When the script runs out: customer service beyond the SOP

A customer rang about a delivery that had been promised twice and arrived neither time. The agent followed the process correctly: apologised, checked the system, confirmed the new date, offered the standard resolution. The customer got angrier with each step, and eventually asked to speak to someone who could actually do something. The agent, accurately, said there was no one else.

Nothing in that call was done wrong. The agent handled it exactly as trained. It still ended with a customer who will tell people about it, because the script covered the situation and did not cover the person.

The twenty per cent that makes the reputation

Most customer interactions are routine and a process handles them well. The calls that shape what customers say about a business are the exceptions — the ones where something has already gone wrong, where the customer is upset before the conversation starts, or where the correct answer is one they do not want.

Training tends to concentrate on the eighty per cent, because it is easier to specify and easier to assess. The twenty per cent is where judgement is required, and judgement is what people are least often given permission to use.

Acknowledge first, explain second

The most reliable single change in service conversations is the order of two things almost everyone gets the wrong way round.

EXPLANATION FIRST  (reads as excuse)
  "The delay happened because the warehouse had a
   systems issue over the weekend, so your order
   didn't get picked. I'm sorry about that."

ACKNOWLEDGEMENT FIRST  (reads as understanding)
  "You were promised this twice and it hasn't arrived
   either time. That's genuinely poor and I'd be
   frustrated too. Here's what happened, and here's
   what I'm going to do about it."

SAME FACTS. DIFFERENT CONVERSATION.

The reason is straightforward. An upset customer is not yet ready to process information; they are waiting to find out whether the person on the other end has understood. Explanation delivered before that point is heard as defence, however accurate it is, and it tends to escalate rather than settle.

Acknowledgement also has to be specific to be credible. "I apologise for the inconvenience" is a phrase customers have heard so often that it registers as the absence of an apology. Naming the actual thing — promised twice, arrived neither time — demonstrates that someone listened, which is most of what is being asked for.

Authority is a training issue

A great deal of what is diagnosed as poor service is in fact an authority problem wearing a service costume. An agent who cannot waive a charge, cannot authorise a replacement, and cannot commit to a date has only one tool, which is sympathy, and customers correctly perceive that as insufficient.

• Give a defined discretionary limit. A value or a set of remedies the agent can apply without asking. The amount matters less than its existence; an agent who can do something behaves differently from one who cannot.
• Say what is outside it and who decides. Vague limits get used cautiously, which wastes them. Clear ones get used well.
• Do not punish reasonable use of it. If every exercised discretion is reviewed, it stops being exercised within a month and the training is undone.
• Track what it gets spent on. Recurring use for the same underlying cause is telling you about a broken process rather than about generous agents.

The last point is the one that connects service training to the rest of the business. Where the same complaint arrives repeatedly, no amount of customer service skills training will resolve it, because the skill is being used to absorb a failure that keeps happening upstream.

Saying no without losing the customer

The hardest conversation in service is the one where the customer's request is reasonable and the answer is still no. Most people handle it in one of two poor ways: soften it until the customer believes there is still room, or deliver it flatly and end the conversation.

1. 1

   Say the no clearly and early

   Buried refusals produce a second call. The customer should not have to work out from context that they are being declined, and drawing it out is experienced as evasion rather than kindness.

2. 2

   Give the actual reason, once

   Not policy as an abstraction. "We can't replace it after ninety days because we no longer have the batch" is a reason. "That's our policy" is a refusal to give one, and it is what makes people ask for a manager.

3. 3

   Offer what you can do

   There is nearly always something — a partial remedy, a different route, an escalation with a realistic timeline. Moving to this quickly changes the conversation from adversarial to practical.

4. 4

   Do not apologise repeatedly for the no

   Repeated apology signals that the decision might be negotiable. One clear acknowledgement, then the alternative.

What to measure, and what measuring the wrong thing does

Average handling time is the most commonly tracked service metric and among the most damaging when it is the primary one. It trains people to end conversations, which is a different objective from resolving them.

The visible effect is repeat contact. A call closed in four minutes that generates a second call two days later has cost more than a seven-minute call that finished the issue, and only the first appears in the metric as good performance.

Better measures exist and are not much harder to collect: whether the issue recurred within a fortnight, the proportion resolved without escalation, and what the recurring reasons for contact actually are. The last is the most valuable and the least collected, because it points at the product or the process rather than at the agent.

Internal customers, same rules

The same skills apply to service between teams, where they are almost never trained. HR responding to a payroll query, IT handling a request, finance declining an expense — each is a service interaction, and each tends to go badly for exactly the reasons above: explanation before acknowledgement, policy quoted instead of a reason given, and no clarity about who can decide.

It is worth including internal-facing teams in service training for that reason. The volume of internal frustration in most businesses is substantial and largely invisible, because nobody surveys it and nobody escalates it to a manager the way an external customer would.

Practising the difficult twenty per cent

Service training that consists of principles produces people who can describe good service. Practice produces people who can deliver it under pressure, which is a different capability and the only one that matters at the moment the call goes wrong.

The practical version is to collect the ten hardest real situations from the last quarter — not invented scenarios — and work through them in a room. What should the first sentence be. Where does this one go wrong. What would you have needed to be allowed to do. Those conversations surface both the skill gaps and the authority gaps at once, and the second list is usually the more useful of the two. That combination of practice and permission is what customer service skills work has to deliver to change anything, and it is why sessions built on generic scenarios so reliably fail to transfer.

Where the same ten situations keep recurring quarter after quarter, the honest conclusion is that the problem is not in the conversation at all. That is a process finding, and an HR process audit or an equivalent look at the upstream operation will do more for customer experience than another training day — a judgement worth making explicitly before commissioning one, which is the first thing training consulting should be asked to test.