Interview Recipes
Menu

Problem solving & judgment

How a candidate breaks down ambiguous, complex problems: framing, decomposition, use of evidence, and knowing when to decide with incomplete information.

You can't hand a candidate a real month of ambiguity in an interview, but you can make them replay one. Past decisions are the most testable evidence of judgment there is: how someone framed a tangled problem, what they checked when time was short, and whether they can separate a sound decision from a lucky outcome.

These problem solving interview questions work that evidence from several angles. Some dig into method: breaking a problem down, turning analysis into a recommendation someone else had to act on, choosing between two genuinely workable options. Others test intellectual honesty, which is the half of judgment that polished candidates hide: being confidently wrong about a cause, wanting one answer so badly the analysis bends toward it, or a defensible decision that turned out badly anyway.

Listen for reasoning you can follow step by step. The probes on each card push past the summary version of the story, and the anchors tell you what separates real judgment from a tidy narrative.

Interview for it when

They will meet problems with no obvious answer and incomplete information.

Where this competency comes from

Problem solving is not our invention: it shows up, under different names, in every major validated competency framework. Our ten-competency taxonomy keeps what at least three of the four agree on. The full reasoning is in why structured interviews work.

The Great Eight (Bartram, 2005)
Analyzing & Interpreting
UK Civil Service Success Profiles
Making Effective Decisions
US OPM general competencies
Problem Solving, Decision Making, Reasoning
Google’s four attributes
General cognitive ability (the interviewable slice)

In the O*NET content model it maps to the work activities Getting Information · Analyzing Data or Information · Making Decisions and Solving Problems.

O*NET mapping adapted from the O*NET content model (U.S. Department of Labor), used under CC BY 4.0.

The questions

All 11 cards in full: the question, why it works, the follow-up probes, and the anchors to score against. Still choosing competencies? Plan the interview first.

problem solving

Walk me through a problem that was too big or tangled to attack head-on. How did you break it down, and where did you start?

Why this works

Anyone can describe a solved problem; the skill lives in the middle — what they split, what they parked, why they started where they did. Asking for the decomposition forces the messy part a rehearsed story skips straight past.

Follow-up probes

  • What made the obvious approach a bad idea?
  • Why start where you started?
  • What part of your first framing didn't survive?

What good looks like

A named structure — pieces, sequence, what was deliberately set aside — a reason behind the first move, and a moment where the framing changed as evidence came in.

Red flags

The "complex" problem is routine work told slowly, the story jumps from mess to solution with no middle, or the method amounts to working hard until it clicked.

Card page · add to recipe

problem solving

Tell me about a time you analyzed your way to a recommendation someone else had to act on. How did you get from the information to the call?

Why this works

A recommendation forces a stance where a summary can hide. The question shows whether the candidate can compress evidence into a position someone else could act on — and whether they know where the data stopped and their inference began.

Follow-up probes

  • What did the evidence not settle, and how did you handle that gap?
  • What would have changed your recommendation?
  • How did you present the option you didn't recommend?

What good looks like

A clear chain from evidence to stance, uncertainty flagged rather than smoothed over, the rejected alternative given its honest due, and a recommendation firm enough to act on.

Red flags

The analysis is a list of tools rather than reasoning, the recommendation matches whatever the requester already wanted, or every caveat is hedged until no actual call remains.

Card page · add to recipe

problem solving

Tell me about a time you were confident about the cause of a problem and turned out to be wrong. How did you find out, and what did you do next?

Why this works

Confident misdiagnosis is where problem-solving actually fails — not for lack of analysis, but analysis aimed at defending a belief. The "how did you find out" clause forces a concrete account instead of a philosophy of humility, and shows whether the candidate tests assumptions or protects them.

Follow-up probes

  • What evidence first made you suspicious of your own diagnosis?
  • What would have happened if you'd been wrong for another month?
  • What do you do differently now when you feel that certain?

What good looks like

A specific incident with a real cost, told without blame; the signal that broke the assumption named precisely; and a changed habit since — disconfirming checks run earlier, confidence treated as a cue to verify rather than a reason to stop.

Red flags

No example available ("my instincts are usually right"), the error pinned entirely on others or on missing information, or the story stops at the discovery with nothing done differently since.

Card page · add to recipe

problem solving

Walk me through a decision where you had more than one genuinely workable option. How did you choose — and what did the winner cost you?

Why this works

Real decisions are rarely good versus bad; they're trade-offs between defensible options. The cost clause is the test — someone who can't say what the chosen option gave up hasn't compared the options, they've rationalized a preference.

Follow-up probes

  • What were the top two, and where exactly did they differ?
  • What would have had to be true for the runner-up to win?
  • Who disagreed with the choice, and on what grounds?

What good looks like

Options compared on criteria named before the choice, the runner-up's real advantages acknowledged, the trade-off stated plainly, and a decision made without waiting for certainty that was never coming.

Red flags

The alternatives are strawmen built to lose, the criteria appear after the choice to justify it, or the candidate re-litigates every angle without ever quite landing.

Card page · add to recipe

problem solving

Describe a decision you had to make before you could get the information you wanted. How did you decide, and how did it play out?

Why this works

Most real decisions are made under uncertainty; this one separates candidates who can reason about reversibility, cost of delay, and what evidence is worth buying from those who either freeze or gamble.

Follow-up probes

  • What information did you decide not to wait for, and why?
  • How would you have known you were wrong early?
  • What did deciding then cost or save, compared to waiting?

