Giving and Taking Feedback

Two halves, usually asked separately.

Receiving. "Tell me about a time you got difficult feedback." The failure modes are the trivial example ("I was told to comment my code more"), the disguised disagreement (the whole answer explains why the feedback was unfair), and the story with no behaviour change at the end. What lands is: what the feedback was — stated in their words, not softened — your honest first reaction, what you did to check whether it was true, and what actually changed. Admitting the first reaction was defensive is a strength; everyone's is.

The strongest version includes feedback that was partly wrong. "They said I was blocking the team by over-engineering. I disagreed with the framing, but when I looked at my last five PRs, three had abstractions nothing else used. So the observation was right even though I'd have described the cause differently." That's judgement, not just compliance.

Giving. "Tell me about difficult feedback you gave." Here they're checking whether you'll actually do it — plenty of senior engineers avoid the conversation forever and let a problem calcify. Good answers are specific about the preparation: what the observable behaviour was, what impact it had, that you gave it privately and promptly rather than saving it for a review cycle, and that you talked about the behaviour rather than the person. Then what happened, including if it went badly.

A useful shape is observation, impact, question: "In the last two design reviews you rewrote the proposal after it was agreed. It meant Sam redid two days of work. What's driving that?" The question is not politeness — it's the part that finds out whether you understood the situation at all.

At a small company there is often no manager layer to absorb this. Whether hard conversations happen at all depends on whether the seniors will have them. Interviewers also use this question to check that you're coachable — a candidate who has never received feedback they acted on is a candidate nobody can correct.

• Pick feedback that stung and that changed something; trivial examples read as evasion.
• Quote the feedback as it was actually given rather than in a softened paraphrase.
• Admitting an initially defensive reaction is credible; claiming you were delighted is not.
• Describe how you checked whether it was true, then what specifically changed.
• Feedback that was partly wrong makes the best story — take the valid part, name the part you disagreed with.
• For giving: private, prompt, about observable behaviour and its impact, ending in a question.
• Say what happened afterwards, including when it didn't work; avoiding the conversation entirely is the answer they're screening against.