Tell me about a time you had to work with a difficult teammate.
answer
- name the shared deliverable and stakes
- behaviours, not labels
- the charitable read you tested
- the direct conversation, in their words
- number plus where the relationship landed
basics
~10 sProbes conflict resolution and self-awareness. Describe one person in observable behaviours, the charitable read you tested, the direct conversation you had, what you changed yourself, and a work outcome with the relationship still standing.
how to answer
6 beats- the shared work and what was riding on itOne or two sentences: what you and this person had to produce together, and the pressure on it. Keep setup and task to roughly fifteen to twenty percent of your airtime — the interviewer needs orientation, not the org chart.
- the friction, described as behaviourSay what a camera would have recorded: comments arriving two days late, being cut off in review, requests answered only when a manager was copied. Never reach for toxic, lazy or arrogant; a label tells the interviewer about your judgement rather than about the situation.
- the charitable read you testedState the hypothesis you formed about their pressures — overloaded, measured on something else, burned before — and that you checked it rather than assumed it. Being wrong about your first read is fine to admit and often makes the answer more credible.
- the direct conversation and what you changedThis is the bulk of the answer, around sixty percent. Give the private conversation: how you opened it, what you asked for, and crucially what you offered to change on your own side. A working agreement you both accepted beats a speech you delivered.
- the result, in the work and in the relationshipClose with a number that moved — a cycle time, a defect count, a quality measure — and one sentence on where the relationship stood afterwards. Result and reflection together are about twenty to twenty-five percent of your airtime.
- what you took into the next difficult relationshipOne short sentence of transferable learning, phrased as a rule you now follow rather than a moral about other people. Skip it if you have already run long — a rushed result beat is a worse trade than a missing reflection.
your answer
5 story prompts- Pick a colleague you genuinely struggled with, ideally in the last two years, and one thing you changed yourself.
- Choose an episode where the work improved, not only the atmosphere, and bring one number.
- Write the opening sentence you used in the private conversation, word for word.
- Check the ending: can you say in one line where the relationship stood afterwards?
- This can be the same episode as your disagreement story, re-angled onto the person rather than the decision.
draft and rehearse your own answer in a learn session
go deeper
Friction between capable people is normal, so the interviewer is predicting how you behave when it happens on their team. The axis is conflict resolution and self-awareness: whether you read the other person charitably, addressed them directly and privately, adjusted your own approach, and kept the work moving — or whether you narrated someone else's flaws and waited for a manager to solve it.
I owned two of the eleven source connectors feeding our reporting warehouse. The engineer who reviewed that area was blunt in a way that landed badly on me — my changes came back with one-line notes like 'wrong grain, read the model' and then nothing for a day or two. I started routing around him and merging with a second approver, which made things worse: my orders connector went out with a join that duplicated rows, and we were carrying drift of about 4,180 rows a night against the source count. So I asked him for fifteen minutes. I opened by saying I thought his comments were just faster than I could read them rather than aimed at me, and asked how he would prefer I brought him work. It turned out he was reviewing for four teams and read changes in one batch each morning. We agreed I would put the grain and the reconciliation query in the description before he ever opened the diff, and that I would ping him once at midday instead of four times. Turnaround went from roughly two days to the same morning, and drift on my two connectors sat under thirty rows within a fortnight. He is still blunt. I stopped reading it as hostility, and I stopped shipping around review.
The signal is in the middle: a junior admitting the workaround made things worse, then testing a charitable read and asking for a protocol rather than for nicer behaviour. The drift number and the review turnaround give the ending teeth. It would downlevel if the conversation were replaced by advice from a manager.
I led ingestion on a fourteen-person programme with a six-month window to retire the old reporting stack. Partway through we had to cut scope, and the analytics lead on the consuming side treated every cut as something being done to her team — she stopped answering in the shared channel and took objections straight to the programme manager, which meant I was learning about her concerns second-hand and days late. My first instinct was to answer her objections in writing, point by point. That went nowhere; it just produced longer threads. So I stopped, booked a standing half hour with her, and changed what I brought: instead of a proposed cut list, I brought the reconciliation numbers for every mart and asked her to rank them for her analysts. Her top three were not the ones I had planned to keep. I also said plainly that going to the programme manager first was costing us a week per decision, and asked her to bring it to me first even when she disagreed. We kept nineteen marts instead of the twenty-eight originally planned, and the retained ones reconciled to source within 0.2 percent, against 1.6 percent on the stack we were retiring. She wrote the second cut proposal with me. We still argue, but it happens in the room now.
