How do you decide what to work on first when everything on your list is marked urgent?
answer
- name the two criteria you rank by
- separate urgent from important
- one real list, ordered out loud
- tell the deferred requester, with a date
- re-check the order as facts change
basics
~20 sTests whether you rank work by a repeatable rule instead of reacting to whoever asked last. Name your criteria, show one concrete call you made with them, and say who you told about the work you pushed back.
how to answer
5 beats- the criteria you rank byOpen with two or three concrete criteria — already live beats not shipped, unblocks someone else beats blocks no one, hard external date beats movable internal one. Keep this to about a fifth of your airtime; it is the setup, not the answer.
- the real list you had to orderName the actual competing items and who wanted each, with enough stakes attached that the comparison is possible. Adjectives like critical do no work here; a defect rate, a blocked colleague or a fixed date does.
- the call you made, and what it costState the order and say plainly what lost. This is the bulk of your airtime. If nothing slipped, you did not prioritize, you just worked hard, and the interviewer will hear the difference.
- what you told the person you deferredGive the message: a date, one line of reasoning, and a smaller alternative where one existed. Saying no with an option attached is the difference between prioritizing and stonewalling.
- how you re-check the orderClose with the maintenance habit — a posted list, a standup confirmation, a re-order when new facts land. One sentence is enough, and it shows the method is routine rather than a story you tell in interviews.
your answer
5 story prompts- Pick one real week where at least three items were labelled urgent by different people.
- Write the two or three criteria you actually rank by, in your own plain words.
- Name the item that slipped and what you sent its owner, including the date you promised.
- Find one number for the stakes: a defect rate, days blocked, or a fixed external date.
- This can reuse the story behind your tight-deadline answer, re-angled onto the ordering decision.
draft and rehearse your own answer in a learn session
go deeper
This probes execution judgment and self-management: whether you have an explicit ranking method or simply respond to social pressure. A strong answer proves you can compare requests on stated criteria, accept a visible cost, and keep the person whose work slipped informed instead of quietly dropping it.
I rank by two things: whether it is already in users' hands, and whether it is blocking another person. On the client app I work on I came in one Monday to three things all flagged urgent — a crash on the checkout screen in the live build, a shared components upgrade the platform squad wanted me to adopt, and a copy change the account lead had promised for a demo. The crash went first because it was already shipping. It was hitting about 4.2% of sessions, and every hour we waited was more people hitting it. The components upgrade went second, because two engineers were stuck behind my module until I took it. The copy change went last: it was one string, and the demo was on Thursday. Then I did the part I used to skip. I messaged the account lead, said the copy change would land Wednesday morning and asked her to correct me if that was wrong. She said it was fine. My lead confirmed the crash call in standup, which took thirty seconds. The patch build went out that evening and came in at 74% seven-day adoption, so the fix reached most users inside the week. What I would change is that I ranked in my head for the first hour before telling anybody. Now I post the order first and let people correct it cheaply.
The signal is a stated rule applied to a real morning, plus the message to the deferred requester with a date on it. The reflection is specific and small, which is right at this level. Ranking silently, or never mentioning the copy change again, would drop this to reactive triage.
I run a two-axis pass — impact on the release we are about to cut against effort — and I do it in front of the people asking rather than alone. On my last engagement I owned the payments flow for a client's app. In one week I had a card-scanning feature the client had already announced internally, an accessibility set our QA lead flagged, and a token-refresh defect our backend counterpart needed reproduced before he could fix his side. The token defect went first even though it was the smallest item: ninety minutes of my time unblocked another person, and cheap-plus-unblocking always jumps my queue. Accessibility went next, because it gated the client's store checklist and card scanning did not. Card scanning slipped. The part that mattered was not the ranking, though. I walked the client's product owner through both axes on a single page and offered a trade: card scanning without the offline path, in this release, or the whole feature in the next one. He took the reduced version, so scope moved rather than just order. We cut the release with both blockers closed and it settled at 68% seven-day adoption. I also started posting the ranked list every Monday, which noticeably cut how often people came to ask me to jump the queue.
The middle-band signal is the trade offered in the room: two options with the cost of each named, so the requester chooses. The queue-jump rule for small unblocking work shows real judgment rather than a rigid framework. Ordering without the scope negotiation or the standing list would read a rung lower.
