What do you do when an employee keeps making the same mistake?

Start by checking whether you ever built a way for the person to find out they were wrong. A repeating mistake is a shared problem: no feedback loop makes it yours, a working loop that changes nothing makes it theirs. Either way, you are the one who has to do something about it.

Why does the same mistake keep coming back?

There is a legend that gets passed around leadership circles about Thomas J. Watson Sr. at IBM. An executive makes a decision that costs the company an enormous amount of money, gets summoned to Watson's office, and arrives expecting to be fired. Watson looks at him and says something like, "Fire you? I just spent a fortune training you."

I like the story. I also can't tell you what it cost, because every version I've found puts a different number on it. Six hundred thousand. Two million. Ten million. When a figure moves that much between tellings, you're holding folklore rather than history, and I'd rather say so than pretend I know.

The principle underneath it still stands, and it's the reason the story survives. A mistake is expensive whether or not you learn anything from it. You have already paid. The only remaining question is whether you collect on what you bought.

Which is fine for a mistake. It falls apart for a pattern.

What's the difference between a mistake and a pattern?

Almost anything can be forgiven if the next one is different.

A person who makes a wide variety of mistakes is usually a person doing real work at the edge of what they know. That's what learning looks like from the outside, and if your business has no tolerance for it you will end up with a team that only does what it has already done. Errors are the cost of anyone growing into a job.

The same error, arriving on a schedule, is a different animal. It isn't range. It's a loop. And a loop means the correction is not reaching the place where the decision gets made.

So the question stops being "why is this person careless" and becomes "what conditions produce this error reliably." Those are not the same question, and only one of them has an answer you can act on.

Who does a repeated mistake actually belong to?

Most owners want a clean verdict at this point, and there isn't one. It's shared. But it's shared in a specific order, and the order matters.

Start with your side of the ledger, honestly:

• Did you onboard them properly, or did they get a login and a general sense of where things are?
• Do they have the training the job actually requires, as opposed to the training you assume anyone would have?
• Is there a feedback loop, or is feedback something that happens when you're annoyed?
• Are there real reviews, on a schedule, that they know are coming?

If you're getting angry about repeated mistakes and you have not built any of that, the cycle is yours. You are watching a problem you constructed and calling it a personality flaw. That's not a hard judgment; it's just arithmetic. Nobody self-corrects toward a standard they were never given.

Now the other side. If the loop exists, if you have sat down with this person more than once, if the standard has been said out loud and in plain language, and the same mistake still shows up, you are no longer dealing with a communication gap. You're dealing with teachability.

When I was a pastor I had a rule about volunteers, and twenty years later I would say exactly the same thing about employees. Two characteristics matter more than any skill on the résumé. They have to be teachable, and they have to work well on a team. Everything else can be built. Those two either exist or they don't.

Repeating the same mistake after real feedback is the clearest available evidence that someone is not teachable. That's a serious problem, and it doesn't get better with a fourth conversation.

Even then, the action is yours. Shared responsibility for the cause; sole responsibility for the response. Nobody else in that building can make the call.

What does a feedback loop look like in a business with eleven people?

A friend of mine runs a construction company and checks in with his crews at the end of every day. Where did the project get to, what moved, what didn't. He knows his numbers at a level that mildly alarms me. He can tell you roughly how long it takes to install casing around a door, so he knows within a few minutes whether a crew is ahead or behind, and he knows it that day rather than at the end of the job.

I think a daily cadence is a bit much for most businesses. But he has something almost nobody in his industry has, which is information arriving fast enough to be useful.

The version I'd actually recommend has four layers, and none of them requires software:

Weekly. A short stand-up on Monday. What happened last week, what we're improving this week. Ten minutes, every week, whether or not anything went wrong.

Immediately. When something goes sideways, you take time out of your day for it. Not a lecture, and not an email. A conversation, close to the event, while the details are still sharp.

Quarterly. A real sit-down with each person who reports to you. An hour. On the calendar in advance so it isn't a surprise or a threat.

Annually. A review with no surprises in it. This is the test of whether the other three worked. If someone learns something new about their own performance at their annual review, the review isn't the problem. The eleven months before it were. And compensation should attach to the quarterly and annual conversations, because a standard with no consequence attached is a preference.

Gallup's research on this is blunt. Employees who get feedback from their manager a few times a week or more are engaged at roughly twice the rate of those who hear from them a few times a year, and only about a quarter of employees strongly agree they receive valuable feedback from the people they work with. Valuable is doing a lot of work in that sentence. Frequency alone is surveillance. Frequency plus something worth hearing is a loop.

How do you actually run the conversation?

Blanchard and Johnson's The One Minute Manager has been in print since 1982 and most of what people remember from it is the compliment sandwich, which I don't particularly love. Praise, criticism, praise, and everyone in the room knows the first slice was structural.

