In this article
Before you start
Why interviewers ask this
The question tests whether you can disagree upward and sideways without damaging the relationship or the decision. A leader from another org has no authority over your team but can slow your plan, so you cannot rely on rank. Interviewers listen for two failures: defending the plan before understanding the concern, and giving way with no check. A weak answer says "I would stay calm and explain my approach". A strong answer shows how you find the real concern, test it, and end with a named decision.
The question can arrive as a story prompt, "tell me about a time a leader challenged you", or as a hypothetical like the one above. The leadership interviews hub places it among the other org-level questions.
A senior leader challenging your approach, from first principles
- Disagree and commit
- Disagree and commit is a decision rule in which people who still disagree after a decision is reached agree to back it fully, so the group can move without waiting for consensus.
- Principled negotiation
- Principled negotiation is a method of negotiating on the merits of an issue that separates the people from the problem, focuses on underlying interests rather than stated positions, and judges options against objective criteria both sides accept in advance.
In plain words
A challenge in a meeting is like a smoke alarm. It goes off because something worries the alarm's owner. You do not argue with the alarm. You check what set it off. The analogy breaks in one place. A smoke alarm has no interests of its own, while the leader does, and their reasons may have little to do with your plan.
Stated precisely, the position is what the leader said, such as "this duplicates our service", and the interest is what sits underneath, such as "I do not want to support two systems". The Program on Negotiation (PON) at Harvard Law School puts it this way: "Positions are surface-level statements. Beneath them lie interests—the underlying needs, motivations, and concerns" (PON, June 2026). Your job is to answer the interest.
Step 1: Restate the challenge before you answer it
The first step is to say back what you heard, in your own words, until the leader agrees you have it. The input is the challenge; the output is a restated concern the leader has confirmed. Do not defend, explain or correct yet. The Program on Negotiation, summarising Fisher, Ury and Patton, lists "separate the people from the problem" as the first element of principled negotiation, so that emotions and relationship friction are handled apart from the substance (PON, June 2026). Restating does that: it treats the leader as a person with a concern and the plan as a separate thing under review. What you now know: the leader feels heard and you know the exact claim.
Step 2: Find the interest behind the position
The second step is to ask what outcome the leader is worried about. The input is the restated claim; the output is the interest. A question such as "what would make you comfortable with this?" turns a verdict into a request. Positions and interests often differ: the leader may say "do not build this" when the worry is support load, risk to their roadmap or a past incident. The PON article defines interests as "the underlying needs, motivations, and concerns driving those demands" (PON, June 2026). Write the interest down in one line, and check it with the leader. What you now know: you are solving the leader's worry, not winning the meeting.
Step 3: Test the concern with evidence
The third step is to separate what is known from what is assumed and to offer a check. The input is the interest; the output is a test both sides accept before it runs. PON's fourth element is "insist on objective criteria": decisions grounded in fair, independent standards that both sides accept in advance (PON, June 2026). In a work setting that means a measurement, a pilot or a data pull with an agreed pass mark. Say plainly what you have not yet measured. Amazon's leadership principles say leaders "seek diverse perspectives and work to disconfirm their beliefs" (Amazon Jobs, read October 2026). What you now know: the disagreement is now a question with a deadline.
Step 4: Agree the decision owner and the next step
The fourth step names who decides, and when, so the disagreement ends in a decision. The input is the test and its date; the output is one named decision owner and a next step. A decision owner is the single person who makes the final call after hearing the others. Atlassian's framework of Driver, Approver, Contributors, Informed (DACI) makes the same point: the approver is "the one person (yes: one!) who makes the decision", and contributors have "a voice, but not a vote" (Atlassian, read October 2026). Propose the owner without pulling rank, for example your shared manager or the leader of the affected service. What you now know: the answer ends in a decision, not a stalemate.
