Hiring · August 2, 2026 · 10 min read

Interview Feedback Examples: How to Write It Well

Interview feedback examples for hiring managers: copy-adaptable phrases for strong-hire, mixed and decline decisions, each anchored to evidence you can defend.

← Part of The five pillars of hiring: what assessments measure

If you are the hiring manager or interviewer typing up a scorecard at the end of a long day, the feedback you write is doing more work than you think. It is the raw material of the hiring decision, the record you will defend if a rejected candidate asks why, and, increasingly, a document that can be requested in a dispute. Most interview feedback fails all three tests at once, because it records conclusions instead of evidence: 'not a culture fit', 'strong communicator', 'seemed junior'. Those phrases feel like judgement and are actually noise. This guide gives you interview feedback examples you can adapt today, each shown as the weak version most people write next to the evidence-anchored rewrite that holds up. The rule underneath them all: write what you observed, not what you inferred.

Good feedback is not about being kinder or harsher. It is about being traceable. Every line should point back to a moment in the interview, so a colleague who was not there, or you in three weeks, or the candidate, can see what you saw and why it mattered for the role. This sits downstream of a good interview scorecard template and a genuinely structured interview: the scorecard tells you which competencies to write about, structure ensures you gathered comparable evidence to write from. Get those upstream and the feedback almost writes itself.

Why does most interview feedback say nothing useful?

Most feedback fails because it records the interviewer's impression rather than the candidate's behaviour. An impression is a compressed verdict, formed fast and defended later, and it cannot be checked by anyone who was not in the room. When four interviewers each write a verdict, the debrief becomes a negotiation between opinions rather than a comparison of evidence. Nobody can tell whether 'seemed junior' means a missing skill or someone who simply looked younger than expected.

The fix is a discipline borrowed from good note-taking: separate observation from inference. An observation is what was said or done; an inference is the meaning you assign to it. Both belong in feedback, but kept apart and in that order, evidence first, judgement second, so the reader can agree with your evidence while questioning your conclusion. That separation is also what makes feedback fair: it exposes the leap from behaviour to verdict, where similarity bias and the halo effect struggle to survive.

What does the weak-versus-useful pattern look like in practice?

The bank below is the artefact: feedback lines to lift and adapt, grouped by the decision you are recording and, within that, by competency. Each pair shows the weak phrasing almost everyone reaches for first, then the evidence-anchored rewrite. The rewrites are longer, and that is the point: specificity costs words. Adapt them to your role and actual observations; never paste a rewrite you did not witness, because a fabricated specific is worse than an honest generality.

Strong-hire feedback: name the evidence, not the halo

The danger with a strong candidate is that one impressive moment colours the whole scorecard and you write praise you cannot substantiate. Anchor the enthusiasm to specific behaviours, so the debrief hears why they were strong, not just that you liked them.

• Communication — Weak: 'Excellent communicator, great culture add.' Useful: 'Explained the caching trade-off to me, a non-specialist, without jargon, and paused to check I had followed before moving on. Adjusted the level of detail when I asked a basic follow-up.'
• Problem-solving — Weak: 'Very sharp, clearly senior.' Useful: 'When the brief was deliberately vague, asked three clarifying questions that narrowed the problem before writing anything, then stated the assumption they were proceeding on out loud.'
• Ownership — Weak: 'Real leader, takes initiative.' Useful: 'Described stepping back from a decision so their team could own it, and named the specific risk that created for them personally. Gave concrete detail on what they did when it went wrong.'
• Domain judgement — Weak: 'Knows their stuff, strong technically.' Useful: 'Chose the simpler of two designs and explained the maintenance cost that drove the choice, rather than reaching for the more impressive option to signal expertise.'
• Collaboration — Weak: 'Team player, everyone will love them.' Useful: 'Credited a named colleague for the part of the project I asked about, then drew a clear line around their own contribution when I probed, without overclaiming.'

Mixed-evidence feedback: describe the split honestly

Most real candidates land here, and mixed feedback is where the weak-versus-useful gap matters most. The temptation is to average the impression into a mushy 'good but not sure'. Resist it: record the strength and the gap as separate, specific observations, so the debrief weighs them against the role rather than a vibe.

• Competence versus communication — Weak: 'Smart but a bit awkward.' Useful: 'Solved the core task cleanly and explained the algorithm well, but when I asked how they would present the result to a customer, the answer stayed technical and did not adjust for the audience.'
• Depth versus breadth — Weak: 'Strong in some areas, weaker in others.' Useful: 'Deep and specific on the data-modelling questions, with real examples; on the stakeholder-management scenario, gave a generic textbook answer with no story from their own experience.'
• Potential versus current level — Weak: 'Not quite there yet but promising.' Useful: 'Reasoning and questions were strong for the role; hands-on experience with the specific stack is thin, evidenced by hesitation on the practical follow-ups. A gap ramping could close, not a judgement gap.'
• Pace versus rigour — Weak: 'Fast but sloppy.' Useful: 'Reached a working answer quickly, then did not check the edge case I had planted; when I pointed at it, spotted and fixed it immediately, so the gap looked like habit rather than ability.'

Decline feedback: the observable gap, never the person

Decline feedback is where careless phrasing does the most damage. 'Not a culture fit' is the phrase to hunt down and destroy: unfalsifiable, correlated with bias, indefensible if challenged. Replace every verdict about the person with the specific behaviour that was missing for the job.

