In this article
Before you start
The interview process guide explains the stages of a loop and what each round tests.
Why interviewers ask this
The question is reported by candidates in engineering-manager loops, and the cost of a poor answer is that it shows how you treat your own mistakes. A manager's decisions land on other people, so interviewers use this prompt to see whether a candidate can examine one of them without defending it or punishing themselves. The structure of such loops is covered in the Meta engineering manager interview page. Other companies lay out their loops differently; see the Anthropic interview process and OpenAI interview questions.
A weak answer picks a trivial slip, blames the context or a boss, or says nothing would be reversed. A strong answer picks a decision with real stakes, shows what was known at the time, owns the part that was missed and states a concrete change. Companies publish the values a question like this touches: Amazon's leadership principles include Ownership and Learn and Be Curious, and a candid reversal story is evidence for both.
Reversing a management decision from first principles
- Two-way decision
- A two-way decision is a choice that can be undone at a modest cost, in contrast to a one-way decision that is hard or impossible to undo. Amazon's 2015 shareholder letter calls them Type 2 and Type 1 decisions. The letter argues that most decisions are two-way, and team and process changes usually are, which is why an honest reversal story is normal.
- Outcome bias
- Outcome bias is the habit of judging a decision by how it turned out rather than by what was known when it was made, so a sound call with a bad result looks like an error and a lucky call looks wise. Baron and Hershey (1988) studied it in decision evaluation.
In plain words
A reversal answer is a short review of one of your own decisions. You tell it out loud. Think of a chess player replaying a lost game. The useful replay asks if the move made sense given the board then. It does not only ask if the game was lost. The analogy breaks in one place: a manager's board includes people who lived with the result, so the review names them.
In one precise sentence, a strong answer picks a real decision that was reasonable when made, separates it from its outcome, names what you would reverse, and ends with a change in how you decide. Each step below adds one part of that shape.
Step 1: Pick a real decision that was reasonable when made
A strong answer picks a real decision that was reasonable when made and had consequences for other people. Good choices are a team reorganization, a hire or a promotion, a roadmap cut, a process change, or a choice to keep or end a project.
A decision that is too small, such as when a meeting was scheduled, signals that you are avoiding the question. A decision that was careless signals a judgment problem. Edmondson (2011) separates failures that were avoidable from those that came from reasonable attempts under uncertainty, so say honestly which kind yours was. Team and process changes are usually two-way decisions, so the cost was real but recoverable, which makes them honest material.
Input: your history of decisions. Output: one decision with stakes and a defensible reason behind it.
Step 2: Judge the decision apart from its outcome
A fair verdict separates the decision from the outcome before it names what to reverse. Say what you knew when you chose, which options you weighed and why you picked one. Then state the outcome plainly, including the cost to people.
Outcome bias pulls speaker and listener to judge by the result, so show that you can pull the two apart: "with what I knew it was reasonable; what I missed was this." The blameless postmortem stance in the Google site reliability engineering (SRE) book asks what conditions let the failure happen rather than who to blame, and a candidate can borrow it while still owning their share. Ownership means "I decided", not "the situation forced it".
Input: the decision. Output: a verdict that names one thing you missed.
Step 3: Name the change in how you decide
The strongest part of the answer is the change in how you decide now, stated as a habit you can point to. A strong answer ends with a change in how you decide, because that is where the follow-up goes: "what do you do differently?" is natural to ask, so say it before it is asked.
A change is concrete when it has a trigger and an owner: a written rollback condition before a reorganization, a short pre-mortem before a hire, a named check-in partway through a change, or a list of who holds which knowledge. "I communicate more" is not concrete. If you have used the habit since, one sentence on that later case turns a lesson into evidence. The answer is a habit, not a regret.
Output: one habit that a listener could check.
Step-by-step description
- Choose a real, reasonable decision.
- Separate the decision from the outcome.
- Name the change in how you decide.
- A strong answer picks a real decision that was reasonable when made.
- A fair verdict separates the decision from the outcome before it names what to reverse.
- A strong answer ends with a change in how you decide.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →The Staff-level answer
The answer below is an illustrative example, not a real person's history. It follows the three steps and is built to be said aloud.
Tell me about a management decision you'd reverse if you could.
I would reverse how I split my team into two groups to hit a deadline. It was a reasonable call: the work divided cleanly and I asked both leads first. What I missed was that on-call knowledge sat with three people who ended up in different groups, so incidents took longer to resolve for a quarter. I moved the shared services back under one owner. Now any reorganization I plan starts with a list of who holds which knowledge, names a check-in partway through, and carries a written condition for rolling it back.
What makes it Staff-level is the last sentence: the lesson is a method with three parts that anyone could audit, and the failure is described as a gap in information rather than a gap in effort.
