When Automation Meets Emotion: Why Customers Still Escalate to Humans Even When Bots Work

A child wakes in the night to the sound of thunder.

The parents have explained thunderstorms before. School has explained them. The child may even know the science better than previous generations ever did. Warm air rises. Moisture condenses. Electrical charges build within clouds. The explanation for the thunder is readily available and, likely, known.

Yet the child still walks down the hallway and climbs into their parents’ bed.

The journey is not motivated by a lack of information. It is motivated by something information alone cannot provide.

For all our advances in technology, humans remain profoundly social creatures. We do not simply use other people to acquire knowledge. We use them to regulate uncertainty. We look to others to help us interpret risk, to validate our experiences, and to reassure us that our understanding of the world is shared by those around us.

This tendency becomes particularly visible in customer service environments, where organisations are investing heavily in automation while simultaneously discovering that some customers continue to seek human interaction even after receiving the answer they need.

The assumption often made is that the technology has somehow failed. The chatbot misunderstood the question. The self-service journey was poorly designed. The customer could not find the relevant information. Sometimes that is undoubtedly true.

Yet there are many occasions when the automation has performed exactly as intended and the customer still asks to speak with a person.

From an operational perspective, this can seem puzzling. From a psychological perspective, it is perhaps one of the most predictable behaviours we could observe.

The Limits of Information

For much of the twentieth century, customer service was constrained by access to information. Customers contacted organisations because the organisation possessed knowledge that they did not.

The internet altered that relationship. Search engines placed vast amounts of information within easy reach and mobile devices made that information constantly accessible. More recently, AI-powered systems have become increasingly capable of interpreting customer intent, retrieving relevant answers and guiding people through complex processes.

In many circumstances, technology now performs these tasks exceptionally well.

A customer can check a balance, track a delivery, amend a booking or reset a password without ever speaking to another human being. In most cases, that is exactly what they prefer.

This is important because discussions about automation often become distorted by an assumption that customers inherently desire human contact. Most evidence suggests otherwise. Customers generally prefer convenience. If a problem can be solved quickly and effortlessly, many are perfectly happy to avoid a conversation altogether.

The challenge emerges when the interaction is no longer solely about obtaining information.

A customer discovers an unfamiliar transaction on their account. A flight is cancelled hours before departure. A claim is rejected, a payment fails or a loved one’s account becomes inaccessible following a bereavement.

In such moments, information remains important, but information is no longer the only thing being sought. The customer is attempting to understand what has happened, what it means and whether they should be concerned.

Those questions are practical. They are also emotional.

Humans as Social Reassurance

Psychologists have long understood that human beings use relationships as a mechanism for managing uncertainty.

When faced with ambiguity, we rarely rely exclusively on objective evidence. Instead, we seek social confirmation. We ask friends for advice before making decisions we have already researched. We consult colleagues about plans we have already developed. We ask somebody else to review an email we have already written.

The purpose of these interactions is not always to obtain new information. Sometimes we are seeking confidence in the information we already possess. We look for validation.

This tendency is closely connected to the way our brains evolved.

For most of human history, uncertainty was rarely a private experience. If a rustle in the bushes might indicate a predator, the question was not simply whether you had seen it. The question was whether everyone else had seen it too. Consensus was often a proxy for safety.

Thousands of years later, the consequences have changed, but the instinct remains. A customer questioning an unexpected bank transaction is not facing a predator on the savannah, yet the psychological process is remarkably similar. They are looking for confirmation that their interpretation of events is shared by somebody else.

Understanding the environment mattered. Understanding how other people interpreted the environment mattered just as much.

As a result, uncertainty is rarely experienced as a purely intellectual problem. There is also an emotional component.

A customer may understand what a chatbot has told them, but still feels uncertain. An adviser may provide exactly the same answer as the chatbot. Yet the presence of another person changes the experience of receiving it. The answer becomes socially validated.

Someone else has looked at the situation. Someone else understands the context. Someone else is willing to stand behind the explanation.

The information itself may be unchanged, yet the customer’s relationship with that information has altered considerably. What was previously an isolated interpretation has now been shared, acknowledged and validated by another person.

Consider a customer contacting their insurer after a claim has been declined. By the time they reach the adviser, they may already know the outcome. They may have read the policy wording, reviewed the correspondence and even anticipated the decision. What remains unresolved is not the claim itself, but their attempt to make sense of it. The conversation becomes less about changing the outcome and more about understanding it and feeling heard.

Being Heard

The desire to speak with a human adviser is sometimes portrayed as a need for empathy.

While empathy is certainly part of the equation, another factor may be equally important: People want to feel heard.

This distinction matters because customers are not always seeking sympathy. Frequently, they simply want acknowledgement.

Attachment theory proposes that humans seek a “secure base” during periods of uncertainty or threat. While the theory is most often associated with childhood development, its influence extends far beyond the parent-child relationship. Throughout our lives, we continue to seek reassurance from trusted others when navigating unfamiliar or emotionally significant situations.

Viewed through this lens, a customer requesting an adviser after interacting with a chatbot may not be rejecting the technology. They may simply be seeking a secure base from which to interpret what the technology has already told them.

Imagine a customer who has been charged a cancellation fee.