Step 5: Commit, then report back
The fifth step is to back the decision and follow up. The input is the decision; the output is a visible result sent to the leader. Jeff Bezos describes the phrase this way: "Look, I know we disagree on this but will you gamble with me on it? Disagree and commit?" (Amazon 2016 shareholder letter, April 2017). If the decision goes against you, commit fully.
If it goes your way, share the result and credit the check. Google's team-effectiveness research lists psychological safety, "a shared belief held by members of a team that the team is safe for interpersonal risk taking", first among five dynamics (Google re:Work, originally 2015); reporting back keeps the door open. What you now know: the relationship ends stronger than it began.
Step-by-step description
- Say back what you heard.
- Ask what outcome worries them.
- Agree a check before it runs.
- Name who decides and when.
- Commit, then share the result.
- Restate the challenge before you answer it.
- Find the interest behind the position, then solve for it.
- Name one decision owner and a date.
Knowing it is half the job. Saying it is the other half.
Interview Clozo is an AI interview copilot for real interview rounds: it hears the question and puts a strong answer on your screen, and you say it in your own words. See the features or how it works.
Start Free Trial →Worked example
Two illustrative answers to the same prompt show the difference. Both are composites written for this page, with no real person or company, and every detail is made up. The prompt: a director from another org has challenged your approach in a planning meeting. Word counts were computed in assets/data/answer-timing.py, and speaking times use the 152 to nearly 170 words per minute that Yuan et al. (2006) report for English conversation topics.1
The weak answer rests on rank and process: the plan was approved, so the challenge is a nuisance. It restates nothing, tests nothing and ends by passing the problem to a manager. The strong answer takes five moves in order. It restates, asks for the interest, admits an unmeasured gap, offers a one-week test, names one decider and then follows up. The strong answer also changes the plan in a visible way, which is what the interviewer remembers.1
Text description of the diagram
- The strong answer is about twice as long as the weak one.
- The strong answer still fits in under a minute.
| Fact | Weak answer | Strong answer |
|---|---|---|
| Words | 59 | 116 |
| Seconds at 170 words per minute | 21 | 41 |
| Seconds at 152 words per minute | 23 | 46 |
| Restates the concern | No | Yes |
| Finds the interest | No | Yes |
| Tests with evidence | No | Yes |
| Names one decision owner | No | Yes |
| Follows up with results | No | Yes |
The same prompt is scored on a ladder. A level is the seniority band the interviewer is hiring for, and the dimension that moves most between bands is who owns the decision.
| Dimension | Weak | Strong | Staff-level |
|---|---|---|---|
| Hearing the challenge | Defends the plan at once | Restates the concern and asks for the interest | Also finds who else the concern affects |
| Evidence | Cites approval, not data | Offers a short test and admits what is unmeasured | Agrees the pass mark before the test runs |
| Decision | Passes it to a manager | Names one decision owner and a date | Sets the owner and the rule for future clashes |
| Relationship | Treats the leader as an obstacle | Credits their concern, reports back | Shares the result with both teams |
| Failure to avoid | Rank as the argument | Yielding with no check | Over-process for a small call |
Trade-offs and edge cases
The first case is when the leader is right. A challenge that holds up should change your plan, and saying so is a strength. Amazon's principle says leaders are "right, a lot" partly by seeking out differing views and working "to disconfirm their beliefs" (Amazon Jobs, read October 2026). Treat the test as a real test, and say in the interview what you changed.
The second case is when the test does not settle it. Then the decision owner decides, and you commit. The rule has limits: Amazon's "Have Backbone; Disagree and Commit" principle says leaders "respectfully challenge decisions when they disagree, even when doing so is uncomfortable or exhausting" and then commit wholly (Amazon Jobs, read October 2026), so commit after a real challenge, not instead of one. How far to escalate varies by company, so ask the owner for a date rather than going over anyone's head.
The third case is when the challenge comes in public or with an audience. The same five steps apply, and restating works especially well in front of others. You can offer to continue the detail one-to-one after the meeting, as long as you give the owner and the date aloud.