What I do use is the shape underneath it, and it goes roughly like this. Name the thing that happened, plainly. Say why it's a problem. Say what it cost, in money or time or reputation, and say how it landed on you personally. Then stop talking and let them respond. Then move on, and actually move on.

The part people skip is the cost. "You made a mistake" is abstract and easy to absorb without changing anything. "That one cost us the callback and I had to phone the client myself on a Saturday" is a fact with weight in it, and weight is what gets remembered.

The other part people skip is moving on. If the conversation happened, it happened. Bringing it up again three weeks later teaches the person that correction is permanent and the safest move is to hide the next one.

Where does the architecture come into it?

Everything above is management practice, and management practice is not the whole of it.

Underneath a repeated mistake there is almost always a system that either doesn't exist or isn't being lived by. Those two failures look identical from the outside and they need different fixes.

If the system doesn't exist, the mistake is being manufactured. There's no written standard for how an estimate gets checked before it goes out, so every estimate is checked according to whoever happened to touch it. The error rate isn't a people problem; it's the predictable output of a process nobody designed. You will hire and fire your way through several good people before that changes.

If the system exists and nobody lives by it, that is usually a leadership problem and it usually starts at the top. The founder who wrote the checklist and then skips it when he's in a hurry has taught the whole company that the checklist is for other people. You cannot delegate a standard you don't keep. That's not a moral point, it's a mechanical one. People do what they see done.

This is the part I'd sit with longest if I were you. Most of the founder-led businesses we work with have never built the four systems a company actually runs on, and they never noticed, because the founder was the system. He was the qualification standard, the quality check, the pricing authority and the final word on everything. It worked beautifully at four people. At eighteen it produces exactly the symptom this post is about, which is a business full of people guessing at criteria that live in one person's head.

You don't fix that by adopting somebody's framework. You fix it by building your own version of each one, for how your business actually works. Chess has principles. It doesn't have a formula.

Putting it to work

Take the last repeated mistake in your business and run it through three questions in this order. The order is the whole exercise.

One: was the standard ever stated out loud? Not implied, not demonstrated, not "obviously." Said, in words, to this person, in a moment when nothing had gone wrong. If the answer is no, you have your answer and it isn't about them.

Two: how would they have found out they were wrong? Trace the actual path. If the only mechanism is you noticing and getting annoyed, there is no loop, there is just weather.

Three: has anything changed since the conversation? Not compliance for a week. Change. If you've said it clearly, more than once, with the cost attached, and the pattern is intact, you're looking at teachability and you should stop treating it as a training problem.

Then borrow the question I used on every young volunteer I ever supervised: are you thinking like a parent? It was shorthand for a whole standard nobody could have written down in a manual. Peer behaviour is fine among peers and wrong when you're responsible for someone. One question carried more than a rulebook would have, because it made the person do the reasoning instead of memorising the rule.

Find your version of that question for the role where the mistakes keep happening. A standard someone can apply to a situation you didn't anticipate is worth more than a procedure for the situation you did.

Sources

• Organizations Can Redefine Feedback by Including Recognition, Gallup
• This Fixable Problem Costs U.S. Businesses $1 Trillion, Gallup
• Kenneth Blanchard and Spencer Johnson, The One Minute Manager (1982), cited at book level for its correction structure.

Frequently asked

How many times should I address the same mistake before letting someone go?

There's no fixed number, and anyone who gives you one is selling something. What matters is whether real feedback has happened: the standard said plainly, the cost named, the conversation had more than once, on separate occasions, with time in between to change. Once that's true and nothing has moved, you have the information you need.

Is a repeated mistake always the employee's fault?

No, and assuming so is the most common error owners make here. If no feedback loop exists, the pattern belongs to the business. Onboarding, training, a way to find out you were wrong, and reviews on a schedule all have to exist before the person's teachability is even a fair question.

What if the mistake is expensive?

Cost tells you how urgent the response is, not whose fault it is. An expensive error still has to be diagnosed. Firing someone the day after a costly mistake often means paying the tuition and then handing the education to your competitor.

Do small businesses really need quarterly reviews?

A business with eight or ten people needs them more than a large one does, because there's no HR function catching what the owner misses. An hour per person per quarter is a small cost against the alternative, which is finding out at an exit interview.

Should compensation be tied to reviews?

Yes, to the quarterly and annual conversations both. A standard with nothing attached to it is a preference, and people read preferences as optional. That doesn't mean every review moves the number. It means the number moves in that conversation and not somewhere else.

Ready to look at the architecture honestly?

Book the conversation. We'll tell you what we see, and whether the work we do fits where you are.

Book the conversation →