Tell me about a time you pushed back on a deadline or scope a stakeholder insisted on.
answer
- name the ask and who owned it
- state the evidence, not the feeling
- restate their goal in their words
- offer two options with real tradeoffs
- close on what shipped and when
basics
~10 sTests whether you can disagree with the business without stonewalling it. Name the commitment, show the evidence it would not hold, and leave the stakeholder choosing between real options rather than facing a wall.
how to answer
6 beats- the commitment and who was on the hook for itOpen with the date or the scope, who had promised it, and what the business outcome behind it was. Keep this and the next beat to roughly fifteen to twenty percent of your airtime — the interviewer needs orientation, not the whole org chart.
- what told me it would not holdState the evidence in one or two sentences: a measurement you took, a decomposed estimate, a base rate from similar work, or a named risk with its cost. Say how you got the number, because the credibility of the whole story rests here.
- how I opened the conversationShow that you led with their goal, not your objection — what decision the date was serving, what happens to them if it slips. This beat and the next carry the bulk of the answer, around sixty percent, because the how is what is being assessed.
- the two ways forward I handed themGive two or three concrete shapes with the cost of each attached, and say which one the stakeholder chose. Options with no cost attached are not options; the point is that the decision stayed theirs while the information became yours.
- what we shipped and what it costLand the result with a number and a date, including what did not happen because of the trade. Result and reflection together should take twenty to twenty-five percent — a story that ends at the meeting is the commonest structural failure here.
- the conversation I now have soonerClose with the change you made so the same conversation happens sooner or not at all — sitting in the planning review, attaching estimates to requests, measuring before a date is quoted. One sentence, no moralising.
your answer
5 story prompts- Pick a commitment you personally argued against, ideally within the last eighteen months, where you can name the date.
- Write down the one number you used as evidence: a measurement, a decomposed estimate, or a costed risk.
- Name the two options you offered and what each one actually cost the stakeholder.
- This can be the same project as your missed-deadline story, re-angled to the conversation before the slip.
- Check you can say how that working relationship stood a month later.
draft and rehearse your own answer in a learn session
go deeper
This prompt probes influence without authority and judgment under commercial pressure. Interviewers want evidence that you treat a stakeholder's goal as legitimate, that your disagreement rests on measurement rather than preference, and that you hand back a choice instead of a refusal. A strong answer shows you can protect delivery quality and the working relationship in the same conversation.
I owned the marketing attribution pipeline on a seven-person data team at an early-stage company. Our head of growth had told the board that a channel-level dashboard would be live by the end of the quarter, and the spec had nine fields — three of which needed session stitching rebuilt from scratch. Before I said anything I ran the numbers. With the joins that spec implied, the nightly backfill runtime went from four hours nine minutes to just over eleven hours in my test run, which meant the dashboard would be stale every morning for exactly the people meant to use it. So I asked for twenty minutes with her, and I started with her goal rather than my objection: she needed channel spend decisions on Monday mornings. Then I showed the timing and gave her two options — six of the nine fields on her date with a refresh that lands before seven, or all nine a cycle later once we partitioned the session table. She took the six fields, and asked that the remaining three be scheduled explicitly rather than left as a promise. We shipped on her date, the refresh landed at 6:40 in the morning, and the last three fields followed five weeks later. What I changed afterwards is that I run the cost estimate before the commitment gets made in a room I am not in — I now sit in that planning review.
The signal is carried by the measurement taken before the conversation and by opening with the stakeholder's Monday decision rather than the engineering objection. The two costed options keep the choice with her. It would downlevel if the numbers were absent or if the story ended at her agreeing, with no shipped date and no follow-through on the deferred fields.
I led the data platform group when our revenue lead wanted six months of restated history behind a new pricing model, in the same cycle as the pricing launch itself. I did not open with no. I asked what decision the history was for, and it turned out finance needed only the last two closed months to sign off the model; the older history was for a board narrative a cycle later. That reframing was most of the work. Then I put the cost in his terms. A full restatement meant re-running every downstream model, and our measured chain took thirteen hours forty on a good night, so one mistake cost two days of reporting and the freshness commitment we had already made to sales. I offered three shapes: the two closed months on his date; the full history a cycle later; or the full history on his date if we paused the churn-model work for four weeks — and he made that trade with the other team himself, rather than me absorbing it quietly. He took the first, and wrote the trade into the planning doc instead of settling it in a hallway. We hit the pricing date. Afterwards I changed our intake so anything touching historical data arrives with a runtime estimate attached before a date is promised to anyone.
Senior here means uncovering the real decision behind the request, pricing the risk in the stakeholder's own commitments, and refusing to absorb a cross-team trade privately. The intake change shows prevention. It would downlevel to mid if the third option had been dropped or if the trade had been made silently inside the team.
