Tell me about a time two people each insisted their request was your top priority.
answer
- both requests, and what each protects
- the comparison you actually ran
- recommendation, not a question, upward
- what the deferred side got instead
- the result, then the fix that followed
basics
~20 sProbes whether you settle a priority collision with evidence and a clear decision-maker rather than quiet heroics. Name both requests and their real stakes, show the comparison you ran, and say what the deferred side got instead.
how to answer
6 beats- the two requests and who wanted eachSet both up in a few sentences with roles rather than names, and make clear they were genuinely both reasonable. Setup and framing together deserve roughly a fifth of your airtime.
- why they collidedName the scarce thing: one of you, one release slot, one shared component. Without a stated constraint the story sounds like poor planning rather than a real prioritization call.
- the comparison you ranSay what each request protected — a contractual date, a demo, a live defect — and put a rough effort figure next to each. This and the next beat carry most of your airtime.
- the decision and who made itBe exact about your role: recommended and had it approved, or decided outright. Either is fine; being vague is not. Say the order plainly and name what slipped.
- what the deferred side got insteadThis is the beat weak answers skip. A date, a reduced version, a workaround that needs none of your time — saying no with an alternative is the whole skill being tested.
- the result and the change you made afterClose with an outcome number and one structural fix — an intake channel, a written rule, a standing sequencing meeting. Result and reflection together are about a quarter of your airtime.
your answer
5 story prompts- Pick a collision where both requesters were genuinely reasonable, not one where somebody was clearly wrong.
- Write one number for what each request protected: a date, a defect rate, a blocked team.
- Name your exact role in the decision — recommended, or decided — and do not blur it.
- Record what you offered the side that waited, with the date you gave them.
- Note the change you made afterwards so the same two people stopped colliding on you.
draft and rehearse your own answer in a learn session
go deeper
This targets stakeholder management and ownership under conflicting demands. The interviewer wants to see that you make the collision visible quickly, compare the two requests on stated criteria rather than on who outranks whom, and take responsibility for what the deferred party hears next. A strong answer ends with both a result and a relationship intact.
Our delivery group was seventeen people across three squads, and I sat on the shared components squad, so every client app in the building pulled our module. In one sprint two account leads came to me on the same afternoon. One needed a redesigned onboarding component for a client demo; the other needed an analytics SDK swapped, because their client's contract required the new one before a release that was eight days out. I did not argue about which mattered more in the abstract. I wrote down what each one protected. The SDK swap protected a contractual date; the onboarding work protected a demo that could be rescheduled. Then I estimated both: the swap was about two and a half days including a regression pass across four apps, onboarding closer to six because it touched every screen size we support. I took that to our delivery manager as a recommendation rather than a question — SDK first, onboarding starting the following Tuesday — and I went to the onboarding lead myself with a date and an offer to ship a stripped version for the demo using the existing styling. She took the stripped version. The swap landed with two days of margin and that release reached 79% seven-day adoption. Afterwards I asked for one intake channel into our squad, because both leads had come to me directly, and that was the real defect.
The comparison carries the signal: what each request protects, an effort figure beside each, then a recommendation handed upward rather than the decision itself. The fallback for the deferred lead is what keeps it from being a refusal. Skipping the date and stripped version would drop this a rung.
By then I owned the release train across three client apps, so a collision like this stopped with me instead of going up. We were nine weeks into an eleven-week engagement. The client's marketing lead wanted a store-refresh feature in the build we were cutting on Thursday. Our own platform lead wanted the same week to finish a background-sync rewrite that was already nine days late, because the client's backend team had only just shipped the endpoint we were blocked on. I made the call myself and said its cost out loud, which is the part I used to avoid. I told the marketing lead the refresh would miss this train and go on the next one, twelve days later, and I told him exactly what he was buying: sync was why 22% of returning sessions were loading stale content, and putting his feature on top of that would launch it onto a bad experience. Then I gave him something real. The copy and asset changes his campaign actually needed went out as server-driven config, no build required, so his date held without touching the train. Sync shipped on the following train and pulled us to 83% seven-day adoption, the best on that account. The durable change was a standing rule: anything blocked on a team outside our building gets declared in the first week of an engagement, not in cut week.
Deciding rather than escalating, naming the cost to the losing party's face, and engineering a path that protects their date without consuming the contested capacity is what reads as senior. The standing rule at the end shows the collision was treated as systemic. Softening the cost sentence would flatten it.
At this level nobody expects you to arbitrate. Show that you surfaced the collision the same day, brought a suggested order rather than an open question, and then executed the call without sulking about it.
You are expected to run the comparison yourself: what each request protects, a rough effort estimate for each, and a recommendation handed upward instead of a decision handed upward. Then a concrete fallback for the side that waits.
Make the call yourself and say its cost out loud to the person who loses it. What separates this band is the alternative you engineer for them and the follow-up change to how work arrives, so the same collision is cheaper next time.
Treat a recurring collision as a system defect. Talk about who owns the arbitration, what criteria are written down, and how you removed the structural reason two parties both believed they held the top slot.
saying these in an interview costs you the question
- Doing both by working nights, presented as the resolution
- Letting the more senior requester win on rank alone
- Casting the other stakeholder as unreasonable rather than differently informed
- No date or alternative offered to whoever was deferred
- The story ending at the decision with no outcome named
- Sitting on the collision for days before telling either side
- What would you have done if neither of them accepted your order?Show a real escalation path rather than defiance or collapse. Strong handling names the person who owns the trade-off, brings both requesters into one conversation instead of shuttling messages, and puts the cost of not deciding on the table. Say what you would commit to once the decision goes against your recommendation.
- How long did you sit with it before you escalated?Speed is the signal. Aim to show hours, not days, and explain the threshold you use: once you can see that the two cannot both land, it stops being your problem to absorb quietly. If you were slow that time, name the delay and what it cost rather than smoothing over it.
- What did the person who lost that call say to you afterwards?Interviewers are checking the relationship survived. Give a real reaction, including an unhappy one, and what you did with it — a follow-up date you then hit, a fallback you delivered, or a later request from the same person, which is the strongest possible evidence that trust held.