Skip to content
←

필사 모드: The Order of Operations After a Mistake — Contain, Report, Record, Prevent

English
0%
정확도 0%
💡 왼쪽 원문을 읽으면서 오른쪽에 따라 써보세요. Tab 키로 힌트를 받을 수 있습니다.

Introduction

You make mistakes at work. That is the default, not the exception. Which makes how not to make mistakes an unhelpful topic.

The useful topic is elsewhere. The cost of a mistake is set not by the mistake itself but by the next few hours and days. Two people make a mistake of the same size, and one of them loses trust while the other actually gains it. The difference is neither personality nor luck; mostly it is order.

There are four steps in the order: contain, report, record, prevent. The order itself is not remarkable, and most people already know it. The problem is that in real situations it gets inverted in almost exactly the same way every time. This post looks at why each stage sits where it does, and where it usually breaks down once you are standing in it.

1. Why the Order Is What It Is

The four stages are ordered not by importance but by time sensitivity.

Containment comes first because the damage grows as time passes. Reporting comes second because the options left to the other person shrink over time. Recording comes third because memory distorts over time. Prevention comes last because it is the only one that is not urgent.

So this order has nothing to do with how important each stage is. Prevention being last does not make it least important. Over the long run it is worth the most. It just is not urgent.

And because it is not urgent, in practice it barely ever happens. That is the trap in this order.

2. Stage One, Containment — Stop the Bleeding, Not Diagnose

The most common inversion here is going after the cause and failing to stop the spread.

The natural reaction the instant you find a mistake is to work out how this happened, because not understanding is unsettling. But during those fifteen minutes the wrong data keeps going out, the wrong email keeps sending, the wrong setting keeps applying.

What is needed right now is not diagnosis but stopping the bleeding. In most cases you can stop it without knowing the cause. And once it is stopped, you can find the cause accurately, with no time pressure on you.

The second most common failure runs the other way. Rushing to undo it and creating a bigger mistake. Trying to delete what you sent by accident and deleting the original too; trying to fix the wrong setting and touching a different one. Cases where the second mistake is bigger than the first are genuinely common.

So it helps to keep one rule inside the containment stage. You do not decide an irreversible action alone. Reversible actions, fast; irreversible actions, cleared with at least one other person. That one rule prevents most second mistakes.

The line you should not cross alone usually looks like this. Actions that delete data, messages that go outside, actions that move money, and actions whose undoing would require someone else's work. If it falls into one of those four, stop for a moment and call someone.

3. Stage Two, Reporting — Do Not Wait for the Containment to Finish

The inversion here is always the same. Holding the report until the containment is done.

The reason is always the same too. You are hoping that once it is all fixed there will not be much left to report. And occasionally that is what happens. It gets quietly repaired and nobody ever knows. That successful experience is what makes you hold off longer next time.

The problem is that the payoff structure of this bet is bad. Win, and nothing happens. Lose, and you have two things instead of one: the mistake and the concealment. In that structure, a high win rate still leaves the expected value negative.

Judging the timing of a report is covered in detail in Bad News Is Judged by Timing More Than by Content. Here I will only touch the part that overlaps with containment.

A report made in the middle of containment does not have to be a finished report. One line is enough.

I just found this problem and I am stopping the spread right now. I will update you on the situation in thirty minutes.

That one line does two things. It keeps the other person from hearing it through another path first, and it lets them prepare if there is something they need to prepare now. And if you wrote thirty minutes, you actually send something in thirty minutes. You send it even with no progress. No progress is information too.

4. Stage Three, Recording — The Stage That Gets Dropped Most

Once the containment is over you feel relief. And then you skip the record. Of the four stages, this is the one that actually disappears most often.

Why you should not skip it becomes visible the moment you put the note you scribbled at the time next to the write-up you produced two days later. Try it with something you have already been through. The sequence has moved around, what you did at which hour has been rounded off, and something you started and abandoned has been entered as something you did. This happens without you telling a single lie. Memory seems to be less a matter of storing an event and pulling it back out, and more a matter of reassembling it each time, and the reassembly tidies things in the direction that makes them cohere.

So the careful document you write later cannot replace the rough note you left in the moment. Believing that it can is exactly what makes people skip this stage.

Four things go in.

  • Times. When it happened, when you found out, and when you did what. A record with no times cannot be reconstructed into a sequence later.
  • Actions taken. The steps you actually executed. Kept separate from the ones you started and abandoned.
  • Observations. What appeared on the screen, what the value was. Observation, not interpretation.
  • Who you told. Who you told what, and when.

And one important framing. The primary use of this record is not defense. It is the raw material for prevention. Start writing it as a defensive document and the sentences distort, and a distorted record is useless at the next stage. Write only facts and the defense takes care of itself.

In practice the easiest approach is to keep a scratch file open while you contain and stack one line at a time. It does not have to be in sentences and the order can be tangled. Tidying can happen later; the only thing that cannot happen later is the observation.

5. Stage Four, Prevention — A Resolution Is Not a Measure

This is the most uncomfortable part of the post.

It will not happen again. I will be more careful. I will double-check from now on. These sentences sound like prevention measures, but in practice they are closer to a declaration that no prevention will occur. Because the conditions are unchanged.

The same person, in the same state of fatigue, with the same tools, working in the same order, produces the same mistake. Attention is a resource, and resources run out. If it did not happen this time, that is not the measure working; that is luck.