At an illustrative pace of 130 words a minute, a 60-second slot holds 130 words. The illustrative split of a 60-second answer is 12, 18, 12 and 18 seconds for context, the decision and its reasoning, the reversal, and the change.1 The change in how you decide gets as much time as the decision. A longer illustrative slot of up to 90 seconds suits an interviewer who invites detail; add that detail to the reasoning and to a later use of the habit, not to the context.
Text description of the diagram
- The illustrative split of a 60-second answer is 12, 18, 12 and 18 seconds.
- The change in how you decide gets as much time as the decision.
A second example: a hiring decision
In a second illustrative case, a manager made the strongest individual engineer a team lead without testing how that person coached others. The call was reasonable because the engineer had asked for the role and had led projects well. What the manager missed was that no one had checked coaching skill, and two teammates quietly moved teams. The reversal is not the promotion itself but the missing test of coaching skill. The changed habit is a short, named stretch assignment with feedback from the team before any promotion to a lead role.
The rubric
The table below is our own reading of what separates answers, written as advice; it is not a published company rubric.
| Dimension | Weak answer | Strong answer | Staff answer |
|---|---|---|---|
| Choice of decision | Trivial slip or "nothing to reverse" | A real decision with consequences for others | A real decision, and says whether the failure was avoidable |
| Judgment at the time | Hindsight only | States what was known and why the call was reasonable | Names the missing information and why it was hard to see |
| Ownership | Blames context or a boss | "I decided" and owns the cost | Owns the cost and names who was affected |
| Learning | "I learned to communicate more" | A specific change in method | A trigger, an owner and a later case where it worked |
| Communication | Rambles or stops at the regret | Clear structure using STAR | Short, ends on the habit, ready for the follow-up |
Three variants of this question
Tell me about a time you changed your mind on a decision.
Use the same shape. Say what you decided, what new information arrived, why you reversed, and what you now check before deciding. The emphasis moves from regret to updating.
Tell me about a bad decision you made, or a decision you regret as a manager.
Pick the same kind of decision, but spend more time on the cost to people and on the repair you made. The lesson is still a habit, not a feeling.
What would you do differently in your last role?
The question "What would you do differently in your last role?" is broader and lets you choose a decision, a process or a relationship. Keep the same rule: one specific example, a fair verdict and one concrete change.
Common mistakes
- "I wouldn't reverse anything." (wrong) It reads as a lack of self-awareness. (right) "There is one decision I would change, and here is what I learned."
- "The deadline and my manager left me no choice." (wrong) It shifts the blame away from your choice. (right) "I chose the split; here is what I missed."
- "It was a small scheduling mistake." (wrong) It avoids the stakes. (right) Pick a decision that affected a team's work.
- "It turned out badly, so it was a bad decision." (wrong) It confuses the decision with the outcome. (right) "With what I knew it was reasonable; this is what I missed."
- "I learned to communicate more." (wrong) No one can check it. (right) "I now write a rollback condition and name a check-in before any reorganization."
Follow-up questions
What would you do differently?
State the habit from step 3 again with its trigger, then give a later case where you used it. If there is no later case, say how you would apply it next time.
How do you answer a question about a decision you regret without sounding careless?
Separate the decision from the outcome: say what you knew, why the call was reasonable, and what you missed. A candid account of the gap sounds careful, not careless.
Is it acceptable to say there is nothing you would reverse?
It is risky, because it removes the evidence the interviewer asked for. A reversal story from a reasonable decision shows judgment and ownership at the same time.
How long should the answer be?
Plan on about a minute, with an illustrative upper limit of 90 seconds, and spend as much time on the change as on the decision. Detail for the interviewer to explore belongs in the follow-up.
What if the decision was not yours alone?
Say who else was involved, but describe your own part: what you argued, what you agreed to and what you would do differently. Describing shared decisions fairly reads as maturity.
- Pick a real decision with consequences that was reasonable when made.
- Separate the decision from the outcome, and own the part you missed.
- End with a change in how you decide that has a trigger and an owner.
- Give the change as much time as the decision, and say it before the follow-up.
- Avoid saying nothing would be reversed, and avoid shifting the blame.
Sources
- 2015 Letter to Shareholders (Type 1 and Type 2 decisions), exhibit 99.1 to Form 8-K — Amazon (SEC filing), 6 Apr 2016
- Leadership Principles — Amazon, 26 Oct 2026
- Postmortem Culture: Learning from Failure — Google, Site Reliability Engineering book, 1 Apr 2016
- Strategies for Learning from Failure — Harvard Business Review (Edmondson), 1 Apr 2011
- Outcome bias in decision evaluation — Journal of Personality and Social Psychology 54(4), 569-579 (Baron and Hershey), 1 Apr 1988
