Structured interviewing
How to build an interview scorecard people actually fill in
The Itya team · Updated · 7 min read
The short answer
A scorecard people fill in is short: four to six competencies, a 1–5 scale with written anchors, a box for the evidence behind each rating, a "not assessed" option and a separate overall recommendation. Completion is a process problem, not a form problem. Book time to score, make scorecards due within 24 hours and before the debrief, and start from a draft.
Why scorecards go unfilled
Ask a recruiter what slows a hiring decision and you will usually hear the same answer: waiting for feedback. Scorecards are rarely skipped out of laziness. They are skipped because the form and the process make them hard to finish.
- The form is long. Twelve competencies, each with a free-text box, is a 30-minute job. It gets deferred, then written from memory, then written badly.
- The numbers mean nothing. If nobody has defined a 3, each interviewer invents the scale every time. Many give up and write "solid, lean hire".
- There is no deadline. "Before the debrief" is not a time. A task with no due time loses to every task that has one.
- No time was booked. The interview was in the calendar; the scoring was not. The next meeting starts at the top of the hour.
- Nobody reads it. If the debrief runs on what people say in the room, interviewers learn that the form is paperwork.
The cost is not just admin. Anchored ratings and independent scoring are what make an interview structured, and in the 2022 re-analysis by Sackett and colleagues structured interviews predicted job performance at .42, against .19 for unstructured ones (SIOP summary). A scorecard that is never filled in quietly turns one into the other.
Anatomy of a scorecard that works
Google's re:Work guide names "scoring with standardized rubrics so that all reviewers have a shared understanding" as one of the four parts of structured interviewing. The scorecard is where that rubric meets a busy interviewer. Seven parts are enough:
| Part | What it holds | Why it matters |
|---|---|---|
| Header | Candidate, role, interview stage, interviewer, date | Anyone reading it later knows which interview it describes |
| Competencies | The two or three competencies this interviewer was assigned, each with a one-line definition | Keeps each interviewer on their part of the loop |
| Rating | A 1–5 score per competency, with anchors written for 1, 3 and 5 | A 4 from one interviewer means a 4 from another |
| Evidence | What the candidate said or did that earned the rating, quoted where possible | Turns a number into something the panel can check |
| Not assessed | A tick box per competency | Stops interviewers guessing a 3 for something they never asked about |
| Recommendation | Strong no, no, yes or strong yes, with one sentence of reasoning | Keeps the overall call apart from the ratings |
| For the next interviewer | What to probe that this interview could not | Hands gaps forward instead of losing them |
Keep the recommendation separate
Ratings describe evidence; the recommendation is a judgment about the role. Mixing them invites what the US Office of Personnel Management's structured interview guide calls the halo effect: "allowing ratings of performance in one competency to influence ratings for other competencies." Score each competency first, then make the overall call. Offer four options with no middle, so nobody can hide in "maybe".
Make "not assessed" easy
A good loop gives each interviewer part of the rubric. If the form demands a number for every row, interviewers fill the gaps with a polite 3, and the average starts to look like evidence that does not exist. The same guide lists this as central tendency: "a tendency to rate all competencies at the middle of the rating scale".
An interview scorecard template you can copy
Paste this into your ATS, a form or a document. The example competencies are for a product manager; swap in your own, or start from the product manager kit or another role kit.
Header
- Candidate name and role
- Interview stage (for example, "product sense, round 2")
- Interviewer
- Interview date and time
- Scorecard submitted at (date and time)
Ratings and evidence
| Competency | 1: weak | 3: solid | 5: strong | Score | Evidence |
|---|---|---|---|---|---|
| Problem framing | Jumps to a solution; cannot say who has the problem | Defines the user and the problem; names a constraint or two | Separates symptom from cause, sizes the problem, and says what they would not do | ||
| Prioritization | Ranks by opinion, or by who asked loudest | Uses a stated criterion such as impact or effort | Makes the trade-off explicit and says what evidence would change it | ||
| Working with engineering | Describes handing over requirements | Describes regular collaboration and settling a disagreement | Shows a plan they changed because of an engineer's input, and why | ||
| Using data | Cites numbers without saying what they mean | Picks a sensible metric and explains it | Questions the metric and names what it would miss |
Overall recommendation
| Recommendation | Use it when |
|---|---|
| Strong yes | You would argue for this hire, with evidence on most of your competencies |
| Yes | The evidence clears the bar, and any concern you have is minor and named |
| No | The evidence falls short on a competency the role cannot do without |
| Strong no | You would argue against this hire, with evidence |
- Reason, in one sentence: the main reason for your recommendation, in terms of evidence.
