer in the thread
  2. How do we know? What can this kind of evidence show, and what can’t it?

    The evidence This rests on a controlled comparison with a sample of 735, from Journal of Consumer Behaviour.

    Answer in the thread
  3. What does it leave open? What else would you want to know before acting on it?

    The record’s own caveat The studies used samples of university students, professionals and online consumers, ranging from 75 to 300 participants, and tested preferences in experimental settings rather than real company complaint systems.

    Answer in the thread
  4. Have you seen AI used this way where you study, work or live? Did it go the same way?

    Answer in the thread

The questions are the same on every record. The details under them are taken from this record as written.

What readers say