Tell me about a time a teammate wasn't pulling their weight on a project you shared.
answer
- name the gap and what it cost
- diagnose skill, will, or context
- direct conversation before escalating
- what you changed: scope, pairing, checkpoint
- result plus where the relationship landed
basics
~20 sTests whether you address a delivery gap directly instead of silently absorbing it or escalating first. Diagnose skill, will, or context, talk to the person before anyone else, change the work, and show delivery recovered.
how to answer
6 beats- the shared commitment and the gap you sawName what you were both on the hook for, your own stake in it, and the observable gap — stalled work, missed handoffs, a date that slipped — with a count or a duration rather than an adjective. Keep this and the task framing to roughly fifteen to twenty percent of your airtime.
- what you checked before you concluded anythingSay how you separated skill, will and context: you read the actual work, looked at the history, or considered their load. Name the hypothesis you formed and the evidence for it. This beat is where diagnostic judgment shows, and it is the one most candidates skip.
- the direct conversation you had with themGive roughly the words you opened with — a question lands better than a verdict, and quoting yourself proves the conversation happened. Make clear you went to the person before you went to anyone else, and say where you had it, privately.
- what you changed about the workThis is the bulk of the answer, around sixty percent. Concrete moves only: pairing, re-splitting the scope, removing a blocker, a written definition of done, a checkpoint date, or handing part of the work elsewhere with agreement. Say who decided what.
- the delivery result and where the relationship landedOne number for delivery and one sentence for the person. The number ends the story at the outcome rather than at the conversation; the sentence shows you can be direct without leaving damage behind. Result and reflection together are about twenty to twenty-five percent.
- what you do differently nowTwo sentences at most. Name the earlier signal you would act on, or the norm you added so the same silence surfaces sooner. Avoid a generic lesson about communication being important.
your answer
5 story prompts- Pick a gap you actually raised with the person, not one you only complained about privately.
- Choose a case where you can name the cause: a missing skill, missing context, or genuinely missing motivation.
- Have one number ready — days recovered, work reassigned, or the delivery figure after your fix.
- This can be the same story as your missed-deadline answer, re-angled onto the conversation you had.
- Check your draft for pronouns: the actions should be I, the agreement can be we.
draft and rehearse your own answer in a learn session
go deeper
This probes ownership and interpersonal courage: whether you name a delivery gap early and directly, or quietly absorb it until a deadline exposes it. Interviewers also listen for diagnostic judgment, since a missing skill, a missing motivation and a missing piece of context need different responses. A strong answer proves you protected a shared commitment without humiliating a peer.
I was one of two contributors on the key-normalization milestone of an open-source data-ingestion toolkit, which I maintained part-time as part of my job. My co-owner had self-assigned three issues in that milestone, and over five weeks none of them moved — no draft pull request, no comment — and the null-key rate on our reference corpus was stuck at 6.4% because his piece gated mine. My first instinct was that he had lost interest, so I made myself check before deciding. I read his last few merged changes, and the work was good. Then I noticed all three stalled issues needed the integration fixtures, which only ran if you had a local sample warehouse. That looked like a setup problem, not a motivation problem. I messaged him directly and led with a question rather than a verdict: these three have not moved in a month, is something blocking you, or should we re-split them? He said he had spent two evenings failing to get fixtures running and had not wanted to admit it on a public issue. We paired for about forty minutes, I published a downloadable fixture bundle with a short setup note, and we broke the largest issue into three smaller ones so progress would be visible. He shipped the first two within nine days. The null-key rate on the corpus came down to 0.9% before we cut the release, and he is still contributing. Now, when something I depend on goes quiet for two weeks, I ask in week one.
The signal sits in the check-before-concluding beat: he formed a wrong hypothesis, tested it, and found a context blocker. The opening line of the conversation is quoted, which makes it credible. It would downlevel if the fix were only the pairing session with no change to how the work was split or measured.
I was the maintainer lead for the schema-validation subproject, with twenty-two active contributors and a six-week release train. One long-standing reviewer owned the dedupe area, and review turnaround there had drifted from under a day to roughly nine days. Two correctness regressions reached a release candidate because nobody with real context had looked at them. I treated this as a delivery problem before a people problem. I pulled review-latency figures for each area, confirmed the slowdown was one area and one person rather than a general trend, then called him instead of posting anything in public. He had picked up an on-call rotation in his day job and his volunteer hours had evaporated. Capacity, not commitment, and he was relieved someone finally asked. We renegotiated in the open. He kept design review on new proposals, where his judgment mattered most, and handed the dedupe queue to two newer contributors with a written handover and a fortnight of shadowing. I also stood up a four-person review rota with a stated three-day target, so no critical area sat one volunteer deep again. Median review wait in that area settled around a day and a half, and null-key regressions found after a release fell from five in that cycle to one across the following two. He is still a maintainer, and he is the one who trains the rota. The permanent change: everyone states the hours they can give at the start of a cycle, so a drop shows up as a number instead of as silence.
Sequencing is what puts this at senior — evidence first, private conversation second, structural fix third — and the fix outlives the incident. The capacity declaration is the beat that a mid-level version of this story would not contain. It would downlevel if the answer stopped at the renegotiated split.
