How to build your own neutral candidate comparison checklist (copy-ready, with the reason for every item)
On this page
- How do you build your own neutral comparison checklist?
- The copy-ready checklist
- Why is each part of the checklist there?
- Where does the evidence for each criterion come from?
- Can you check the comparison for group-level bias?
- What should the decision record say?
- Common mistakes
- Questions people ask
Build a neutral candidate comparison checklist in five parts, and write the first part before anyone is interviewed: (1) job-related criteria, each tied to a duty of the job and marked must-have or nice-to-have; (2) an evidence line for every criterion and candidate, in the candidate's own words; (3) independent scores, submitted before any discussion; (4) bias checks run at the comparison itself; and (5) tie-break rules agreed in advance. Then record the decision against the checklist. The copy-ready version is below, with the reason each item is there.
The candidate comparison matrix template gives the table layouts, and how to compare candidates after interviews covers the decision method and weighted scores. This page is the checklist that keeps either of them neutral.
How do you build your own neutral comparison checklist?
- List the duties. From the job description and the intake meeting, write the four to six things the person must do well in the first year.
- Turn each duty into a criterion with visible evidence. "Owns renewals" becomes "has owned renewals for a book of accounts and can give their own retention figure".
- Mark must-haves and set their minimum. A must-have is a gate with a stated minimum score, not a heavy weight; see how to weight an interview scorecard.
- Assign the same question or exercise per criterion for every candidate, and the same interviewer seat where you can.
- Write the bias checks and the tie-break order into the checklist now, not at the debrief.
- Sign and date it with the hiring manager, then lock it for the search.
Job-relatedness is the legal anchor as well as the practical one. Title VII's disparate-impact provision asks whether a challenged practice is "job related for the position in question and consistent with business necessity" (42 U.S.C. 2000e-2(k)). The Uniform Guidelines, still printed in the eCFR, define selection procedures to include "informal or casual interviews" (29 CFR 1607.16(Q)); as of November 2026, federal agencies have stepped back from those Guidelines, with OPM removing its references after a June 9, 2026 Justice Department opinion found them unlawful (91 FR 48234), while the statute and state laws still apply. This page is not legal advice.
The copy-ready checklist
NEUTRAL CANDIDATE COMPARISON CHECKLIST
Role: [ ] Req: [ ] Locked on: [date] Signed: [hiring manager], [recruiter]
A. BEFORE THE FIRST INTERVIEW
[ ] Criteria listed, each tied to a duty in the job description
[ ] Each criterion says what evidence would show it
[ ] Must-haves marked, each with a minimum score (a gate, not a weight)
[ ] Weights for the rest, if any, written as percentages that sum to 100
[ ] Same question or exercise assigned per criterion for every candidate
[ ] Same interviewer per seat for every candidate, where feasible
[ ] Rating scale with written anchors for every point
[ ] Tie-break order written (section E)
[ ] No criterion about personal characteristics, "culture fit" or
anything not tied to a duty
B. DURING EACH INTERVIEW
[ ] Every assigned question asked, as written
[ ] Notes record what the candidate said or did, not impressions
[ ] Personal information the candidate volunteers is not pursued or noted
C. AFTER EACH INTERVIEW
[ ] Scorecard submitted before reading anyone else's
[ ] Every score has one line of evidence in the candidate's words
[ ] Criteria not covered marked "not assessed", never guessed
D. AT THE COMPARISON
[ ] Candidates labeled A, B, C in the comparison table
[ ] One criterion read across all candidates at a time, not one
candidate down all criteria
[ ] Candidates reviewed in alphabetical order, not interview order
[ ] Must-have gates applied before any total is read
[ ] For every claim in discussion: "which line of evidence shows that?"
[ ] "Would we say the same about the other candidate's equal evidence?"
[ ] Interviewer patterns checked (one interviewer always high or low)
[ ] "None of them clears the bar" kept as a possible outcome
E. TIE-BREAKS (in this order)
[ ] 1. Depth of evidence on the most important must-have
[ ] 2. Open questions settled with the same extra step for each tied
candidate (same work sample, same reference questions)
[ ] 3. Still level: both are hireable; hiring manager decides and
writes the job-related reason
[ ] Never: interview order, who we liked more, who referred them
F. THE RECORD
[ ] Decision, criteria, scores, evidence and tie-break used, written
the same day
[ ] Kept with the hiring record for at least the legal retention period
Why is each part of the checklist there?
