Conflict and Disagreement

you disagreed, and something got decided

What is scored: can you hold a position, engage seriously with the opposing one, and reach a decision — including one that goes against you — without damage to the relationship?

This is the competency most likely to reveal something the candidate did not intend to reveal.

What the interviewer is looking for

A real disagreement, on something that mattered. A trivial one signals you avoid conflict.
You can state the other position fairly. The single strongest signal in this competency. If you can only describe their view as wrong or uninformed, you have shown how you behave in every design review.
A mechanism for resolution — data, a prototype, a time-box, a trial, escalation. Not "we talked until we agreed", which usually means someone gave up.
A decision, and you committed to it. Especially if it went against you.
The relationship survived. Ideally evidence of that: you worked together again afterwards.

Having a story where you lost and committed anyway is extremely valuable, and most candidates do not have one prepared. Disagree-and-commit is a specific, assessable behaviour.

Weak and strong, same prompt

Prompt: tell me about a disagreement with a colleague.

Weak.

"A teammate wanted to use MongoDB for a service where the data was clearly relational. I explained why that was wrong and showed him some articles about it. Eventually he came around and we used Postgres, which turned out to be the right call — we needed joins within a month."

Three problems. The other person has no reasoning, only an error. The resolution mechanism is "I explained until he agreed", which is not a mechanism. And the ending is self-congratulatory, which makes the whole account read as edited. Interviewers hear "I was right and he was wrong" as a description of the candidate, not the situation.

Strong.

"Our tech lead wanted to cut over to the new payments provider in one step. I wanted a dual-write phase with reconciliation, which would cost about two weeks of an eight-week contractual deadline. Her reasoning was good, and it was not just about the deadline: she had been on a dual-write migration that ran nine months because nobody would commit to turning the old path off. That is a real failure mode and I had not accounted for it. So I reframed it. Rather than arguing about the strategy, I proposed a dual-write phase with a hard expiry, written down: we would cut over as soon as reconciliation had been clean for nine days, and no later than week six. That addressed her actual concern, which was open-endedness, not verification. She agreed, and we cut over in week six, after nine clean days. What I took from it: my first version of the argument was about being right, and it would have lost. The second version was about her concern, and it took five minutes. I now try to find out what someone is actually worried about before I make a case, because it is usually not the thing they are arguing about."

The other position is steelmanned and has a real basis. The resolution is a mechanism — a time-box that addressed the underlying concern. And the reflection is about the candidate's own approach, not about winning.

Question bank

Tell me about a disagreement with a coworker.
Describe a time you disagreed with your manager.
Tell me about a time you were overruled. What did you do?
How do you handle it when someone rejects your technical approach?
Tell me about a time you had to work with someone difficult.
Describe a time you changed your mind after pushing back.
Tell me about a decision you disagreed with but had to implement.
When did you escalate a disagreement? Would you do it again?

"Tell me about a time you were overruled" is the one that separates candidates. The expected answer is not that you were secretly right.

Where these answers go wrong

Being right. A story that exists to show you were correct is weaker than one that shows you disagreed well. If you were right, downplay it; the process is what is scored.

A straw man. The other person "did not understand the requirements" or "just wanted to use a new technology". Real colleagues have real reasons. If you cannot produce theirs, either you did not understand their position at the time — which is the actual finding — or you are editing.

No resolution. "We escalated to the manager and he decided" is only complete if you say what you did next. Did you commit? Did you support the decision publicly? That is the assessable part.

Making it personal. Any hint that the disagreement was about the person rather than the position is damaging and hard to recover from in the debrief. Describe positions.

Avoiding conflict entirely. Some candidates answer this by explaining that they prevent disagreements by communicating early. That is not an answer to the question and is read as conflict-aversion, which for senior roles is a genuine concern — a senior engineer who never disagrees is not doing the job.

How to say it in the interview

"We disagreed about the cutover strategy. She had seen dual-writes drag on for months, which was a fair concern - here is how we resolved it."

Checklist

The disagreement was about something that actually mattered
You stated the other position fairly and with a real basis
You resolved it with a mechanism, not by out-arguing them
A decision was reached and you committed to it
You have one story where you lost and committed anyway
The relationship survived, with evidence
You described positions rather than characterising the person

Practice this interactively