The Art of
De-Escalation

Imagine a customer whose account has been restricted after an unusual activity alert. They need access today. The security rule exists because fraud and impersonation are real, and the agent cannot safely waive verification. The customer cannot complete the required check before tomorrow. Both the need for protection and the customer’s immediate loss are real. This is an imagined case, used here to follow the book’s argument. It is not an example taken from the manuscript.

At first, the agent searches for the missing option. Perhaps another form of verification will work. Perhaps somebody with more authority can help. That search is responsible. But if the available options are exhausted, a better tone cannot produce an option that the system does not allow. The agent has reached the edge of the design.

This is what the book calls a no-win situation: not merely an unpleasant conversation, but a collision between legitimate claims or constraints that cannot both be satisfied in the current arrangement. It does not follow that the security boundary is unnecessary, or that every refusal is an injustice. It means that the cost of the boundary is real even when the boundary has a defensible purpose. The book traces this pattern beyond customer service, into operations, compliance, trust and safety, healthcare and other work where someone must deliver the limits of a system they did not design.

Once the decision cannot change, the work changes. Someone must explain what will happen, acknowledge what will not happen, and stand beside an outcome they did not design. The organisation may record a closed case. The customer still cannot use the account. The agent still had to be the voice of the decision. As the book puts it: “Some situations do not have a right answer. They only have an ending.”

“Let me escalate this” sounds like a promise of wider possibilities. Sometimes it is. The book asks us to look at what an escalation route is actually authorised to do. In a scaled organisation, a higher decision-maker may have to protect consistency, legal obligations and the risks created by making an exception. That can narrow the available outcome rather than expand it.

In our imagined case, a specialist reviews the restriction and confirms that the check cannot be waived. The specialist has the authority to decide. The agent is asked to tell the customer. The decision travelled upward; its explanation and the customer’s reaction travelled back down. An appeal may still be meaningful, but the existence of a path does not tell us whether it can reconsider the underlying trade-off, how long it will take, or what evidence the customer can reasonably provide.

From the outside, the escalation may look like diligence. For the person waiting, it can be the closing of the last door they thought might open. For the agent, it adds the weight of finality to a conversation they were already carrying.

The question is therefore more precise than “Was the case escalated?” Who had the authority to change what happened? Who decided that the boundary was proportionate? Who is exposed to its effects? If those people never have to encounter one another’s evidence, escalation can confirm a decision without connecting its makers to its consequences. That is a design of ownership, even when nobody deliberately set out to evade responsibility.

The agent delivers the decision accurately and without cruelty. They do not make a promise they cannot keep. They listen to the customer’s distress and remain composed. That takes judgement. It is also the point at which the organisation may conclude that the situation was handled well.

The book is interested in the gap between those two statements. The interaction may have been handled with care, and its underlying conflict may still be intact. When composure becomes the principal evidence of success, the person delivering the outcome must manage more than words. They decide how much to acknowledge, when explanation becomes false hope, and how to be honest without making the loss sharper than necessary. None of that changes the customer’s position. All of it costs attention.

Repeated refusals can alter how a person works. They may learn to keep conversations shorter because open ones rarely change the answer. They may sound more cautious in a meeting because they have seen concerns go nowhere. A manager may read that restraint as declining engagement and coach the person’s attitude or tone. Adaptation to repeated moral conflict is another possible reading. It will not explain every quiet colleague, but a performance review focused on behaviour and output will struggle to see it at all.

This is why the author challenges the ease with which “burnout” becomes the explanation. Rest and support may be necessary, but a person may also be struggling with the repeated demand to represent decisions they cannot reconcile with their sense of fairness. The book calls attention to moral injury in that pattern. It is making a structural argument, not diagnosing any particular worker. The unsettling part is that a person can continue to perform well while the work changes what they feel able to say, notice or carry home.

The account restriction is confirmed. The call ends calmly. The case is marked resolved. Perhaps the customer does not call again. Every one of those records may be accurate, yet none establishes that the customer could access their account, understood the reason for the boundary, or still trusted the organisation.

Silence is especially ambiguous. A customer who does not return may be satisfied, may be waiting, or may have given up. A worker who stops escalating may have found a better way to resolve cases, or may have learned that escalation changes nothing. The same lower contact or escalation count can accompany very different realities. Calm may signal care. Alone, it cannot prove that harm was avoided.

The book follows the unresolved outcome into three places. The customer carries a loss or a story that does not make sense. The agent carries the memory of delivering a final answer without being able to repair what it cost. The organisation carries a record flattened into a status code. When the customer leaves later, when the worker becomes guarded, or when a similar restriction produces another no-win case, those events may be analysed separately. The connection has been lost. The case is closed. The work is not.