| Item | What it prevents | Support |
|---|---|---|
| Criteria fixed before interviews | Criteria bending toward a favorite after meeting them | Weights and criteria locked at intake; see weighted interview scorecard |
| Same interviewers per seat | Differences between interviewers posing as differences between candidates | OPM's structured interview guide (Appendix D): use the same interviewers across all candidates when feasible |
| Independent scores before discussion | The first or most senior voice setting everyone's score | OPM Appendix D: each panel member rates independently, without discussion, before ratings are compared |
| Evidence line per score | Scores nobody can check | OPM: ratings should be defensible and supported by examples of actual answers |
| Alphabetical review order | Contrast effects | OPM Appendix F: the order in which candidates are interviewed can affect their ratings |
| One criterion across all candidates | Halo from one strong answer | OPM Appendix F names the halo effect |
| Gates before totals, totals by rule | Debate overriding the evidence | Kuncel et al. (2013): mechanical combination predicted job performance better than holistic judgment |
| Tie-break order written in advance | Ties settled by gut feel | A rule set before anyone knows who is tied |
The fuller list of comparison-time biases, with what each sounds like in a debrief, is in how to compare candidates after interviews. When finalists were seen by different interviewers, comparing candidates who had different interviewers shows how to spot lenient and severe scorers before you compare.
Where does the evidence for each criterion come from?
From the scorecards, written by each interviewer from their own notes, not from memory at the debrief. Each cell in the comparison should carry a score and the one line that justifies it, as evidence-based interview feedback describes. If an AI tool drafts the notes, check every quote against what was actually said before it goes into the comparison. Interview Signal, for example, removes any quote it cannot find in the transcript and exports each scorecard as a CSV row, so several candidates can be pasted into one comparison sheet; the scores and the decision still belong to the interviewers.
Can you check the comparison for group-level bias?
Not inside an individual decision. Title VII makes it unlawful to "adjust the scores of, use different cutoff scores for, or otherwise alter the results of" employment-related tests on the basis of race, color, religion, sex or national origin (42 U.S.C. 2000e-2(l)), so the checklist never adjusts one candidate's evidence by group. Group-level review belongs to the process: selection rates by stage, reviewed periodically. The four-fifths rule printed at 29 CFR 1607.4(D) is a long-used rule of thumb for that review, with the federal status noted above; if a stage shows a large gap, look at whether its criteria and questions are job-related, with counsel. How selection rates are calculated is covered in selection ratio in hiring.
What should the decision record say?
DECISION RECORD: [role], [date]
Checklist version: locked [date]
Decision: offer to Candidate [ ] / no offer / reopen search
Must-have gates: A clears / B clears / C needs a plan ([criterion], score 2)
Deciding evidence: [criterion]: "[candidate's words]" ([interviewer])
Tie-break used: none / step [1-3], with reason
Open questions settled by: [reference call / work sample], [date]
Recorded by: [name]
Keep it with the rest of the hiring file. EEOC regulations require covered employers to preserve records having to do with hiring for one year from the date the record was made or the personnel action, whichever is later (29 CFR 1602.14, as of November 2026); a pending charge extends that. A debrief agenda that produces this record is in the interview debrief template.
Common mistakes
| Mistake | Effect | Fix |
|---|---|---|
| Checklist written after the interviews | It describes the favorite | Lock section A before the first interview |
| Ticks instead of evidence | Nobody can check a tick | One line in the candidate's words per score |
| "Culture fit" as a criterion | Similarity bias with a label | Name the job behavior, such as how they handle disagreement |
| Tie-breaks invented on the day | The loudest preference wins | Section E, in order |
| Blank cells filled with guesses | Unassessed criteria decide the outcome | "Not assessed", and an open question |
Questions people ask
How can I build my own neutral candidate comparison checklist?
Write it in five parts, and finish the first before anyone is interviewed: job-related criteria tied to the job's duties and marked must-have or nice-to-have; an evidence line per criterion per candidate; independent scores submitted before any discussion; bias checks run at the comparison itself; and tie-break rules agreed in advance. Then record the decision against the checklist and keep the record.
What makes a candidate comparison neutral?
Every candidate is judged on the same job-related criteria, fixed before the first interview, using the same questions or exercises, by interviewers who score independently and cite what the candidate said or did. Nothing about protected characteristics, and nothing that is not tied to the job, appears in the comparison.
How should a tie between two candidates be broken?
With a rule written before the interviews: first compare the depth of evidence on the most important must-have, then settle the open questions with the same extra step for both candidates, such as one work sample or the same reference questions. If they are still level, both are hireable, and the hiring manager decides and writes down the job-related reason.
Can I use the four-fifths rule to check my candidate comparison?
Not for an individual decision. The four-fifths rule in 29 CFR 1607.4(D) is a long-used rule of thumb for comparing selection rates between groups across a whole process, and as of November 2026 federal agencies have stepped back from the Uniform Guidelines that contain it. Title VII also makes it unlawful to adjust scores or cutoffs on the basis of race, color, religion, sex or national origin. Review process-level rates periodically with counsel instead.