A debrief lands on hire, but when you read the written feedback every interviewer has vague evidence and nobody can point to something the candidate actually did. How do you handle it?

Short answer

A hire recommendation with no evidence behind it is not a weak yes — it is an absence of a decision. The panel has told you how they felt, not what they saw. The move is neither to overrule them nor to sign it off, but to work out precisely which required signal was never tested and get that signal, usually with one targeted follow-up rather than another full loop.

Separate what was observed from what was inferred

Read the written feedback with one filter: which sentences describe something the candidate did or said, and which describe a conclusion about them.

"Communicated well" is an inference. "Restated the ambiguous requirement in his own words before starting, and asked which of the two readings we cared about" is evidence — and it is evidence a reader can disagree with, which is what makes it useful. "Seemed strong technically" is an impression. "Chose a hash map, then noticed the ordering requirement and switched to a tree map without being prompted" is a fact.

Do this and the pattern usually becomes visible fast. Thin debriefs are rarely uniformly thin. Typically two interviewers have real evidence on one or two competencies and everyone else has written impressions about the same charismatic surface. What looked like five signals is actually one signal repeated five times.

Find the hole, not the verdict

The productive question is not "should we hire" but "what were we supposed to learn in this loop, and what did we actually learn". Lay the required competencies against the evidence:

Signal the role needs	Who was meant to test it	Evidence produced
Designs for failure	System design	"Good discussion" — no failure mode named
Writes production-grade code	Coding	Solution worked; no tests, not discussed
Handles disagreement	Behavioural	One anecdote, no conflict in it
Domain judgement	Hiring manager	Not covered — ran out of time

Now the decision is concrete. You are not weighing a mood; you are looking at two unfilled boxes. That reframing also makes the conversation with the panel much easier, because you are not telling anyone they were wrong.

Why consensus can be an artefact

Be suspicious of unanimity that arrives with no substance behind it, because debriefs manufacture it in two well-documented ways.

If interviewers can read each other's notes before writing their own, the first submitted assessment anchors the rest. If the debrief starts with the most senior person in the room stating a view, juniors adjust toward it. Neither produces five independent judgements; it produces one judgement with four echoes, and it looks exactly like consensus.

The structural defences are cheap and worth naming: written feedback submitted before anyone can read another's, and the most junior interviewer speaking first in the room. If your loop lacks both, a unanimous thin verdict is close to uninformative and you should say so out loud.

Get the missing signal deliberately

For an unfilled box, run one focused session with a clear brief — not "meet the candidate again and see what you think", which reproduces the original problem. Specify the competency, the interviewer, and what would count as evidence either way.

Choose an interviewer who has demonstrated they can produce evidence, and if the missing signal is domain judgement, that is often the hiring manager rather than another engineer. Tell the candidate honestly that you are adding a session on a specific area; competent candidates read that as rigour, and it is more respectful than a rejection they cannot learn from or an offer nobody can defend.

The time cost is real and worth naming rather than dismissing. Two days in a competitive market can lose the candidate. That trade is legitimate to make — but make it explicitly, and if you choose speed then say you are hiring on partial information and accept the accountability for it, rather than pretending the debrief was sufficient.

Handle your own authority carefully

A manager who overrules a hire decision without engaging the evidence teaches the panel that the debrief is theatre. The next loop will produce even thinner feedback, because interviewers who believe their assessment is decorative stop investing in it. The problem compounds.

Overruling a no hire is the more damaging direction, because it signals that the bar bends under pressure and it disproportionately silences the interviewer who did the most careful work. If you genuinely disagree with a well-evidenced no hire, the honest move is to name what evidence would change your mind and go get it — the same procedure as above.

Where the panel is thin and you are not, present your evidence to them and let the group re-decide. Being persuadable in the room is what keeps the process real.

Calibrate against the actual cost

"Thin signal means no hire" is a decent default and a poor rule, because the cost of a false positive is not constant. On a two-person team where a bad hire consumes the senior engineer's entire quarter, the bar for proceeding on incomplete evidence should be very high. On a large team with a real probation process, strong onboarding, and a manager who has actually performance-managed someone before, the cost is recoverable and the bar can be lower.

Say which situation you are in. A candidate who defaults to "no hire on thin evidence" without asking about the team is applying a heuristic; one who asks what a mistake costs here is making a decision.

Fix the loop, not just this candidate

Five thin write-ups in one debrief is a system telling you something. The usual causes are that interviewers were never told which competency they own, that no rubric defines what strong and weak look like for it, that feedback is submitted late from memory rather than during the interview, or that nobody has ever received feedback on the quality of their feedback.

The concrete interventions are cheap: assign each interviewer one or two named competencies so coverage is explicit and gaps are visible before the debrief; give each a rubric with behavioural anchors; require evidence quotes in the written feedback and reject submissions without them; and periodically review write-ups for quality the way you would review code. Proposing that — rather than only resolving this one candidate — is what separates an answer about a decision from an answer about running hiring.

---

© 2026 Preptima. Originally published at preptima.com.