What to Do When a Job Goes Wrong

There is a piece of advice that circulates in every small business group, and it is wrong in a way that costs money.

The advice is that a complaint handled well leaves the customer happier than if nothing had gone wrong at all. Fix it brilliantly and you come out ahead. Problems are opportunities.

This idea has a name. It is called the service recovery paradox, and it has been studied for more than thirty years. The research is a lot less encouraging than the advice.

In the study that gave the idea its most careful test, researchers Stefan Michel and Matthew Meuter segmented retail banking customers by whether they had experienced a failure and how well it was handled. Overall satisfaction was consistently highest among the customers who never had a problem. The paradox showed up only in a narrow case: when the original service had been rated poorly to begin with, so the recovery had a low bar to clear. Across the broader literature they reviewed, roughly half the studies denied the paradox exists at all. Their own title called it "true but overrated."

The practical translation for a service business is short. A great recovery does not put you ahead of the version of the job where nothing broke. What it does is keep a customer you were otherwise going to lose, and prevent a review you were otherwise going to get. That is worth a great deal. It is just not an upgrade, and treating a bad job as an opportunity is how people end up spending too much fixing it.

So here is what to actually do.

The first hour matters more than the fix

You do not know yet what the right remedy is. You do know the customer is currently deciding whether you are the kind of business that disappears.

Acknowledge it before you understand it. A text within the hour that says "I heard, I am on it, I will call you by five" costs nothing and buys you the entire rest of the process. The temptation is to wait until you have an answer so you only have to call once. That instinct is backwards. Silence while you investigate reads identically to silence while you avoid.

The expectation data here is old but directionally useful. A GatherUp survey of 480 US adults, run back in 2016, asked how long is reasonable for a local business to respond and resolve a complaint. The largest group, 38 percent, said responding within one day mattered. Almost 18 percent expected a few hours or sooner. Those numbers have not moved toward patience since 2016.

Do not explain before you acknowledge

This is the single most common unforced error, and it is not about tone. It is about order.

The supplier sent the wrong part. The previous installer did it wrong. The weather moved. All of these may be completely true and none of them belong in the first sentence. An explanation offered before an acknowledgment reads as a defense, and the customer stops listening to everything after it.

The order that works: what happened, that it is not acceptable, what you are doing, when. The reason, if it still matters, goes at the end or nowhere. Most of the time the customer does not want the reason. They want to know it is handled.

Decide the remedy yourself, and decide it once

Asking "what would make this right?" sounds generous and often makes things worse. It hands the customer a blank check they did not ask to hold, and in a lot of cases they will name something larger than what you would have offered, because they now think they are negotiating.

Name the fix. Redo it, refund the portion that failed, credit it, or come back Thursday with the right part. Then do that thing.

The research has something specific to say here. Michel and Meuter noted that recoveries satisfying enough to produce genuine delight are expensive and difficult to manage, and that standardized recoveries cannot generate the surprise that delight requires. In other words: the big generous gesture is not reliably buying you loyalty, it is just expensive. There is a whole line of research on avoiding overcompensation for exactly this reason.

Proportional and immediate beats large and slow. Almost every time.

Redo versus refund

Rough rule: redo when the work is fixable and the relationship is intact. Refund when the customer no longer wants you in the house.

The mistake is pushing a redo on someone who has already decided they are done with you. You get a second chance to fail in front of a person who is now watching for it, and if it goes sideways the second time you have converted a recoverable job into a certainty of a review. When someone has emotionally exited, the cheapest outcome available is a clean refund and a courteous exit.

The reverse mistake is refunding reflexively on a job you could have fixed in an hour. That buys silence, not a customer.

The window closes when they post

Everything above assumes you still have a private conversation. Once a review is posted, the audience changes from one person to everyone, and so does the playbook. That is a different job, and we wrote it up separately: how to respond to a bad review.

Which is the real argument for speed. The recovery window is the gap between the thing going wrong and the customer deciding to tell people about it. That gap is usually measured in days. Sometimes hours. Every hour you spend assembling the perfect response is an hour spent inside a closing window.

Write it down after

One bad job is an incident. The same bad job twice is a process problem, and the service recovery research is clear that whatever benefit recovery has does not survive you doing it again.

So keep a note. What failed, why, what you did, what it cost. Four lines. After a year you will have a list, and the list will be short and repetitive, which is the useful part. Most service businesses fail in two or three ways, over and over, and nobody ever writes the ways down.