Tell Me About a Time You Were Wrong: Use a Position You Argued, Not a Bug

To answer “tell me about a time you were wrong,” pick a position you argued and later dropped. A bug you misdiagnosed is a weaker pick, because the question asks how you change your mind. The interviewer wants four things. First, what you believed and why it was reasonable at the time. Second, what evidence showed you were wrong. Third, how long you held on after that evidence appeared. Fourth, what you did about the people who had been right. Amazon writes this into its “Are Right, A Lot” leadership principle. It says leaders “work to disconfirm their beliefs.” GitLab’s handbook has a value called “Articulate when you change your mind”. It asks people to say so openly when new data changes their stance. A strong answer shows both habits. You looked for the evidence. Then you announced the reversal.

A misread bug answers a different question

Most engineers reach for a debugging story. They were sure the crash came from the database. It came from image decoding instead. That story shows you can debug. It says little about judgment, because nobody argued with you. A wrong hypothesis that a profiler overturns in an hour costs you nothing. The interviewer can’t see how you behave when being wrong is expensive.

A failure story has the opposite problem. Failure questions score ownership of a bad outcome. A “wrong” question scores the belief behind the outcome. The two can share a story. The emphasis is what moves. Here, spend your time on the position you held and the moment it broke. Spend less on the cleanup.

So the best material is a disagreement you won and later regretted. You had reasons. Someone argued the other side. Then events proved them right.

Marek argued that feature flags weren’t worth it

Take Marek, an Android engineer on a food delivery app. His team shipped to the Play Store every two weeks. A staff engineer proposed remote feature flags for every new checkout change. Marek argued against it in the design review. His case was sound on paper. Flags add code paths. Stale flags rot. The team had a solid QA cycle. The review sided with Marek. The team skipped the flags.

Two releases later, a change to the saved-card screen broke payment for one card type. QA hadn’t covered that card type. The fix took a day to write. Getting it through review and the staged rollout took four more days. During those days, users with that card couldn’t pay in the app. With a flag, someone could have turned the change off in minutes.

The story works because each part is concrete. Marek held a position in public. He had reasons a reasonable engineer would accept. The cost of being wrong was measurable. A named person had argued the other side.

Say when the evidence arrived and how long you held on

Interviewers listen for the gap between the evidence and the update. A short gap is a good sign. A long gap is fine if you can explain it honestly.

Marek’s honest version has a delay in it. After the outage, his first reaction was that QA had missed a case. He spent a day arguing that better test coverage was the real fix. Then he looked at the rollout timeline. Better tests might catch this bug. They wouldn’t shorten the five days for the next one. That was the point where his view changed.

Keep that day in the answer. It shows you defended your position until the evidence was clear, then let it go. Cutting it makes the reversal sound instant and rehearsed. Amazon’s principle asks leaders to work to disconfirm their beliefs. The rollout timeline was Marek doing that, a day late.

Tell the people who heard the first position

A private change of mind is half the answer. The team still thinks your original argument stands. GitLab’s handbook treats new data as a fine reason to change position. The same value asks you to state plainly that your earlier stance no longer holds.

Marek did three things. He told the staff engineer first, in person. He then wrote a short note to the design review channel. The note admitted he had argued against flags. It said the payment outage changed his view and explained why. Last, he offered to build the first flag himself, on the saved-card screen.

Name the person who was right. Then say what you told them. Interviewers often probe this part. “How did your teammate react?” is a common follow-up. A vague “we were fine” suggests the conversation never happened.

A sample answer to “tell me about a time you were wrong”

Marek and his app are invented for this post. Borrow the order of the parts, not the story.

“Last year I argued against adding remote feature flags to our checkout code. We shipped every two weeks and had a good QA cycle. I thought flags would add code paths we’d never clean up. Our staff engineer, Anja, wanted them. The design review went my way.

Two releases later, a saved-card change broke payment for one card type. The fix took a day. Review and the staged rollout took four more. For those five days, those users couldn’t pay in the app.

My first reaction was that QA had missed a case. I pushed for more test coverage for about a day. Then I laid out the rollout timeline. More tests might catch this bug. The next one would still take five days. A flag would have turned it off in minutes. Anja had been right about the risk I was discounting.

I went to her first and admitted it. Then I posted a note in the design review channel. It admitted I’d argued against flags. It explained why the outage changed my mind. I built the first flag on the saved-card screen myself.

The bigger change was in how I argue. When I oppose a safeguard now, I name the failure that would change my mind. I write it in the design doc. With flags, that would have been ‘a bug we can’t roll back within a day.'”

The last paragraph is where the answer earns its score. It shows a lasting change in how Marek makes decisions. “I learned to listen more” would claim the same thing with nothing to check.

“How did you find out you were wrong?” separates luck from habit

Expect this follow-up. It asks whether you went looking or got lucky. In Marek’s story, the outage forced the question. That is common. It’s also honest. But the interviewer will notice that nothing in his process would have caught it earlier.

The design-doc habit in the sample answers this. Writing down the failure that would change your mind is a way of looking for disconfirming evidence. If the follow-up comes, point to it. Say that before the outage, he had no test for his own position. Now every design doc he writes has one.

If your own story has a moment where you went looking on purpose, use it. Maybe you ran a spike to test your own estimate. Maybe you asked the skeptic for their strongest case. A story like that is stronger than an outage, because the habit was already there.

When you never told anyone you’d changed your mind

Many honest stories lack the public reversal. You changed your view quietly and moved on. The person who was right never heard it from you. Don’t invent a conversation that didn’t happen. Follow-ups about reactions and timing will expose it.

Tell it as it went. Treat the silence as part of what you were wrong about. Say you changed your position without saying so. The team kept building on your old argument. Then give a later case where you did announce a reversal. Keep that second case to two or three sentences. If you have no later case, describe the next reversal concretely. Name the person you’d tell first. Then name where you’d write it down. That is usually the design doc or channel where you first argued.

Leave a Comment