• The banned phrase — Weak: 'Not a culture fit, wouldn't gel with the team.' Useful: 'Asked how they handle disagreement with a manager, and the answer was to defer without pushing back even when they believed the manager was wrong, which the role needs the opposite of.'
• Missing skill — Weak: 'Just not strong enough.' Useful: 'On the two role-critical questions about incident response, could describe the theory but not a specific incident they had handled; the role requires someone who has run one unaided.'
• No clarifying instinct — Weak: 'Seemed lost, not proactive.' Useful: 'Started building against the ambiguous brief without asking a single clarifying question, and did not flag the ambiguity later, so shipped an answer to the wrong problem.'
• Overclaiming — Weak: 'Something felt off, didn't trust it.' Useful: 'Described leading a project that, on three specific follow-ups, they appeared to have contributed to rather than led; the detail did not hold up when probed from a different angle.'
• Wrong level for the role — Weak: 'Too junior.' Useful: 'Strong individual contributor answers, but the role is majority people-management and the candidate had no examples of managing performance or a difficult report, which was the core of two planned questions.'

How do you turn a weak line into a useful one, every time?

You do not need to memorise every example, just one move you can run on any line. When you catch yourself writing a verdict, stop and ask what the candidate did that produced it, then write that instead. 'Poor communicator' becomes 'gave a five-minute answer to a yes-or-no question'. 'Great fit' becomes 'named the exact trade-off we argue about weekly and took a defensible position on it'. The verdict can stay, but only after the evidence, never instead.

• Lead with the behaviour. Start the sentence with what happened, not your conclusion; the grammar forces the evidence to come first.
• Quote where you can. A remembered phrase from the candidate beats a paragraph of your paraphrase, because it is theirs, not yours.
• Tie it to a competency. If a note does not map to something the role needs, it is bias or filler; cut it.
• Name what was missing, not who was lacking. 'Did not ask about the deadline' is fair; 'not detail-oriented' is a label the candidate cannot rebut.
• Write it for three readers: the debrief, a future you, and the candidate. If any would be misled or offended, the line is not finished.

Why does AI-drafted feedback read so plausibly and say nothing?

A new failure mode has arrived, and it is seductive. Interviewers paste a few scrappy notes into a chatbot, ask it to 'write up my feedback', and out comes a fluent, professional-sounding paragraph. The problem is that fluency is the thing that fools us. A language model can only elaborate on what you gave it; if your notes were 'seemed fine, bit quiet', the polished output is a longer, more confident way of saying nothing, now authoritative enough that nobody questions it. The AI cannot add evidence it never had, only plausibility, which is the opposite of what feedback needs.

There is a second, subtler risk. Generated feedback tends toward smooth generality and safe positives, sanding off the awkward, decision-relevant detail a real observation carries. The rehearsed-sounding candidate and the genuinely strong one produce similar AI write-ups, because the model matches a register, not evidence. The evidence-first structure in this guide is the defence: if your raw notes already contain quoted behaviour tied to competencies, a tool can tidy the prose without inventing substance. Feed it verdicts and it launders them into confident verdicts; feed it evidence and it stays honest. The discipline lives in your notes, not the write-up.

Where does the scorecard fit, and how do you build one?

Feedback quality is capped by what you were writing against. Interviewers who score into free-text boxes with no competencies produce the vague verdicts this guide is trying to cure, because nothing on the page prompts them for evidence. A scorecard with named competencies and an evidence column does the opposite: for each thing the role needs, it asks what did they actually say or do? Build the competencies and evidence prompts before you interview, and the feedback has somewhere structured to land. Lay one out following the interview scorecard template, or build it in the browser with the interview scorecard builder so every interviewer captures evidence the same way and your debrief compares like with like.

One deliberate note from someone who reads a lot of these: the best interview feedback I ever received as a candidate was a two-line decline that named the one competency I had missed and quoted my own hedging answer back to me. It stung, it was fair, and I have never forgotten which company sent it. That is the standard: specific enough to be uncomfortable, honest enough to respect the person on the other end.

How does feedback tie back to fairness and to the work itself?

Evidence-anchored feedback is not just cleaner writing; it is your main defence against bias and your audit trail if a decision is challenged. When feedback records behaviour, you can review it across candidates and groups to see whether a question or a rater is producing skewed outcomes, the discipline in reduce bias in hiring and monitored through adverse impact. Vague verdicts hide bias; specific evidence exposes it, because you can inspect the reasons and ask whether they were job-relevant.

There is also an honest limit to what an interview can give you feedback about. An interview samples what a candidate says about their work; a work sample test samples the work itself. The least contestable feedback comes from something the candidate produced in front of you, a concrete artefact to describe rather than a story to interpret. Where you can, run the work first and let the interview probe it, and your feedback shifts from 'they claimed they could' to 'here is what they did'. For the questions themselves, behavioural interview questions generate the most quotable, evidence-rich answers, because they ask what happened, not what someone would hypothetically do.

Pull it together and the standard is undramatic. Write down what you saw before what you concluded, tie every line to a competency the role needs, keep inference and anything touching a protected characteristic out of it, and assume the candidate will read it and a lawyer might. Do that, and your feedback stops being noise and becomes the evidence the decision rests on.

Interview feedback is not where you record your verdict. It is where you record the evidence someone else could use to reach it, or to challenge it. Write for that reader and the verdict takes care of itself.