Interview Feedback
Argue from evidence, not from 'they felt fine'
Write candidate feedback from my interview notes. Key requirements for the role: [paste JD requirements]. Interview notes: [paste Q&A]. Requirements: 1) rate each requirement as exceeds / meets / below, and every rating must quote the candidate's exact words or a specific behavior as evidence; 2) add a separate paragraph on what I'm unsure about and what needs to be tested further; 3) give a clear call: advance, another round, or reject; 4) no unevidenced adjectives like 'strong communicator' or 'high potential'.
How to use
Fill it in the same day — a day later only impressions remain, and there's nothing left to quote.
Good for
- Structured feedback is due the same day, the company wants evidence, and your notes are four keywords.
- Interviewers disagree and you need to turn 'felt fine' into item-by-item judgments against the JD before the hiring meeting.
Not for
- There are no quotes in your notes, only 'fine' or 'so-so' — run another assessment or decline to evaluate; do not have the model manufacture evidence.
- Your impression may be biased, or the conversation touched discriminatory ground — have another interviewer write independently first, then merge.
Example
Input
JD: 1) has owned a 0-to-1 product; 2) can write SQL to check data; 3) has driven schedules with engineering. Notes: on 0-to-1 he said he followed the last one from the spec but was moved off before launch; on window functions, that he uses the ready-made reports and complex stuff goes to the data team; on schedule conflicts, that their lead usually negotiates with engineering.
What you get
Rate each requirement and quote the evidence: 0-to-1 not met (never saw a launch), SQL not met (depends on the data team), collaboration uncertain (the lead drove schedules). Separate what you are unsure about (he may just be a cautious speaker) from follow-ups worth testing (a small data task, one schedule conflict he owned). End with a clear call — another round or reject — and no adjectives like strong communicator without evidence.
Common mistakes
- Feeding in 'felt fine' and asking for a positive write-up — the model supplies the adjectives and the review becomes an impression score with no evidence.
- Writing from memory days later and asking the AI to add detail — the added detail may never have happened, which is factual contamination.
Try tweaking it like this
- Add 'Tag each judgment with evidence strength: direct quote / observed behavior / impression only', so the hiring committee knows how much to trust it.
- Add 'Give the strongest case for advancing and for rejecting', useful when you have not decided and want to avoid anchoring.