Rank your own queue and show that you asked your lead to confirm the order rather than guessing in silence. Naming two honest criteria and one real morning you applied them clears this bar.
Show the ordering reaching outside your own queue: an effort estimate next to an impact call, and a conversation with the requester where you offered a smaller version instead of a flat no.
Talk about protecting a team's capacity, not just your own day. What is expected is a visible intake and sequencing others can see, and a willingness to say what is not getting done this cycle.
Speak to the mechanism: criteria tied to business outcomes, a forum where cross-org collisions get settled, and evidence that the ranking held after you stopped personally running it.
saying these in an interview costs you the question
- Claiming you simply work faster rather than naming any ranking rule
- Whoever asks loudest goes first, described as being responsive
- Never mentioning the person whose request was pushed back
- Naming a framework but no decision it actually changed
- Escalating every collision upward without proposing an order first
- An answer entirely in the abstract, with no list you really ordered
- What did you say to the person whose request you pushed back?Give the actual words, briefly. Strong handling names a date, gives one line of reasoning the person can check, and offers a smaller version or a workaround. Show you told them before they noticed, not after they chased you. If you handled it badly at the time, say so and say what you would send now.
- How often does your ranking turn out to be wrong?Do not claim it never is. Say roughly how often you re-order, what usually triggers it (new data on impact, an unblocked dependency, a date that moves), and how quickly you tell people when the order changes. Treating a re-order as normal maintenance rather than a failure is the answer they want.
- Who gets to overrule the order you set?Name the role, not a person, and show you understand the difference between deciding and recommending. A good answer says you bring a recommendation plus the trade-off, then commit fully to the call once it is made, without relitigating it in the next standup.
## The trap in the prompt This prompt is asked in several wordings that all want the same thing: - "How do you prioritize your work?" - "Everything on your plate is P1 — what now?" - "Walk me through how you planned your last week." It looks like an invitation to recite a framework, and that is the trap. The interviewer is not checking whether you can spell Eisenhower; they are checking whether an order exists in your head before someone senior imposes one, and whether the people you deprioritized found out from you. ## The two halves of a strong answer **Half one is the rule**, stated in a sentence or two. Two or three criteria is plenty: - something already in users' hands beats something not yet shipped; - work that unblocks another person beats work only you are waiting on; - a fixed external date beats a movable internal one; - small-and-unblocking jumps the queue regardless of size. **Half two**, which is where most answers collapse, is one real list you ordered with that rule. Without the list, the rule is a slogan. The list is also where your judgment becomes visible: they get to see what you decided was less important, which is far more informative than what you decided mattered. ## Evidence that lands Anything with a number attached to the stakes: - how many users a live defect touches, - how many days another person stays blocked, - how much of a release checklist an item gates. Effort estimates count as evidence too — "ninety minutes to reproduce" is what turns a low-impact item into a legitimate queue-jump. Weak evidence is adjectives: critical, huge, business-critical. If both items in your example are described only with adjectives, you have not shown a comparison, you have shown a preference. ## Urgent versus important, used honestly The distinction is worth naming because it explains the hardest call in the answer: the thing that is **loud but movable** losing to the thing that is **quiet but dated**. If you name the framework, immediately show it changing an outcome. A named framework with no consequence reads as preparation, not practice. ## Weak versus strong - **Weak:** "I make a list, I talk to my manager, and I make sure everything gets done." It promises no trade-off, which means no prioritization happened. - **Strong:** the same list, one item explicitly not done this week, and a message to its owner with a date on it. Interviewers reward **the visible cost**. A candidate who says nothing slipped is either not busy or not telling you what slipped. ## How the bar moves - **Early on**, the scope is your own queue and the winning move is confirming your order with your lead quickly and cheaply — ranking silently for a day is the classic junior miss. - **In the middle band**, the ordering has to survive contact with the requester: you are expected to negotiate scope, not just sequence, offering a reduced version rather than a binary yes or no. - **Further up**, the question stops being about your list at all and becomes about capacity: whose work you protect, what you publicly declare is not happening this cycle, and whether the sequencing is legible to people who never talk to you. - **At the top of the range**, the interviewer wants a mechanism that outlives your attention — criteria written down, a place collisions get resolved, and evidence the ranking still held once you were not in the room.