Tell me about a time you knew something your manager did not, and had to change their mind.
answer
- name the decision and the stake
- the evidence you brought
- raised privately, framed in business terms
- the decision that was actually made
- what you did once it was decided
basics
~10 sProbes candour and judgement about where dissent belongs. Answer with one real disagreement, evidence rather than opinion, raised privately first, a decision that actually got made, and you committing to it afterwards.
how to answer
5 beats- the decision you disagreed with, and what was at stakeOpen with what your manager decided or proposed and the concrete consequence you were worried about. Keep this and your role in it to roughly fifteen to twenty percent of your airtime — enough for the listener to feel the stake, not a tour of the org.
- what you knew that they did notState the evidence in one or two sentences: a measurement, an estimate with its assumptions, a customer signal. Vague concern is the most common reason this answer fails, so put a number or a specific failure mode here.
- how and where you raised itSay the room out loud — a one-to-one, a direct message, before the wider meeting — because interviewers score venue as much as content. Frame the argument in the terms your manager is judged in, and attach a sized alternative rather than a pure objection.
- the decision that got made, and the resultThis beat plus the reflection should be about a quarter of your airtime, and the action beat before it the bulk. Name who decided, what was decided, and the outcome with one number. Losing is fine; ending without a decision is not.
- what you did once it was decidedShow commitment: you executed it, you did not relitigate it downward, and if you set a revisit trigger you say what it was. Close with one sentence on what you would do earlier or differently next time.
your answer
4 story prompts- Pick a disagreement with a manager where a decision was actually made and shipped.
- Choose one where you lost the argument and still executed it cleanly.
- Write down the single number or fact you brought into the conversation.
- Note the room you raised it in, and why that room.
draft and rehearse your own answer in a learn session
go deeper
Managers hire people who will tell them the truth without making it expensive. This probes candour, judgement about when and where dissent belongs, and whether you separate the decision from the person. A strong answer shows an argument made with evidence, in private first, a decision that actually got made, and clean execution afterwards.
At my previous company, a security startup with about a dozen engineers building phishing-simulation tooling, my manager asked me to ship a campaign-scheduler change without a per-campaign send cap. His reasoning was that the cap was a nice-to-have and a customer demo was already on the calendar. I disagreed, and I raised it in our one-to-one rather than in the team channel. I said I might be wrong about the priority, but here is what worried me: if a customer points a campaign at their whole directory by mistake, we send thousands of simulated phishing emails in one burst and their help desk finds out before we do. I brought the number — our largest tenant directory held just over nine thousand addresses — and I proposed the smallest version I could build rather than the full feature: a hard cap of two hundred sends per hour per campaign. He pushed back on the cost, so we agreed I would timebox it to one day and drop it if I overran. It took about seven hours. Eleven days after the demo, a customer did upload their full directory by accident, and the cap held. What I took from it is that pushing back lands much better with a cost estimate and a smaller option attached than as a concern on its own. I also learned to raise it privately first — he was far more receptive one-to-one than he would have been with the team watching.
The signal carries in three places: the disagreement is raised in a one-to-one, it comes with a sized alternative instead of a veto, and it ends with a decision and a verified outcome. Dropping the day-and-a-half cost estimate, or ending at 'he agreed', would downlevel it to an opinion story.
I owned reporting for the phishing-simulation product at a startup of roughly a dozen engineers, and my manager and I disagreed about how we computed the number customers actually looked at. He wanted each campaign's click rate reported on its own, which made one account's trend read as a fall from 12.6% to 4.1%. The problem was that the second campaign had gone to a smaller cohort that had already been through training twice, so most of the improvement was selection, not learning. I disagreed because that chart was about to appear in a renewal conversation, and I said so to him privately before it went anywhere near the customer. What I brought was a rerun of the same data on a consistent cohort: the honest movement was 12.6% to 9.3%. Still a good story, just not a spectacular one. His objection was real too — the honest chart was harder to explain inside a fifteen-minute call. So what we shipped was both: the cohort-consistent trend as the headline, per-campaign detail one click down, and a line of copy explaining the difference. The renewal closed. It became a product feature, and two other accounts asked for the cohort view inside the quarter. What I learned is that the win came from taking his objection seriously rather than treating my number as self-evidently correct. The explanatory copy was his idea, and it is the part customers mention.
Strong because the disagreement is settled by rerunning the data, raised before the number reached a customer, and resolved by absorbing the manager's objection into the fix rather than defeating it. It would downlevel if the ending were 'he saw I was right' with no shipped artefact and no number.
Your scope is your own task, and that is fine. Show that you raised it at all, in the right place, with a cost estimate attached — and that you executed cleanly once the call was made.
