Why it matters
Conflict questions exist because engineering work is full of disagreements that have no correct answer in advance: build versus buy, ship versus harden, one owner versus two. The interviewer is testing whether you can hold a position, hear a counter-position and still move the work forward. Stories that end with a manager resolving it show no mechanism, and that is what the question is looking for.
Mental model
A conflict story is a decision record with a relationship attached: options, positions, the evidence that broke the tie, and the follow-up. The metaphor breaks because a decision record is written after the fact, while the skill under test is what you did while the disagreement was live.
How it works
Each rung tests one part of the answer; a failure stops the story before it is told.
Step 1 of 6
The person is not the story
Start with the disagreement: two defensible positions and real stakes. The interviewer scores the decision, not the personalities.
question
Is it a disagreement about a judgement, not a person?
Both sides need a defensible position and real stakes. A story about a difficult colleague scores no judgement.
Path walked
- You are preparing a conflict story
- Is it a disagreement about a judgement, not a person?
Figure. A conflict answer is judged on the decision process, not on who won. Four rungs test whether the story is about a judgement, states both positions, shows the step before escalation, and leaves a mechanism behind; a failure stops the story until it is fixed or replaced.
What to notice. The rungs are the page's rubric, not a scoring scale. The figure draws no word count, duration or outcome measure; the ninety-second rehearsal in the page is a practice limit, not a threshold on the answer. The figure does not claim every conflict needs an escalation — it claims the answer must show what you did before one.
Text version
A decision ladder for preparing a behavioural conflict answer. The start is a conflict story being prepared. Rung 1 asks whether it is a disagreement about a judgement, not a person; no leads to picking a judgement conflict, because a personality clash scores no judgement. Yes continues to rung 2: can you state both positions fairly? No leads to rehearsing their case first, because a counter-position they would recognise shows you listened. Yes continues to rung 3: did you take a step before escalating? No leads to naming the step you took, an experiment or a written trade-off or the decision's owner. Yes continues to rung 4: did the process change afterwards? No leads to closing with the mechanism, so the conflict does not recur. Yes leads to the story that survives: a decision you owned, an outcome, and a change that outlasted the project. Every stop is marked Choose, because each is the recommended move for that deficiency.
- Choose a disagreement where both sides had a defensible position and the stakes were real, not a story about a difficult colleague.
- Open with one sentence of context, then state your position and the evidence you had for it.
- Describe the counter-position fairly, in terms its holder would recognise, before saying why you disagreed.
- Name the step you took to resolve it: a smaller experiment, a written trade-off, a review with the owner of the decision.
- Explain how the decision was made and what you did when it went against you.
- Close with the outcome and the process change that prevented the same disagreement from recurring.
Trade-offs
| Choice | Good for | Cost | Fails when |
|---|---|---|---|
| Disagree and commit | Fast decisions and preserved trust | A wrong decision still ships | The disagreement was about safety or ethics |
| Escalate early | Clear authority and unblocked work | Manager time and a signal you cannot resolve alone | The disagreement is a judgement call you could test |
| Run an experiment | Evidence instead of opinion | Time and a temporary fork in the code | The decision is irreversible or expensive to test |
| Defer to the owner | Respects ownership boundaries | A decision you think is wrong stands | You have information the owner lacks and never shared it |
Good for
Fast decisions and preserved trust
Cost
A wrong decision still ships
Fails when
The disagreement was about safety or ethics
- The disagreement was about safety or ethics
- The disagreement is a judgement call you could test
- The decision is irreversible or expensive to test
- You have information the owner lacks and never shared it
Figure. Each choice with what it buys, what it costs, and the case where it loses.
What to notice. The fourth column is the decision: if the losing case describes your situation, the choice is wrong.
Text version
Four trade-offs: Disagree and commit is good for Fast decisions and preserved trust, costs A wrong decision still ships, and fails when The disagreement was about safety or ethics. Escalate early is good for Clear authority and unblocked work, costs Manager time and a signal you cannot resolve alone, and fails when The disagreement is a judgement call you could test. Run an experiment is good for Evidence instead of opinion, costs Time and a temporary fork in the code, and fails when The decision is irreversible or expensive to test. Defer to the owner is good for Respects ownership boundaries, costs A decision you think is wrong stands, and fails when You have information the owner lacks and never shared it
Failure modes
- Villain story. The answer describes a colleague's behaviour and never your own decision. Debug: count sentences where you are the subject of a choice.
- No resolution. The story ends with the conflict unresolved or resolved by someone else. Debug: check whether the answer contains a decision, a date and an outcome.
- Trivial conflict. A disagreement about formatting does not demonstrate judgement under stakes. Debug: ask whether a reasonable engineer could have chosen either way and whether it mattered.
In production
Rehearse this answer aloud with a timer and record it once. If the story cannot be told in ninety seconds without losing the decision, the story is too complicated to use.
Common misconceptions
People think: The best conflict story is one you won.
Actually: Interviewers learn more from a decision that went against you and how you handled it than from a victory.
People think: Conflict answers should avoid emotion.
Actually: They should show that you noticed the tension and chose a response; pretending it was purely technical reads as unaware.
Interview angle
A strong answer names the disagreement, both positions, the evidence or process that resolved it, and the outcome including what you changed. It is specific about your actions. Weak answers describe a difficult person, avoid saying what they wanted, or end before the decision.
Sources
- Will Larson, Staff Engineer guides — influence and disagreement without authority.
- interviewing.io blog — what interviewers record from behavioural answers.
- Amazon Leadership Principles — a published example of the competencies behavioural loops score.