The failure question has one disqualifying answer, and most candidates give a version of it: the failure that wasn't — the team let me down, the timeline was impossible, we technically missed the target but learned so much. Interviewers ask this question to watch you hold ownership of something that genuinely went wrong, and every deflection teaches them how you'll behave when something goes wrong on their payroll.
A real failure answer has four load-bearing parts: a failure that was actually yours (a decision, an omission, a misjudgement — named plainly); the harm, sized honestly, including who felt it; what you did about the harm at the time, which is different from what you learned; and the system change — the thing that now exists so this failure class can't simply recur. Learning without changed behaviour is a book report.
This page is the by-situation answer bank: marked specimens across early-career, professional, and commercial contexts. It pairs with our weaknesses guide — the questions overlap but differ in tense: weaknesses probe an ongoing trait under management; failure probes a completed event and your conduct inside it. Interviewers who ask both are checking your self-account stays consistent across the two.
Practise these questions out loud with a live AI interviewer — free. Start with Try Me
How the marking guidance works
Each model answer below is marked against the four criteria a live aurate session scores:
See how a full session is scored
aurate is a practice tool. Marking guidance describes what strong practice answers show — it isn't an employability assessment.
Why it's asked: The base question, and the interviewer's checklist is simple: was it real, was it yours, what did it cost, how did you behave, what exists now because of it. The commonest fail is choosing a safe non-failure; the second commonest is choosing a real one and then narrating it defensively. The specimen below does neither.
I ran a customer event that eleven people attended — in a venue set for eighty. That number still makes me wince, so let me own the decisions that produced it.
The context: my first solo event for a software firm, a breakfast seminar for prospects. The failure wasn't bad luck. It was three choices of mine: I picked the date without checking the industry calendar — we clashed with the sector's biggest annual conference, which a ten-minute search would have caught. I sized the venue on the invite list rather than on any realistic conversion rate, because eighty felt more impressive to book than thirty. And when registrations were visibly stalling at nineteen, a fortnight out, I told my manager it would 'pick up' — not because the data said so, but because admitting the clash felt worse than hoping past it. That third choice is the one I'd actually call the failure. The first two were inexperience; the third was character, and it was the wrong call.
What I did once reality arrived: told my manager the full picture four days out, including the conference clash I'd found too late — her first question was 'when did you know?', and the honest answer ('ten days ago, partially') was the worst moment of the whole thing. We salvaged what could be salvaged: converted the format to a round-table, which eleven people actually suits, and two of those eleven became customers — the event's one accidental mercy.
What exists now because of it: my event checklist opens with an industry-calendar check, venue sizing runs off conversion assumptions I have to write down, and — the real change — I report registration numbers weekly with no adjectives attached. Numbers don't 'pick up'. They are what they are, and my manager hears them while there's still time to act.
The failure cost about £4,000 and a piece of my manager's trust that took two well-run events to earn back. Fair price, honestly, for what it taught.
Marking guide
Failure stories wobble under the second question
The specimens above survive because they hold ownership under probing — 'when did you know?', 'why did you sit on it?' — which is exactly where rehearsed failure answers collapse. A live aurate session asks those second questions, in real time, and marks your composure on the same four criteria used across this page. Two free sessions. No credit card.
Try it freeWhy it's asked: For junior candidates the question tests whether small-scale failures get treated with adult seriousness: disclosure over concealment, repair over panic. Assessors know your failures will be small; what they're pricing is the behaviour pattern before your failures get expensive.
In my first year I missed a client's confirmation-statement filing deadline — a small document, a real consequence, and entirely my error.
How it happened matters, because it wasn't dramatic: the reminder lived in my personal task list rather than the practice's central deadline system, because I'd taken the client over mid-year and never migrated their dates across. Fourteen small deadlines transferred correctly; the fifteenth didn't. The statement went overdue, the client received a compliance warning letter from Companies House, and they learned about my mistake from a regulator before they learned about it from us. That last part is the piece I'd rank as the true failure.
What I did inside the same hour of discovering it: told my manager before I touched anything else — first-year me badly wanted to quietly fix it first, and I'm glad I didn't, because the fix needed her authority anyway. Then I called the client myself, with her on the line: apologised without qualifiers, explained the filing was already being submitted, and confirmed the warning carried no penalty at this stage. The client was gruff and fine. The firm waived that quarter's fee — about £150 — as a goodwill gesture I'd recommended from my own initiative, which my manager approved.
The system change went beyond me: migrating a client now has a written checklist, and 'all statutory dates entered centrally' is its first line — my manager asked me to draft that checklist myself, which I understood as both a kindness and a sentence. Every client I've taken on since has gone through it.
What the episode actually taught me sits underneath the process: bad news travels best at maximum speed and minimum editing. Everything I've seen since has confirmed it.
Marking guide
Why it's asked: The sharpened version, which excludes the safe mechanical-error answers: the interviewer wants a considered decision, made with available information, that turned out wrong — because that's the failure class senior work actually produces. It probes whether you can distinguish a bad decision from a bad outcome, and own the former without hiding behind the latter.
I over-bought a summer range on a signal I wanted to believe — a genuine judgement failure, not a slip, because I'd been given the counter-evidence and reasoned my way past it.
The decision: I recommended tripling our order on a linen collection after its first fortnight sold ahead of plan. The counter-evidence existed and I had it: the early sales spike coincided with a heatwave, our two most seasoned store managers both flagged that the repeat-purchase signal was thin, and the supplier's own lead time meant my reorder would land in late August — buying summer stock for autumn weather. I read all of that and constructed a story around it: 'the range is building a following; the weather just revealed demand that was already there.' It was a clever story. It was also me falling in love with my own early call and hunting for reasons it deserved tripling.