Task scope is enough. You noticed a handoff you depended on was late, asked the person directly with facts rather than assumptions, and told your lead once your own attempt did not move it. Nobody expects you to redesign the work split.
Feature and peer scope. Show the diagnosis — why they were stuck — and the concrete renegotiation that followed: re-split scope, paired, set a visible checkpoint. The interviewer is listening for whether you kept your shared commitment honest without going around the person.
Team and delivery scope. You spotted the pattern before the deadline exposed it, protected the release, rebalanced ownership so the dependency was not one person deep, and changed how work gets assigned so the same silence surfaces earlier next time.
The individual case is an instance of a system problem. Talk about where accountability was ambiguous across teams, the mechanism you introduced for surfacing stalled work, and how you coached other leads to have the conversation in week two rather than month three.
saying these in an interview costs you the question
- Absorbing the extra work in silence and calling that teamwork
- Going to their manager first, without ever talking to the person
- Diagnosing laziness from behavior alone, with no conversation
- Describing the teammate's character instead of the specific missed work
- No result — the story ends at the conversation and never reaches delivery
- Venting about the person rather than describing what you changed
- How did their manager get involved?Show sequence, not avoidance. The strong shape is: you talked to the person, gave the change a fair window, and then brought the manager in with facts and without ambush — ideally telling the teammate you were doing it. Never involving anyone reads as conflict avoidance if delivery kept slipping; going straight there reads as escalation reflex.
- What would you do differently?Pick one real thing, usually timing. Most people wait too long because they are searching for proof rather than asking a question. Say which earlier signal you would act on now, and what you would ask sooner. Avoid answering that you would have escalated faster — that usually says you skipped the diagnosis.
- How did the working relationship end up afterwards?Interviewers use this to test whether you can be direct without burning people. Give evidence rather than a claim: you kept collaborating, they asked you to review later work, they told you afterwards that the conversation helped. If it did stay cool, say so honestly and name what you would have done to soften it.
- What if the conversation hadn't changed anything?Answer with an escalation ladder, not a threat. Second conversation with a specific expectation and a date, then bring in the person who owns the assignment — a lead or manager — with the record of what you already tried, then protect delivery by reassigning the work. Say plainly that you would not let a shared commitment fail quietly to avoid awkwardness.
## One question, many wordings This prompt appears in almost every collaboration loop, and it is one of the few where the interviewer is listening as much for what you did *not* do as for what you did. Common wordings: - a teammate who was not pulling their weight; - a time you worked with someone who consistently missed deadlines; - a time you had to carry someone else's work; - how you handle a peer who is not delivering. They are the same question — pick one story and let the wording steer the emphasis. ## The evaluation has three axes ### Courage The first is **courage**: did you say the thing to the person's face, early, or did you route around them — absorbing the work, quietly rewriting their code, or going to a manager as the first move. - Absorbing is the most common weak answer, and candidates often tell it as a strength ("I just picked it up and shipped it"). - It reads as an inability to have a normal adult conversation, and at senior level it also reads as a delivery risk, because the gap is now invisible to everyone above you. ### Diagnosis The second axis is **diagnosis**. Skill, will, and context are three different problems with three different fixes, and strong candidates say which one they were looking at and how they checked. - A **skill** gap is fixed by pairing, a smaller first slice, or a different assignment. - A **context** gap — no environment, unclear definition of done, an unstated dependency, a load they never mentioned — is fixed by removing the obstacle, and it is by far the most common real cause. - **Will** is the rarest, and candidates who reach for it first without evidence look uncharitable. The line that lands well is some version of: I assumed I was missing information until I had asked. ### Outcome ownership The third axis is **outcome ownership**. The story must end at delivery, not at the conversation. - Weak versions finish with "we had a chat and things improved." - Strong versions name what actually changed in the work — a re-split of scope, a written definition of done, a checkpoint date, a review rota — and then give one number: days recovered, the delivery figure after the fix, the volume of work that moved. One number is enough; two is better than an adjective. ## The evidence you can offer shifts by level - At **mid level**, the credible evidence is the mechanics of the collaboration: what you noticed, what you asked, what you both agreed, what shipped. - At **senior level**, interviewers expect a second layer — the structural change that stopped the situation recurring, because a senior engineer who solves the same interpersonal problem five times has not learned anything. - At **principal level** the individual case becomes an illustration of an ownership or assignment problem across teams, and the interesting content is the mechanism and the coaching, not the anecdote. ## Two traps are worth naming 1. The first is **relitigating fault in the room**: the interviewer cannot verify who was right, and time spent proving the teammate was wrong is time not spent on your own judgment. 2. The second is **over-anonymizing to the point of vagueness** — you can and should strip names and identifying details, but keep the specifics of the work, the timeline, and the numbers. A story with no texture sounds invented, and a story that is mostly complaint sounds worse than one that is mostly repair. ## Finally, watch your pronouns This story tends to slide into "we" because it is inherently about two people. Keep the actions in the first person — I checked, I asked, I proposed — and reserve "we" for the agreement you reached together. The interviewer is scoring you, and can only score what you claim.