The chatbot accurately explains the policy. The customer understands the policy. They may even accept that the policy has been applied correctly. Yet they still want to explain why they are frustrated.

The interaction is no longer about discovering the rules. It is about communicating an experience.

Researchers studying procedural justice have repeatedly found that people are more accepting of outcomes when they feel they have had an opportunity to express their perspective. Even when the final decision remains unchanged, the process itself influences perceptions of fairness.

This is one reason why a customer may leave a conversation satisfied despite receiving an answer they did not want.

The adviser did not necessarily solve the problem. They helped the customer feel that their problem had been understood. As automation becomes increasingly sophisticated, this distinction may become more rather than less important.

Policies are written at an organisational level, but they are experienced in individual moments, often under conditions of uncertainty or frustration. While technology may be highly effective at explaining the former, customers often seek another person to help them process the latter.

The Emotional Labour of Modern Service

There is another implication for contact centres. As routine interactions become increasingly automated, the conversations reaching advisers begin to change.

Historically, advisers handled a mixture of simple and complex contacts. Some required empathy and judgement. Others required little more than access to information.

Automation steadily removes many of the latter and the remaining interactions are often characterised by greater uncertainty, greater complexity or greater emotional significance.

Customers whose needs can be fully addressed through self-service rarely reach an adviser. Those who do make contact are increasingly likely to be navigating situations that technology alone has not fully resolved.

This creates an interesting paradox where the more successful automation becomes, the more psychologically demanding human interactions may become.

For organisations operating hybrid and homeworking models, this presents particular challenges.

When advisers worked side by side, difficult conversations were often buffered by the presence of colleagues. Support could be sought informally. Experiences could be shared immediately after a challenging interaction.

In remote and hybrid environments, an adviser may spend much of the day handling emotionally complex conversations from a spare bedroom or home office, separated from the social support that once existed naturally within the contact centre environment.

If the future role of advisers increasingly involves emotional regulation, reassurance and relationship building, organisations may need to think differently about adviser wellbeing, coaching and support.

The skills becoming most valuable may not be the ones traditionally associated with efficiency. What appears to be inefficiency in a dashboard may, in practice, be the point at which a customer stops escalating frustration and starts resolving it.

Designing for Human Needs

Perhaps the most common mistake organisations make is treating escalation as evidence of failure.

If a customer requests a human adviser after interacting with a chatbot, the assumption is often that the automated journey was unsuccessful.

Sometimes it was. And sometimes the customer simply needed something the technology was never designed to provide.

This distinction offers an opportunity of a different way of thinking about customer journeys. Rather than asking how to prevent customers from reaching human advisers, organisations might ask which moments genuinely benefit from human involvement.

Where are customers likely to experience uncertainty rather than confusion?

Where are they seeking reassurance rather than information?

Where might acknowledgement be as important as resolution?

The answers will vary across industries and customer groups, but the underlying principle remains remarkably consistent.

Human interactions create value in ways that extend beyond the transfer of information.

Contact centres have spent decades treating escalation as a symptom to be eliminated. Every transfer, callback and adviser-assisted interaction is typically viewed as a source of cost and inefficiency.

Yet if customers sometimes seek human interaction for reasons that are psychological rather than informational, escalation may not always represent a failure in the customer journey. In some circumstances, it may represent the successful completion of a different kind of journey altogether.

More Than Answers

The child who climbs into their parents’ bed during a thunderstorm is not searching for a meteorological explanation. They are seeking reassurance in the presence of uncertainty.

Most adults like to imagine they have outgrown such instincts – but the evidence suggests otherwise.

We continue to seek reassurance from colleagues before making decisions. We continue to seek validation when experiences feel unfair. We continue to look towards other people when the consequences of being wrong feel significant.

Customers are no different.

As automation becomes increasingly capable, organisations will become ever more effective at delivering answers. The question facing customer service leaders is whether answers alone are always what customers need.

Because when automation meets emotion, the customer’s decision to seek another human being may have very little to do with the quality of the technology and a great deal to do with the psychology of being human.

References:

Ainsworth, M. D. S., Blehar, M. C., Waters, E., & Wall, S. (1978). Patterns of attachment: A psychological study of the strange situation. Lawrence Erlbaum.

Bowlby, J. (1988). A secure base: Parent-child attachment and healthy human development. Basic Books.

Cialdini, R. B. (2009). Influence: Science and practice (5th ed.). Pearson.

Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117–140. https://doi.org/10.1177/001872675400700202

Folger, R. (1987). Distributive and procedural justice in the workplace. Social Justice Research, 1(2), 143–159. https://doi.org/10.1007/BF01048057

Lind, E. A., & Tyler, T. R. (1988). The social psychology of procedural justice. Plenum Press.

Shannon, C. E. (1948). A mathematical theory of communication. The Bell System Technical Journal, 27(3), 379–423. https://doi.org/10.1002/j.1538-7305.1948.tb01338.x

Sillence, E., Briggs, P., Harris, P. R., & Fishwick, L. (2007). How do patients evaluate and make use of online health information? Social Science & Medicine, 64(9), 1853–1862. https://doi.org/10.1016/j.socscimed.2007.01.012