The outcome: the reorder landed August 28th, sold at roughly a third of the spring rate, and we cleared the tail in the October sale. The markdown cost was about £18,000 against my business case. My director's review of it was fair and surgical: 'the buy wasn't wrong because it rained in September. It was wrong because you treated weather as brand loyalty when two experienced people told you otherwise.'
What changed in how I decide: I now write the counter-case before any significant recommendation — one paragraph, steel-manned, in the proposal itself. Twice since, writing that paragraph has changed my own recommendation before anyone else read it. And I weight experienced-operator scepticism as data, not as caution to be argued past — the store managers were right for reasons no spreadsheet showed me.
Bad outcome, yes. But I want to be exact: it was a bad decision, knowably bad at the time, and the distinction is the whole lesson.
Marking guide
Why it's asked: The human-cost version, increasingly asked in people-facing and management roles: it tests whether your failure register includes obligations to people, and whether you can discuss letting someone down without either melodrama or evasion. The strongest answers involve a repair attempted directly with the person affected.
I failed a junior colleague I was meant to be developing — not through anything I did, but through eight months of things I didn't do, which turned out to be the more damaging kind.
The situation: as the team's senior, I had informal development responsibility for two juniors. One was quick, visible, and easy to develop — he asked, so I gave. The other, Sarah, was quieter, competent, and undemanding — and I let undemanding mean unattended. Her mid-year review landed as a shock to her: rated 'developing' on two skills she'd had no idea were considered gaps, because the feedback that should have reached her across eight months had reached only my good intentions. She was visibly upset in that review — not at the verdict, as she told me later, but at discovering the silence had meant something all along.
What I did about it: apologised to her directly, specifically, and without softening — 'you should have heard these things in March, in small pieces, and the fact you heard them in September as a verdict is my failure, not yours.' Then I made the repair structural rather than sentimental: fortnightly twenty-minute sessions with a standing format — one thing going well, one thing to build, evidenced both ways. Her next review: both gaps closed, and she's since told the story herself in a team retro, generously, as an example of feedback culture improving.
The system change beyond us two: I proposed the team adopt 'no review surprises' as an explicit norm — anything appearing in a review must have been said before, in writing or witnessed. It's now in our team charter, and it has teeth: a surprise in a review is treated as the reviewer's failure.
What I actually learned: my development attention was flowing to whoever pulled it. Quiet people don't pull. The job was to push — evenly, on a schedule, whether asked or not.
Marking guide
Why it's asked: The 'biggest' forces a stakes decision: too small reads as evasive, catastrophic reads as alarming, and the interviewer is watching the calibration itself. The working range is a failure with genuine cost that you demonstrably survived and metabolised — big enough to matter, resolved enough to discuss steadily.
Why it's asked: An integrity probe disguised as a failure question: the unnoticed failure tests what you do when disclosure is optional. Interviewers are listening for whether you told anyone, fixed anything, or quietly banked the escape — and the honest answer including the temptation to stay silent scores better than implausible instant virtue.
Why it's asked: A thinking question that separates candidates who've genuinely reflected on failure from those with one rehearsed story: good answers distinguish decision quality from outcome, name what makes a risk well-taken (sized, disclosed, reversible where possible), and can classify their own past examples into each column.
Why it's asked: The organisational-reality version: sometimes you flagged it, escalated it, and watched it happen anyway. The test is conduct inside powerlessness — how you escalated, whether you stayed professional after losing the argument, what you did to soften the landing — and honesty about whether you could have advocated better.
Why it's asked: The reflective version, usually late in an interview: it invites you a level above the story into what failure uniquely provides — calibration of your own judgement, the experience of surviving disclosure, knowing your behaviour under shame. Answers that stay concrete (this failure, this lesson) beat philosophy every time.
Why it's asked: The consistency check, sharpened to reference-call realism: it tests whether your failure account is the shared organisational record or your private edit. The confident answer is yes-with-nuance — same events, possibly harder on you in places — and being able to say what their version would emphasise differently.
A real one that was yours: a decision, omission, or misjudgement with genuine cost — sized honestly, repaired visibly, and followed by a system change you can point at. Scale it to your career stage: a graduate's missed deadline handled like an adult beats a senior's cosmetic near-miss every time. The disqualifier isn't the failure; it's the deflection.
No — but keep the ownership at your own address: what YOU decided, missed, or should have done differently, with others' roles stated factually and without blame. Failures involving people you were responsible for (see the failed-someone specimen above) are among the strongest answers available for management-track roles, precisely because they're harder to own.
Tense and shape: weakness probes an ongoing trait under management; failure probes a completed event and your conduct inside it — disclosure, repair, system change. Interviewers asking both are checking consistency: your failure story should plausibly connect to the traits you've claimed, and contradictions between the two answers get noticed.
You control the selection, not the honesty: choose a failure you can discuss fully and steadily, and be truthful within it. What you must not do is present a trivial slip as your 'biggest' — interviewers calibrate. If a serious failure is likely to surface via references, prepare its telling rather than hoping; controlled disclosure always beats discovered concealment.
On the system change, stated plainly — the checklist, the rule, the habit with witnesses — and, where true, the cost named without self-flagellation: 'fair price for what it taught' lands better than either breeziness or lingering guilt. Avoid the redemption crescendo; interviewers trust answers that end factual over answers that end triumphant.
Two free sessions. No credit card.