This is one of the book’s most useful shifts in perspective: “resolved” is a statement about the organisation’s process. It is not, on its own, a statement about the person’s experience or the fairness of the underlying arrangement. A technically correct measurement can mislead precisely because it records one part of the truth so cleanly.

There is nothing wrong with helping a distressed person understand a difficult outcome. A calmer conversation can make room for explanation and preserve dignity. The book draws a line at the purpose of the calm. Does it help the person encounter the truth without further injury, or does it help the system avoid seeing the truth at all?

Suppose the agent acknowledges the customer’s urgent need, explains that access cannot be restored before verification, and is clear about what remains uncertain. The customer may still be angry. The interaction may still count as difficult. But the person has not been asked to pretend the loss is trivial, and the agent has not claimed a resolution that does not exist.

Now imagine the same decision delivered in perfectly soothing language that hurries the customer toward acceptance, implies that the situation is taken care of, and steers them away from the impact they are trying to explain. The tone may earn an excellent score. The customer’s experience has been made less visible. In the book’s terms, de-escalation has moved from reducing secondary harm toward silencing.

This is a distinction about what remains true in the exchange, not a script for the agent. Language can prevent additional harm. It cannot repair the loss created by a rule or make the underlying decision fair. Trust in a no-win situation may not mean satisfaction or agreement. It may mean that the person was told the truth and was not treated as a problem to be managed. If the organisation calls that an unsuccessful interaction because the customer remained upset, its definition of success deserves scrutiny.

The book’s interlude turns this question toward the reader with authority. If you have praised someone for handling an impossible situation beautifully, what happened to the conditions that made it impossible? Admiration can be sincere and still leave the burden exactly where it was.

An agent dealing with repeated no-win cases may need time off, support or counselling. The book does not ask an organisation to withhold any of these. It asks what happens when the availability of support becomes proof that the organisation has done enough.

If the same role repeatedly requires recovery after enforcing the same constraints, the pattern says something about the work. Treating each recovery as a private wellness event can hide that signal. The person is helped to return to conditions that keep producing the need for help. Compassion reaches the consequence while the decisions about thresholds, authority and escalation remain elsewhere.

This distinction matters in our example. Supporting the agent after a difficult call is appropriate. It does not settle whether the verification route makes reasonable provision for people who cannot complete it promptly, whether appeals can correct mistakes, or whether those who set the rule understand how its cost is distributed. Those questions cannot be assigned to the person delivering the refusal.

The book does not promise that redesign will eliminate every difficult case. Some boundaries must hold. Its demand is for the people with design authority to remain in contact with the consequences of those boundaries. Care for the worker and scrutiny of the system belong together.

When the same collision recurs, the book refuses to call every instance an unavoidable edge case. The frequency of that collision reflects decisions about thresholds, exceptions and which kinds of friction the organisation is willing to tolerate. The discomfort of examining those choices belongs where the choices can be changed.

Imagine that the account restriction is now explained first by an automated assistant. Its response is clear, polite and immediate. The human agent appears only after the customer has already tried the available paths. They cannot alter the restriction; they can only interpret it and absorb the frustration that remains. The organisation sees fewer long conversations and perhaps fewer escalations. It cannot infer from those figures alone that fewer people were harmed.

The book recognises why systems are built this way. Fraud and abuse are real. Consistent verification can protect people. Automation can spare workers repetitive exposure to conflict. The concern is what happens when choices made for legitimate risk control become very easy to apply broadly, very difficult to contest and almost effortless to deliver with perfect composure. A rule designed around the worst case may come to treat an ordinary person as a risk by default.

An appeal can look like a human safeguard while asking the least powerful person to find time, persistence and evidence to challenge a decision they may not understand. The worker who questions the output may be asked why they want an exception to a system presumed correct. In this arrangement, the machine did not invent the trade-off. It made the organisation’s existing priorities faster, calmer and more scalable.

The later chapters take this beyond a chatbot. A boundary can be built into eligibility rules, routes that are hard to reach, or options removed before a request can be made. No one has to say “no” aloud. The person still meets the limit. The absence of an argument then becomes an easy, unreliable measure of success. As routine conflict is intercepted, the remaining human cases may be fewer but heavier: the person at the edge must preserve trust in a decision they cannot change and may not be able to explain fully.

This is the book’s forward drift, drawn from design choices it says are already visible. It is not a prediction that every automated service will work this way. It asks what kind of responsibility survives when a decision can be delivered without anyone having to witness its cost.