What good looks like

The trade-off named explicitly — delay versus risk — reversible and irreversible calls distinguished, and some check set up in advance to catch being wrong early.

Red flags

The story is really about luck, the candidate can't say why waiting would have been worse, or gathering more information is treated as free.

Card page · add to recipe

problem solving

Tell me about a decision you had to make faster than you were comfortable with. What did you check in the time you had, and what did you let go?

Why this works

Speed forces triage: which checks matter most, which risks are survivable. The comfort clause invites the honest version — the interesting material is what got skipped, which a polished "I decide fast" story never includes.

Follow-up probes

  • What was actually forcing the clock?
  • Which check did you most wish you'd had time for?
  • How did you find out whether the call was right?

What good looks like

A real deadline with a named source, checks ranked rather than skipped at random, someone told which corners were cut, and a follow-up afterward to verify the call once there was time.

Red flags

The pressure is drama with no actual deadline behind it, no account of what went unchecked, or speed worn as identity — every decision in every story is fast.

Card page · add to recipe

problem solvingsenior

Tell me about a decision you made knowing it would be unpopular. How did you make the call, and what happened once it landed?

Why this works

Popularity is the easiest wrong input to a decision. This shows whether the candidate can separate right from liked — and whether they did the second half of the job: absorbing the reaction without quietly walking the decision back.

Follow-up probes

  • Who pushed back hardest, and what was fair about their case?
  • What would have made you reverse it?
  • Did it stick — and what did holding it cost you?

What good looks like

Reasoning that stands on its own rather than on rank, the objectors' interests stated fairly, the decision explained in person instead of hidden behind process, and a straight answer about the cost.

Red flags

Unpopularity treated as proof of being right, dissent written off as resistance to change, or a decision that quietly eroded as soon as the pushback started.

Card page · add to recipe

problem solving

Tell me about a decision of yours that turned out badly. With what you knew at the time, was it still the right call?

Why this works

Separating decision quality from outcome luck is the core of judgment — bad outcomes follow good process and vice versa. Candidates who can re-run the decision with only what was knowable then are the ones who learn from results instead of just flinching from them.

Follow-up probes

  • What did you know then, and what only became clear after?
  • What in your process, not the outcome, would you change?
  • What did cleaning it up involve?

What good looks like

A costly example owned without hedging, a clean separation of what was knowable from hindsight, and a specific process change — a check added, a threshold moved — rather than a vow to choose better.

Red flags

Every bad outcome is blamed on information nobody could have had, the lesson is "trust my gut more" with nothing concrete behind it, or the example is picked so they were barely responsible.

Card page · add to recipe

problem solving

Tell me about a decision where you badly wanted one answer to be right. How did you keep that from steering the analysis?

Why this works

Motivated reasoning is the failure mode analysis doesn't protect against — the analysis just gets aimed. A decision with a personal stake tests whether the candidate has working defenses: disconfirming checks, outside reviewers, criteria fixed before the evidence arrives.

Follow-up probes

  • What did you do that could have proven your preferred answer wrong?
  • Who was positioned to call you on bias, and did they?
  • Where did it land, and did anyone disagree?

What good looks like

The stake named openly, at least one genuine attempt to disprove the preferred answer, criteria or reviewers in place before the conclusion, and a believable account whichever way it landed.

Red flags

Objectivity claimed as a character trait with no mechanism behind it, the "check" was asking someone likely to agree, or the preferred answer won and every doubt gets retrofitted away.

Card page · add to recipe

problem solving

Tell me about a decision where you went looking for input before making it. Whose did you want, and what were you hoping they'd challenge?

Why this works

Consultation done well is evidence-gathering; done badly it's endorsement-shopping or outsourcing the decision. Asking what they hoped would be challenged separates the two — and shows whether the candidate still owned the call once the input arrived.

Follow-up probes

  • What did they say that you didn't want to hear?
  • What input did you overrule, and why?
  • Who owned the final decision, and did everyone know it?

What good looks like

Advisers picked for what they knew or would push back on rather than for agreeableness, input that visibly changed the decision or was overruled with reasons, and ownership kept throughout.

Red flags

Input collected until it agreed, consultation used to spread blame in advance, or the decision quietly became a committee's because deciding alone felt exposed.

Card page · add to recipe

problem solvingjuniorsituational

A request lands on you that no process covers: the nearest rule says no, common sense says yes, and the person needs an answer today. What do you do?

Why this works

The gap between the rules and sense is where judgment first shows in junior roles, and the dilemma has no free move — hiding behind the rule has a cost, improvising has a risk. The answer-today clause keeps "I'd find out the policy" from being the whole answer.

Follow-up probes

  • What would make you comfortable deciding without asking anyone?
  • Who do you tell afterwards, and why?
  • What if the same request comes back next month?

What good looks like

The actual risk sized rather than assumed, a decision owned within sensible limits, the outcome flagged to whoever owns the process, and the gap raised so a rule exists next time.

Red flags

The rule applied purely for self-protection, a yes improvised with no curiosity about why the rule exists, or a workaround that stays secret and never becomes a procedure.

Card page · add to recipe

Often assessed alongside

A good interview covers four to six competencies, chosen for what the job turns on. Roles that need problem solving usually also lean on:

Hiring for a role that turns on problem solving?

Plan the interview and three short steps will work out which competencies your role needs, no account required. Or subscribe and get new vetted questions and interview guides as we publish them.