Feature-to-programme scope, a failed first attempt named out loud, and a durable change in how disagreement travelled — that is what puts this above a junior answer. The strongest move is handing the ranking to the other person. It would downlevel if the resolution were the manager imposing the cut list.
Your scope is your own task and your own reaction. Show that you checked your read of the person before deciding they were the problem, asked for one specific change in how you two worked, and kept delivering your piece meanwhile.
Expect feature-level scope and mechanics. Name what you tried, what failed, and the concrete working agreement you landed — a review protocol, a standing slot, a written handoff — and show the collaboration outcome in the delivery, not just the mood.
Own the team-level cost of the friction: what it was doing to delivery or to production quality, and what you put in place so the same clash did not recur with the next pair of people. Say how you protected the other person's standing while fixing it.
Talk about the pattern, not the person. Show that you read the structural cause — split ownership, a review bottleneck, mismatched incentives across teams — and changed the mechanism, while still handling the individual relationship directly and privately.
saying these in an interview costs you the question
- Labels the person toxic, lazy or incompetent and stops there
- No direct conversation — the manager is the first move
- Nothing changed on your side of the relationship
- Story resolves only because the person left or was reassigned
- Relitigates who was right instead of how it was worked out
- No result for the work, only for the feelings
- How do you think that person would describe working with you?Answer generously and concretely. Name one thing they would say you did well and one they would criticise, phrased as behaviour rather than a flaw of character. This is where interviewers catch a candidate who casts themselves as the reasonable party throughout; a specific, non-defensive self-criticism resolves it in a sentence.
- What did you try that did not work?Have one genuine failed attempt ready — routing around them, raising it in a group setting, over-explaining in writing. Say briefly why it failed and what that taught you about the person. A story where the first move worked perfectly sounds rehearsed rather than lived.
- What would you do differently if you faced that person again?Pick one change and be specific: go direct sooner, ask about their constraints before proposing a fix, put the agreement in writing. Avoid answering that you would do nothing differently, and avoid an answer that is really another complaint about them.
## One question, several wordings This is the most common relationship prompt in the whole behavioural set, and it appears in several wordings that all want the same evidence: - *tell me about a difficult person you have worked with* - *describe a personality clash you had to get past* - *tell me about a coworker you did not get along with* - and the softened *how do you handle someone who is hard to work with?* Treat them as one question. The only wording that changes your answer is the one that asks for someone you **still** had to work with afterwards — that version rules out stories where the relationship simply ended. ## What the interviewer is actually screening for They are not curious about your former colleague. They are trying to predict what happens on their team when you hit friction, and there are three failure modes they have seen: 1. the candidate who escalates immediately, 2. the candidate who suffers silently and lets delivery slip, 3. and the candidate who is themselves the difficult person and does not know it. Everything in a strong answer is aimed at ruling those out. ## Weak versus strong on the same story A **weak version** says: *I worked with someone who was really negative and blocked everything, so I raised it with my manager and eventually they moved off the project.* Nothing there is about you. A **strong version** of the identical episode says: *reviews on my work were coming back terse and slow, I was starting to route around him, and that was hurting the delivery — so I asked for fifteen minutes, found out he was reviewing for several teams, and we agreed on a format and a single daily ping.* Same facts, but now the interviewer can see a diagnosis, an action, and a change you made in your own behaviour. ## Describe behaviour, never character Say what a camera would have recorded: comments arriving hours late, being talked over in the design review, requests answered only when a manager was copied. The moment you say *toxic*, *arrogant* or *impossible*, the interviewer has only your judgement to go on, and half of them will privately wonder whether the difficult person in the room is you. ## Assume positive intent, out loud The strongest single sentence you can include is the hypothesis you tested: - they were overloaded; - they had been burned by a previous rewrite; - they had context you did not; - their manager was measuring them on something incompatible with what you needed. Say that you checked it, and say what you found — including when your charitable read turned out to be wrong. ## Go direct, and say so Interviewers listen specifically for a **private, first-person conversation** before any escalation, and for how you opened it. Quote your opening line if you can. If the story did end in escalation, the answer must still show what you tried first and that you told the person you were raising it. ## Land the ending on two things 1. **First a work result** — a defect avoided, a delivery date held, a quality number that moved. 2. **Second the state of the relationship**: still blunt but workable, or genuinely repaired, or civil and productive without warmth. Honesty here reads better than a fairy-tale ending; *we never became friends, but we shipped it and I would work with him again* is a completely acceptable close. ## How the bar moves - **Junior answers** may be entirely about one task and one conversation. - **Middle answers** should produce a durable working agreement. - **Senior answers** should account for the cost the friction imposed on the team and the prevention that followed. - **Principal answers** should identify the structural cause and change it, while still handling the human part in private.