Your scope is your own task, and that is enough. Show that you raised the risk to your lead or PM early, with your own estimate or a quick measurement behind it, instead of quietly absorbing it and missing the date.
You should be running the conversation yourself, not relaying it. The bar is a feature-sized scope, evidence you gathered before speaking, and at least two concrete options you put in front of the stakeholder with the cost of each spelled out.
Expect to be judged on protecting a team and a quarter, not a ticket. Show that you traded something explicitly, that the stakeholder made the call with full information, and that you changed how the next commitment gets made.
Talk about the commitment-making process itself. The signal is that you fixed the pattern across teams — how estimates enter planning, who is in the room when dates are promised — not that you won one negotiation well.
saying these in an interview costs you the question
- Framing the stakeholder as someone who simply did not get engineering
- Pushing back with a feeling rather than an estimate, a measurement or a named risk
- Saying no with no alternative and no partial scope on the table
- Escalating to a manager before ever speaking to the stakeholder directly
- Winning the argument while leaving the relationship or the roadmap damaged
- Ending the story at the meeting instead of at what actually shipped
- What would you have done if they had said no anyway?Answer with a real fallback, not defiance. Say what you would have done to reduce the risk inside the constraint you were given — cut scope you controlled, added a checkpoint, put the risk in writing so the decision was informed. Interviewers are checking that you can lose an argument and still deliver professionally, without sabotage or silent resentment.
- How did the stakeholder react in the moment?Be honest, including if it was tense. Describe what you did with the reaction: slowed down, asked what was driving the date, separated their pressure from yours. A candidate who reports a friction-free reaction to every pushback sounds rehearsed; one who names discomfort and then shows how the conversation recovered sounds like they have actually done it.
- What did you get wrong in your original estimate?Take the invitation. Naming a real error in your own numbers strengthens the story, because your whole case rested on evidence. Say what you missed, how you found out, and what you now measure before quoting a figure. Refusing to concede anything here reads as the same rigidity the question is probing for.
This prompt shows up in several shapes, and they are one question: *tell me about a time you disagreed with a product manager*, *how do you handle an unrealistic deadline*, *describe a time you had to say no to the business*, and *tell me about a time you pushed back on a requirement*. Prepare one story and re-angle the opening sentence to whichever shape arrives. ### What the interviewer is actually sorting for They are separating three kinds of engineer. The first absorbs everything, agrees to the date, and the failure surfaces later as a missed commitment — this candidate is pleasant and expensive. The second refuses on principle, treats the deadline as an insult to craft, and produces a story where the stakeholder is the antagonist. The third treats the date as a legitimate business input, produces evidence, and hands the decision back with the tradeoffs made visible. Only the third one gets the strong signal, and the tell is almost always whether the stakeholder's goal appears in the answer at all. ### Evidence beats conviction, and there are only a few kinds of evidence The strongest answers rest on something checkable, and it is usually one of four things: a measurement you took (a timed run, a load test, a sample), a decomposed estimate (the work broken into parts a skeptic can argue with piece by piece), a historical base rate (the last three things shaped like this took this long), or a named risk with a probability and a cost (if the migration reveals a data mismatch, we lose a reporting cycle). Anything vaguer than these — it feels tight, we always underestimate, the code is not ready — is heard as reluctance rather than analysis, no matter how correct you turn out to be. ### Weak versus strong, in one contrast Weak: *The PM wanted it in three weeks and it clearly needed six, so I escalated to my manager and eventually we got more time.* Nothing here is checkable, the stakeholder is a force of nature rather than a person with a reason, and the resolution belongs to someone else. Strong: *She needed the dashboard live for a Monday spend decision. I measured the run first, and the refresh landed after the meeting she needed it for. So I brought her two options — a narrower version on her date, or the full version a cycle later — and she picked. We shipped the narrow one on time and the rest followed.* Same disagreement, but the business goal is named, the evidence is real, and the stakeholder retains the decision. ### The move that separates senior from mid Mid-level answers negotiate the item. Senior answers negotiate the trade out loud: naming what will not happen if this does, and making the stakeholder own that exchange rather than absorbing it privately. If you are interviewing above mid-level, make sure one sentence in your answer is a cost you said aloud to the person who had to pay it — and one sentence about what you changed so the same argument would not be needed again. ### Two failure modes in the delivery itself First, spending ninety seconds on why the deadline was absurd; the setup should be short and the conversation should be the bulk. Second, ending on the meeting rather than the outcome. Interviewers need the last beat: what shipped, when, and how the relationship stood afterwards.