How to Answer "Tell Me About a Time You Failed"

Most candidates hear this question as a trap, and the worry is fair: you are being asked to describe yourself at your worst. It is not a trap, and this section covers what the interviewer is scoring. They want evidence that you notice failure early, own it without deflecting, and change something concrete afterwards.

Hiring managers ask it because past recovery predicts future recovery. A candidate who can name a failure precisely has usually thought about it. A candidate who cannot remember one has usually not looked. The answer also tests calibration: if your worst professional failure is a missed meeting, the interviewer learns either that you have not been trusted with much, or that you are not being candid.

There is a second signal underneath the first. How you talk about other people matters as much as the story. Blame that lands on a manager, a vendor, or a teammate reads as a warning sign even when the blame is factually fair.

Choose a failure that was genuinely yours, had a visible cost, and is far enough behind you that you can discuss it calmly. Real failures trace back to a decision you made: a scope call, an estimate, a hire, a design you pushed through over objections.

• Own the decision. The failure should come from something you chose, not something that happened to you.
• Show the cost. A slipped launch, a rolled-back deploy, a lost account, a rewrite. Vague damage reads as an invented story.
• Keep it professional. Personal crises are not the material here, and they put the interviewer in an awkward position.
• Skip the disguised strength. I care too much and I work too hard are recognized instantly and cost you the room.

A backend engineer interviewing for a senior role at a payments company used the time she shipped a schema migration with no rollback path. The migration locked a table during peak hours, she called the incident herself, and the postmortem she wrote became the team's migration checklist. That answer works because the mistake is technical, the cost is concrete, and the fix outlived the incident.

If you are short on candidates, look at the projects you would run differently today. Other pages in the practice library cover how to mine your own history for stories.

Use STAR, but weight it deliberately. Situation and Task are two sentences of context. Action is what you did once the failure was clear. Result is the change that stuck. Most people invert this and spend a minute and a half explaining why the failure was not really their fault.

A shape that works: one sentence of context, one sentence naming the mistake plainly, three or four sentences on what you did next, and a closing sentence on what you do differently now. Said out loud that runs about ninety seconds, which is the right length for a behavioral answer.

Name the mistake before anyone asks you to. "I underestimated the migration by a week" opens the story; "there were some challenges with timelines" makes the interviewer dig, and everything after the digging sounds defensive. You can rehearse that opening line against an AI interviewer on the mock interview page until it comes out flat and unhurried.

This answer fails in delivery more often than in content. Written down it reads fine. Spoken cold it wanders, because the topic is uncomfortable and the mouth fills the silence. Rehearsing it aloud three or four times fixes most of that.

Practice against follow-ups you cannot see coming. Questions like "what would you do differently?" and "did anyone disagree with you at the time?" expose the soft parts of a story faster than rehearsing alone does. Setup for mock sessions is covered on the tutorial page.

In a live interview, an assistant that listens to the interviewer and suggests structure while you talk helps when a follow-up goes somewhere you did not prepare for. It cannot supply the failure itself; the story has to be yours. On a proctored assessment, a recorded screen share, or a company-managed laptop, no assistant is in scope at all, so the rehearsed version is the one you will actually give.