Work at feature scope and show mechanics: you brought data your manager did not have, you offered an alternative with its tradeoffs, and you took their objection seriously rather than out-arguing them.
Frame the disagreement in team and production terms — risk, cost, customer impact — and give options rather than a veto. Something should have changed structurally afterwards, not just this one decision.
Disagreeing upward is routine at your level, so the interesting part is the mechanism: written briefs, a risk register, a decision forum. Include a decision you lost and carried anyway without leaking dissent downward.
saying these in an interview costs you the question
- Picking only disagreements you won
- Calling the manager's decision stupid or careless
- Raising it in a group channel before raising it privately
- No decision at the end — the story just trails off
- Relitigating who was right instead of saying what happened next
- A disagreement so trivial it shows no judgement, like formatting style
- What did you do when the decision went against you?This is the real question behind the prompt. Say plainly that you executed the decision, and give evidence: you did not slow-walk it, did not relitigate it in a group channel, and did not tell your team you thought it was wrong. If you set a checkpoint to revisit it with agreed criteria, name that — it shows commitment without pretending you changed your mind.
- How did your manager react in the moment?Answer honestly, including if the first reaction was defensive. Interviewers hear a lot of implausibly gracious managers. Describe what you did with the reaction — gave them a night, came back with a smaller ask, moved from opinion to numbers — rather than grading their behaviour. Avoid any word that makes them sound irrational.
- When do you decide it is worth pushing back at all?Give a threshold, not a personality trait. Most people use some mix of reversibility, blast radius, and cost of being wrong: cheap and reversible decisions get one comment and your commitment, expensive or irreversible ones get a written case. Saying you push back on everything reads as noisy; saying you never do reads as passive.
## One story for every wording This prompt shows up in almost every loop, and it is asked in half a dozen wordings that all want the same evidence: - "Tell me about a time you disagreed with a decision from above", - "a time you were overruled", - "a time you had to give your manager bad news", - "how do you handle it when you think your lead is wrong". Prepare one story that survives all of them, and adjust only the opening sentence. ## What the interviewer is actually screening for Two failure modes get people rejected, and they are opposites. - The first is **the pushover**: someone who saw the problem, said nothing, and watched it happen. - The second is **the litigator**: someone who was right, made sure everyone knew, and left a manager who now routes around them. The hire is the person in the middle — argues hard, argues in the right room, and then commits. Your story has to prove the middle, which means it needs both a real argument and a real ending. ## Where you raised it is scored, not just what you said **Private first** is the default expectation: a one-to-one, a direct message, a walk. Escalating in front of an audience is read as pressure rather than persuasion, and it is one of the few things that can sink an otherwise strong answer. If your story genuinely began in a group setting — a design review where the question was on the table — say why that was the right room, because otherwise the listener assumes you ambushed someone. ## Evidence types, roughly in order of how well they land - A measurement from your own system beats an estimate; - an estimate with its assumptions written down beats an analogy; - an analogy beats an appeal to best practice; - "it felt wrong" is not evidence. The single strongest move available to a junior or mid candidate is **bringing a cost**: not "we should add rate limiting" but "a day of work, here is what it prevents". Managers say yes to sized asks far more often than to concerns. ## Business framing Translate the technical claim into the currency your manager is judged in — customer trust, a renewal, an on-call burden, a date. This is not corporate ventriloquism; it is the difference between asking your manager to take your word and giving them something they can repeat to their own boss. A good test while rehearsing: could your manager relay your argument accurately in a meeting you are not in? ## Weak versus strong endings - The **weak version** ends "and they realised I was right". It is unfalsifiable, it makes the manager the fool, and it tells the interviewer nothing about how you behave when you lose. - The **strong version** ends with a decision, a consequence with a number in it, and one sentence about what you carried forward. Losing is entirely acceptable material — arguably better material — as long as you executed the decision properly and, if it went badly later, you did not spend the aftermath collecting credit. ## How the bar moves with level - At **junior**, the whole signal is that you spoke up in the right room with a cost attached. - At **mid**, the signal is craft: alternatives, tradeoffs, taking the counter-argument seriously enough that the final answer contains a piece of it. - At **senior**, one decision is not enough — the interviewer wants the change to the process that made the next disagreement cheaper. - At **principal**, they are listening for how you disagree at scale without destabilising the org: written positions, a forum where these calls get made, and visible discipline about carrying decisions you lost. ## Rehearsal check Time it. 1. Situation and stake in about twenty seconds, 2. the argument and how you ran it as the bulk, 3. the outcome with a number, 4. and roughly fifteen seconds of reflection. If your setup is longer than your action, cut the backstory — nobody has ever been downlevelled for under-explaining the org chart.