behavioral interview prep
Tell Me About a Time You Made a Mistake: How to Answer (With Example)
The mistake question reads similar to the failure question but tests something different. A failure is an outcome that did not land. A mistake is a discrete decision or action that turned out wrong, usually recoverable, usually owned in writing. Interviewers ask about mistakes because they want to see your error pattern at a finer grain: the calls you make under time pressure, the assumptions you bake into estimates, the corners you cut on a Friday afternoon. The best answers describe a mistake that was visible to others, you flagged before being asked about it, and you fixed with a process change that is still in place. The trap is picking a mistake too small to matter (a typo in a customer email) or too large to recover from (an outage with executive escalation). The sweet spot is something with a name, a fix, and a follow-up.
What interviewers are listening for
They want to see how you handle being wrong in public. Strong signal: the candidate noticed the mistake themselves (or accepted the finding immediately when a teammate flagged it), communicated it upward before being asked, fixed the immediate damage, and introduced a check that catches the class of mistake going forward. Weak signal: passive voice ("the data got corrupted"), missing communication step, or no durable change in process. The interviewer is also watching the emotional read: defensiveness drops your score even when the technical recovery is clean.
A worked STAR answer
Situation
I shipped a database migration to production that had been tested against a 50K-row staging dataset but ran against a 90M-row production table. The migration acquired a long-held lock that blocked customer-facing writes for about 18 minutes during business hours.
Task
I needed to recover the system, communicate what happened to affected customers and stakeholders, and prevent the same class of mistake from happening again.
Action
I rolled back inside the first six minutes once I saw the lock-wait metrics climb, paired with our on-call SRE to rewrite the migration in 5,000-row batches, and re-ran it overnight when traffic dropped to 4% of peak. I wrote the postmortem the same evening rather than waiting for the formal process, named the missing step (I had not run EXPLAIN ANALYZE against realistic row counts) and shared it in the engineering channel before the Tuesday review.
Result
Customer impact was 18 minutes of degraded writes affecting roughly 1,200 accounts. Our support lead drafted three response templates from my postmortem and we closed all escalations inside two business days. The durable change: any schema migration touching a table above 1M rows now requires an EXPLAIN ANALYZE block in the PR description, run against a production-shape snapshot. Eleven migrations have shipped since with zero customer-visible incidents.
Variants of this question you might hear
The same competency comes wrapped in different phrasings. Each variant emphasises a slightly different signal. Prep one strong story and you can stretch it across all of these.
Tell me about a mistake you made and what you did about it.
What they're listening for
Equal weight on the mistake and the response. They want both halves: the specific decision that was wrong and the specific actions you took to recover and prevent recurrence.
Walk me through an incident or bug you caused.
What they're listening for
Technical version of the same question. Bonus signal if the candidate distinguishes the technical fix (small, fast) from the trust-recovery work (slower, more important). Defensive answers about why it was not really their fault score badly.
What's something you'd do differently in your last role?
What they're listening for
Reflective version. The interviewer is testing whether the candidate has done the work to look back honestly. Generic answers ("delegate more") flag a candidate who has not actually sat with the question.
Tell me about a time you were wrong about something important.
What they're listening for
Cognitive humility. Strongest signal: the candidate names a belief they held confidently, the evidence that changed their mind, and how the experience shifted how they form opinions now.
How to answer
Pick a mistake with three properties: it was visible to others, it had a recoverable cost (real but not catastrophic), and it produced a process change you can name. Avoid the two ends of the spectrum. A typo in a customer email is too small to count as a mistake worth interviewing about; an executive-escalated outage tied to a single decision can read as a red flag depending on the company. The middle (a bad estimate, a missed check, a shipped bug caught within hours) is the right scope. Write the answer in STAR shape, with at least two named technical or process details so the story reads as real rather than rehearsed. Close with the specific check or rule that exists now because of the mistake, and ideally a count of how many times it has fired since.
Common traps
- The mistake-that-was-not-really-a-mistake. If you describe the situation and the interviewer cannot pinpoint what you actually got wrong, the answer fails the question. Lead with the specific call you made that turned out incorrect.
- Passive voice across the whole answer ("the deploy got messed up", "the test was missed"). Replace with first-person actives. The interviewer is grading ownership and passive language reads as deflection.
- No communication step. If your answer is "I noticed, I fixed it, no one knew" the interviewer cannot evaluate how you handle being wrong with peers and managers in the room. Always include who you told and when.
- A mistake with no follow-up. If the system or your process is the same as before the mistake, the implicit message is the mistake will happen again. Name the check or rule that exists today because of the experience.
Practice this answer out loud
Reading a worked STAR example helps. Saying yours out loud, with realistic follow-ups, helps more. Interview Pilot runs voice-based mock interviews tuned to your role and stage, and if you paste your interviewer's LinkedIn it tailors questions to their background. Every answer gets STAR-graded so you know which beat was thin before the real call.
2 free sessions · No credit card · No subscription