Interview Feedback Templates That Force Real Reasoning

Why "leave me feedback" produces noise

When an interviewer is given a blank text box, they write a paragraph of impressions. The next interviewer reads that paragraph before submitting their own, and the loop becomes an echo. The debrief ends up discussing the first interviewer's framing, not the candidate.

A template constrains the writing in ways that produce evidence rather than impressions, and a workflow that hides peer feedback until submission keeps each rater independent.

The template

For every rubric dimension:

• Score (1–4).
• One concrete observation that supports this score. Quote, paraphrase, or describe a specific moment.
• One thing that would have changed the score in either direction.

For the overall recommendation:

• Recommendation: strong hire / hire / no hire / strong no hire.
• One thing the next interviewer should probe.
• One thing you'd want to know before extending an offer.

That is it. No paragraph. No story. Evidence and a forward signal.

Why this format works

It forces the interviewer to remember a specific thing rather than narrate. Specific things hold up under cross-examination at debrief. Stories don't.

It captures what the next interviewer should do. Loops produce more signal when each round narrows on the previous round's open questions, instead of repeating the same probes independently.

It separates the rubric score from the overall recommendation. A good interviewer might score a candidate 3/4 on a dimension and still hire them; a different interviewer might score 4/4 on the same dimension and not hire. The split surfaces those judgment calls.

Hide peer feedback until submission

Whatever tooling you use, lock submitted feedback so subsequent interviewers cannot read it before submitting their own. ClarityHire's interview reports do this by default — feedback is locked until either the interviewer submits theirs or the debrief opens. The cost is one engineer occasionally complaining they want context before they interview. The benefit is independent signal.

What to do at the debrief

Read scores per rubric dimension across interviewers, side by side. Disagreements > 1 point on the same dimension are where the discussion goes. Agreements pass without discussion. The "tell me your impression" round at the start of debriefs is dead time — skip it.

Decision in the meeting, not after. If there isn't enough information to decide, name the specific dimension that's ambiguous and schedule one targeted 30-minute follow-up. Do not extend an offer or rejection on ambiguous data.

Templates do not replace calibration

A template tells interviewers what to write. Calibration tells them what good looks like per rubric dimension. Both are required. A template without calibration produces structured nonsense.