- For the next interviewer: what to probe that you could not.
How to get every scorecard submitted
A better form helps. A better process does more. These seven habits move completion from "eventually" to "the same day".
- Book the scoring slot. Extend every interview invite by 10 to 15 minutes and label the tail "Scoring". OPM's guide advises reviewing your notes and rating the candidate immediately after they leave.
- Set a hard due time of 24 hours. Long enough to survive a busy day; short enough that the answers are still remembered, not just the impression.
- Hide other scorecards until you submit. Seeing a colleague's "strong no" first is a quick way to lose an independent view. It also gives everyone a reason to submit.
- Make the debrief wait for it. No debrief until every scorecard is in. A late scorecard then delays the decision visibly instead of silently.
- Remind, then escalate. A nudge when the scoring slot starts, another the next morning, and the hiring manager copied at 24 hours. Automate it; recruiters should not chase by hand.
- Start from a draft, not a blank form. If the interview was recorded with consent and transcribed, a draft can place the relevant quotes under each competency. The interviewer checks, edits and scores.
- Show that scorecards are read. Quote them in the debrief. People fill in forms that visibly change decisions.
Five mistakes that make scorecards useless
- Too many competencies. Ten rows per interview means each gets a glance. OPM's guide puts a typical structured interview at four to six competencies; split them across the loop so each interviewer owns two or three.
- Unanchored numbers. "Rate 1 to 5" with no descriptions produces each interviewer's private scale. Write one sentence for a 1, a 3 and a 5.
- One "culture fit" box. An undefined fit rating is a vibe with a number on it. OPM lists "similar to me" ratings among the common rating errors, and an open fit box invites them. If you mean something specific, such as "gives direct feedback to peers", make it a competency with anchors.
- An average as the verdict. A 3.6 hides the one competency the role cannot do without. Read the profile, not the mean.
- Adjectives as evidence. "Sharp, great energy" tells the panel how you felt. "Explained why she cut the launch to one market first" tells them what happened.
Where software helps
Most of what makes scorecards late is mechanical: booking the time, chasing people, and recalling what was said. Software can carry that load without touching the judgment.
Itya keeps one rubric per job and joins Google Meet, Microsoft Teams and Zoom as a notetaker, with consent captured before recording. It transcribes with speaker labels, drafts each scorecard with the moment behind every rating cited, and sends scorecard reminders. Results sync to Greenhouse, Lever, Ashby or Workable. The interviewer edits and submits the ratings.
Questions people ask
- What should an interview scorecard include?
- The candidate and interview details, the competencies this interviewer was assigned with one-line definitions, a 1–5 rating per competency with written anchors, an evidence field, a "not assessed" option, a separate overall recommendation and a note for the next interviewer.
- Is a 1–4 or a 1–5 scale better for an interview scorecard?
- Either works if every point is anchored in words. A 1–5 scale gives a clear middle for "meets the bar"; a 1–4 scale forces a lean. The US Office of Personnel Management recommends at least three levels and suggests five to seven. Whichever you choose, keep the overall recommendation as a separate call with no middle option.
- How soon should interviewers submit a scorecard?
- Within 24 hours of the interview, and before the debrief. The best time is straight after the candidate leaves, in a slot booked for it. The US Office of Personnel Management advises rating immediately after the interview.
- Should interviewers see each other's scorecards?
- Only after submitting their own. Reading a colleague's rating first anchors yours, and the panel loses an independent view. Once every scorecard is in, share them all before the debrief.
- Can AI fill in an interview scorecard?
- It can draft one: pull the relevant quotes from a transcript and suggest where each answer sits against the anchors. The interviewer should check the quotes against the conversation, set the ratings and submit. The hiring decision belongs to a person.
Sources
Every source was opened and checked on 10 October 2026.
- Structured interviews: a practical guide (2008), US Office of Personnel Management
- Revisiting meta-analytic estimates of validity in personnel selection (Sackett, Zhang, Berry and Lievens, 2022), Journal of Applied Psychology
- Is cognitive ability the best predictor of job performance?, Society for Industrial and Organizational Psychology
- A guide to structured interviewing for better hiring practices, Google re:Work
Run every interview on one rubric.
Itya turns a job into competencies, questions and a scored rubric, then holds every interviewer to it. Free plan, unlimited teammates.