Defending protects the plan but loses the relationship. Yielding protects the relationship but loses the decision. Escalating hands the decision to others. Inquiring, testing and agreeing an owner takes a short delay and keeps both.
Text description of the diagram
- Defending protects the plan but loses the relationship.
- Yielding protects the relationship but loses the decision.
- Escalating hands the decision to others.
The wider craft of working with people you cannot instruct is in influence without authority and driving a big cross-team change.
How to answer it in an interview
A senior leader from a different org challenges your approach in a meeting. What do you do?
"I would first make sure I understand the challenge: I would restate it and ask what outcome worries them, because a challenge usually sits on an interest such as risk to their team. Then I would separate what we know from what we assume, offer a quick check or a small pilot, and say what I do not know. I would propose who decides and by when. If the decision goes their way I would commit fully, and either way I would report back with the results."
The spoken structure is: restate and ask for the interest, test it, name who decides, then commit and report back.
Common mistakes
- Defending first. (wrong) "That was already approved." (right) "Let me make sure I have your concern right."
- Making it about rank. (wrong) "They are senior, so I would just agree." (right) Treat the person as a stakeholder and the plan as the thing under review.
- Giving way with no check. (wrong) "I would change the plan." (right) "I would test their concern on real data before changing anything."
- Escalating over their head first. (wrong) "I would tell my manager to handle it." (right) "I would propose a decision owner and a date."
- Ending at the meeting. (wrong) The story stops at the decision. (right) Add the follow-up: what you sent back and what changed.
Follow-up questions
How do you work through a disagreement?
Use the same five steps with a different example: restate, find the interest, test, name who decides, report back. Keep the example to one meeting and one result, and tell it in the STAR order covered in the STAR method.
What do you do when a peer manager has different priorities?
Priorities clash when two interests compete for one scarce resource. Say what each of you needed, what objective criterion you used, such as user impact or cost, and who decided. The engineering management interviews guide covers the wider set of manager-level stories.
What if the senior leader is simply wrong?
Run the test anyway, because the evidence is what persuades. If the test shows they are wrong, share the result neutrally and credit their concern for prompting the check. The Amazon leadership principles page explains how this style of disagreement is judged.
What if the challenge comes in front of your whole team?
Restate the concern calmly, thank them for raising it, and offer the next step with an owner and a date. Do not argue the detail in front of your team; continue it one-to-one. See also tell me about yourself for how to frame your own scope in the round.
Tell me about a time you genuinely messed up in cross-team collaboration?
Pick a real miss and name the step you skipped, its cost and the change: "I committed my team's date before asking the partner team what they needed, and their work slipped a sprint; now I confirm the interest and the owner first." The behavioral interviews hub covers the neighbouring question types.
- Restate the challenge before you answer it.
- Find the interest behind the position, then solve for it.
- Test the concern with a check both sides accept.
- Name one decision owner and a date.
- Commit, report back, and keep the relationship.
Sources
- Principled Negotiation: Focus on Interests to Create Value — Program on Negotiation, Harvard Law School (Katie Shonk, summarising Getting to Yes), 30 Jun 2026
- DACI: a decision-making framework — Atlassian Team Playbook (date = read date; page shows only a 2025 copyright), 24 Oct 2026
- Amazon leadership principles — Amazon Jobs (no page date shown; date = read date), 24 Oct 2026
- 2016 Letter to Shareholders — Amazon (Jeff Bezos; published 17 April 2017), 17 Apr 2017
- Understand team effectiveness — Google re:Work (originally published 2015), 1 Jan 2015
- Using the STAR method for your next behavioral interview — MIT Career Advising and Professional Development (date = read date), 24 Oct 2026
- Towards an Integrated Understanding of Speaking Rate in Conversation — Interspeech 2006 (Yuan, Liberman and Cieri; University of Pennsylvania), 17 Sep 2006