A measure has to change a condition. There are four directions.

  • Change the order. Put the dangerous step last, or reorder so the check comes first.
  • Add a verification step. Make the procedure do the checking, not the person. Not a resolution to look twice, but a structure where nothing proceeds unless it has been looked at twice.
  • Make it reversible. Making a mistake recoverable is generally cheaper and more reliable than preventing it.
  • Make it visible. Have an alert fire when something is wrong. This does not remove the mistake, but it shortens the time to discovery.

And one honest caveat belongs here. Measures have costs too. Insert one verification step and every task after it becomes slightly slower. A measure that spends ten minutes a day to prevent a mistake that occurs once a year is not a good measure.

So a final judgment is needed. How often does this mistake happen, how much does one occurrence cost, and how much does the measure cost? Running that calculation and deciding to build no measure is also a conclusion. But that conclusion has to be written down explicitly. Leave it unwritten and, the next time the same thing happens, it looks as though no judgment was ever made.

6. The Line Between an Apology and an Excuse

Most people have had the experience of an apology being read as an excuse. It feels unfair, and where exactly it split is unclear.

The line is this. An apology is about the impact, and an excuse is about the cause. And explaining the cause is not itself a bad thing. It is indispensable for the prevention discussion. The problem is position.

Put the cause inside the apology sentence and the whole thing becomes an excuse. The schedule was tight, which is how this happened, I am sorry. What stays with the listener there is not the apology but the claim that the schedule was tight. A single conjunction makes the apology conditional.

So you separate them. The apology gets one paragraph and the account gets another.

The release slipped by a day. It affected the neighboring team's schedule, and it was my misjudgment. I am sorry.

Here is the account. One of the verification steps was skipped, and the reason it was skipped is that the step had always passed in previous releases. For prevention I would propose automating that step.

The same content, and the impression is completely different. The first paragraph contains no reason, and the second contains no apology.

And one point about length. The shorter the apology, the more trust it carries. When it runs long, two things appear. First, room opens up inside it for an excuse to mix in. Second, the other person picks up the burden of feeling they should console you. A long apology ends up demanding emotional labor from the person you wronged. That sits on the opposite side of an apology.

7. When You Are Handling Someone Else's Mistake

This post is about your own mistakes, but you are only seeing half of it if you do not look at the reverse side.

How you react when someone else makes a mistake determines how fast that person will tell you next time. There are no exceptions to this. One experience of being reprimanded first when reporting a mistake and, from then on, that person checks before telling you. And as we saw above, checking before telling is late.

So the order of the reaction is fixed too. The first reaction is to ask what is needed right now. Cause and responsibility come after the containment. Around people who hold that order, incidents get found early; around people who do not, they get found late. A team's incident response speed rests mostly on the lead's habits in the first reaction.

And in the prevention discussion, it matters that a person is not the subject of the sentence. Not who got it wrong, but under what conditions this mistake was possible. Under the same conditions, someone else makes the same mistake.

8. One Thing You Can Do Today

Bring to mind the most recent mistake you made. It does not have to be a large one.

Then write a single line. What was the condition that made that mistake possible? The answer that you were careless is not a condition. There was no time, there was no verification step, the operation was irreversible, you judged it alone — those are conditions.

Once the condition is written down, the next line usually comes out on its own. And that one line beats ten resolutions.

Where This Advice Does Not Apply

In incidents touching safety, money, personal data, or regulated areas, the order is not an individual's to set. These areas already have fixed procedures, reporting deadlines, and reporting lines, and those procedures take precedence over individual judgment. Even stage one, containment, is affected. There are cases where the scene has to be preserved, and there the act of undoing is itself a violation. Checking whether your own work falls under such a procedure before an incident occurs is worth more than this entire post.

Where legal liability is involved, the nature of the record changes. A personal note and an official record may be treated differently, and what you leave where can matter later. The recording stage in this post assumes an ordinary workplace mistake. If a dispute is foreseeable, the right move is to check the organization's rules first and, if needed, ask the responsible function.

In an organization that punishes mistakes, this order can work against the individual. The record especially. There are organizations where a record kept as raw material for prevention gets used as evidence for assigning blame. Even in that environment recording is still better than not, but the form has to change: write only facts and times, leave out your own assessment of your judgment, and use the format of a shared team record rather than a personal note.

If the same mistake keeps recurring and the measures keep failing, the target was chosen wrong. If the same mistake has now happened a third time, it is unlikely to be an individual problem. The workload has exceeded capacity, or the tools are bad, or the procedure itself is designed to induce the mistake. Continuing to produce individual-level prevention measures at that point only pins the problem to the individual. The layer to address is a different one. If the workload itself is the problem, The Structure of Saying No comes first.

Finally, this post covers what comes after a mistake, not the feelings that come with one. After a big mistake, work does not stick for a few days. That is normal, and following the procedure well does not make it go away. A procedure is a tool for reducing damage, not a tool for settling your mind.

Closing

The mistake itself is mostly not what gets remembered. What remains six months later is how the person moved after it.

And that movement reduces to four sentences. Stop it first. Tell them before it is over. Write it down on the spot. Change the conditions, not your resolve.

Only the second one is hard. The rest is a matter of knowing.

Further Reading

현재 단락 (1/64)

You make mistakes at work. That is the default, not the exception. Which makes how not to make mista...

작성 글자: 0자원문 글자: 12,112자작성 단락: 0/64