Tell me about a time you had to work closely with someone you found difficult.

What this question is filtering for

Nobody asking this expects you to have liked everybody. The interviewer is running two checks at once, and only one of them is about the conflict. The first is whether you can work productively with someone whose style you do not enjoy, because a hiring manager is about to put you on a team they cannot reshape around you. The second, and the one that usually decides the score, is how you talk about an absent colleague to a person who has no way of verifying anything you say. You are being watched as a future teammate describing a current one, and the interviewer is quietly imagining you doing it about them.

That makes generosity a competence signal rather than a courtesy. A candidate who can explain what the difficult person was optimising for has demonstrated the thing the question is really about: the ability to model someone whose priorities differ from yours, which is the precondition for getting anything useful from them.

Choosing a story with real friction in it

The most common weak answer is a story with no conflict in it. A colleague who wrote terse pull-request comments, a designer who preferred meetings to documents, a contractor in an awkward time zone — these are inconveniences, and reaching for one tells the interviewer either that you have never had a hard working relationship or that you are unwilling to discuss the ones you have had. Neither reading helps you.

Pick a relationship where the friction had a consequence: work stalled, a decision kept being reopened, quality dropped, or you found yourself routing around the person to get anything done. Say what the cost was within the first two sentences, because that is what licenses the rest of the story. Then make clear that you had to keep working with them, since the interviewer wants to see what you do when walking away is not available. If your only story ends with the other person being moved off the project, you have described an escalation rather than a working relationship, and you should have a second story ready.

Describing the other person without prosecuting them

The register here is narrow. The interviewer needs to believe the difficulty was real, while you never sound as though you are building a case. The reliable technique is to describe the behaviour and its effect on the work, then offer the most reasonable explanation you can construct for why a competent adult would behave that way.

A version that costs you	A version that reads as senior
"He was territorial and insecure about his code."	"He had owned that service alone for four years and had been burned by a change he had not reviewed. My pull requests sat for a week."
"She was impossible in meetings, she just steamrolled everyone."	"She would restate the decision as settled while two of us were still arguing. I think she read the disagreement as slipping delivery, which she was accountable for."
"The QA lead was obstructive for the sake of it."	"He blocked our releases on tests we thought were flaky. He had signed off a bad release the previous quarter and had taken the blame for it."

The right-hand column is not softer, it is more informative. It gives the interviewer a mechanism, and a mechanism is what makes an account believable. It also does something quieter: it shows that you went and found out why, which is the actual work of resolving a conflict and the part candidates most often skip.

Where these answers lose the room

Two failure modes account for most poor scores, and neither is visible to the candidate while they are talking.

The first is the villain narrative. The moment the other person becomes irrational, political or insecure, the interviewer stops assessing the conflict and starts wondering what you will say about your new colleagues in eight months. Nothing you add afterwards recovers it, because the judgement being formed is about you rather than about the story.

The second is the resolution that contains no change. A very large share of these answers end with some version of "so I suggested we grab a coffee, we talked it through, and after that we were fine". As a resolution it is unfalsifiable and it credits nothing but goodwill. What an interviewer wants is the specific adjustment you made: you started circulating the design a day before the meeting so she was not hearing it live, you took his review comments to a call because the thread was reading as combative, you began agreeing acceptance criteria with QA before writing any code. Those are mechanisms someone can imagine you repeating here. "We talked and it was fine" is not, and it is the sentence that turns a promising story into an average one.

When the relationship never warmed up

You are allowed a story that ends in a workable truce rather than a friendship, and it is often the stronger answer. Some colleagues do not come round, and a candidate who says so and then explains how the work still shipped sounds like someone with experience rather than someone with a prepared anecdote. Say what you settled on: a narrower interface between you, agreements put in writing, one shared channel instead of four side conversations, a standing fifteen minutes rather than ad hoc interruptions.

What you should not do is dress a cold truce up as a triumph, or claim a lasting friendship that the first follow-up will dismantle. If the interviewer asks what that person would say about you and the honest answer is unflattering, give the unflattering part first and briefly. Candidates who volunteer their own contribution to the friction are rarely marked down for it. Candidates who have to be led to it almost always are, because the reluctance is itself the data point.

The conflict is the setting, not the subject. What gets scored is whether you can describe an absent colleague fairly enough that the interviewer would be comfortable one day being described by you.