Take a complaint that has been escalated. The customer is angry, and by the time it reaches the complaints team they have probably been in touch more than once about the same thing.
The investigation is thorough. Someone pulls the records and looks at each interaction in turn.
On the first contact, the adviser gave information that turned out to be incomplete, but it matched what the knowledge base said, so the adviser followed the process. On the second, a callback was promised within 48 hours and happened on day four. The notes put it down to a backlog that week, and the customer received an apology. On the third, the customer came across as angry and impatient. The adviser handled it calmly and in line with policy, and the case was passed to another team.
Each finding looks reasonable. The complaint is partially upheld, a goodwill gesture is offered for the late callback, and the file is closed.
From the organisation’s side, three minor things happened.
From the customer’s side, one thing happened, and it went on for three weeks.
Judged one at a time
This can happen even where people care and are competent. Sometimes it’s simply what happens when a service is examined the way it is organised.
Quality assurance often samples individual calls and scores them. Teams are measured on their own part of the process. Complaint procedures frequently ask whether each interaction was handled correctly, which is a fair question and one that can be answered with evidence. None of that is unreasonable. But it is entirely possible for nobody in that arrangement to be asked whether the customer got what they came for.
So each interaction is judged against its own standard, and each one can pass. The first adviser followed the knowledge base. The callback team had a backlog. The adviser who took the third call handled an angry caller well.
Three passes, and one customer who still doesn’t have what they needed.
I’m not sure that counts as a success, however good the individual scores look.
Minor depends on what came before
On its own, a late callback may be a minor irritation.
A late callback that arrives after you have already been given incomplete information is something else. By then you are not really waiting for a call. You are waiting to find out whether anyone has actually got hold of your problem, and the missed call answers that question.
The third interaction then lands on top of both. The customer is no longer reacting only to that conversation. They are reacting to three weeks, a broken promise and the effort of explaining the whole thing again to someone who is seeing it for the first time. If the organisation only looks at the third call, the customer looks disproportionate. If it looks at all three, the reaction looks fairly predictable.
That matters, because angry and abusive are not the same thing. Someone who is short, exasperated or openly fed up after three weeks of this is not abusing anybody. Threats and personal insults are a different matter, and staff shouldn’t have to put up with them. But if every raised voice gets filed under customer behaviour, the organisation has quietly moved the problem onto the person who has been on the receiving end of it.
Three small errors, or one cause
This is the part that interests me most, because assessing interactions separately doesn’t just understate what the customer went through. It can change where the cause appears to be.
Look at the three contacts one at a time and you find three small, local issues: incomplete information, a team that fell behind, a customer who lost patience. Each suggests a local fix. Update the knowledge base, add capacity, remind staff of the policy on difficult calls.
Look at them together and different questions appear.
Why did the customer need a callback at all? Possibly because the first adviser couldn’t see what was needed to resolve it and had no way to pass it on other than promising a call. Why did the third adviser have to start from scratch? Possibly because information from the earlier contacts wasn’t recorded in a way the next person could use. And was the backlog really just a backlog, or was some of it made up of customers like this one, calling back about things that hadn’t been sorted the first time?
Those aren’t conclusions. They’re questions that the combined history makes worth asking.
If the evidence supported them, that wouldn’t be three errors. It might be one decision, made somewhere upstream, producing all three. Fix each interaction separately and that decision stays exactly where it was, ready to do the same thing to the next customer.
Sometimes three failures really are three unrelated things that happened to land on the same person. But you can’t tell which it is unless you look at them together, and not every organisation is set up to.
That isn’t necessarily because nobody inside the organisation is capable of seeing it. The problem may cross the boundaries the organisation uses to manage itself. Different teams own different parts of the journey, different measures tell them whether their part is working, and previous decisions have become part of how things are done. Looking across all of that can be harder from inside it, particularly when each individual part appears to be doing what it is supposed to do.
Who notices?
The question I keep coming back to is how the organisation would know this was happening at all.
If the only place accumulated failure becomes visible is a complaint, then the customer has become the mechanism for detecting it.
The organisation finds out only when somebody is fed up enough to put it in writing, and only about the people who do. Everyone who went through the same three weeks and gave up simply doesn’t appear.
One problem
Organisations tend to be good at asking whether something was done correctly. They are not always as well set up to ask whether it worked.
Three interactions can each be handled correctly and still add up to one problem nobody has handled.
The customer never experienced them as three.
About the Author
Karen Ferguson investigates why problems in human-facing services keep happening. Her background is in human behaviour and communication, and she has designed, built and adversarially tested conversational AI systems. Problems she has identified and pursued have led to documented process changes in an NHS trust, a police force and a national charity.