Candidate comparison matrix template: four versions for every hiring stage

A candidate comparison matrix puts every candidate for one role side by side against the same criteria, so the decision rests on evidence rather than on whoever was interviewed last. You need a different matrix at each stage: a screening matrix after phone screens, an evidence matrix after final rounds, a shortlist matrix for an agency's client, and a panel matrix that shows where interviewers disagree about one candidate. All four are below, ready to copy into a spreadsheet, with a filled example.

This page is the set of templates. For the decision method (must-have gates, weighting, breaking a tie and recording the choice), see how to compare candidates after interviews, which includes the spreadsheet formulas for a weighted score.

Which comparison matrix do you need?

Version 1: the screening comparison matrix

One row per screened candidate. It is built for speed: a hiring manager should be able to agree the interview list in five minutes.

SCREENING MATRIX: [role]     Req: [id]     Range: [pay range]     Updated: [date]

Candidate | Knockouts | Must-have 1 | Must-have 2 | Must-have 3 | Pay ask vs range | Notice / start | Motivation (their words) | Recruiter view | Next step
----------|-----------|-------------|-------------|-------------|------------------|----------------|--------------------------|----------------|----------
[name]    | Pass      | Yes: [fact] | Partial:    | No          | In range         | 2 wks          | "[quote]"                | [1 line]       | Interview
[name]    | Pass      |             |             |             | Above by [x]     |                |                          |                | Hold
[name]    | Fail: [which] |         |             |             |                  |                |                          |                | Decline

KEY  Yes = clear evidence on the screen   Partial = some evidence, needs probing
     No = no evidence or contrary evidence   blank = not covered yet

Write a fact in each must-have cell, not a tick. "Yes: runs month-end close for 3 entities" can be checked in the interview; a tick cannot. If your screen notes are thin, the interview note-taking tips for recruiters cover what to capture so this matrix fills itself.

Version 2: the final-round evidence matrix

One row per competency, one column per candidate, and every cell holds a score and one line of evidence. The evidence line is the point: it lets the room check a score instead of arguing about it.

EVIDENCE MATRIX: [role]      Scale: 4 strong / 3 meets the bar / 2 below / 1 no evidence
                             blank = not assessed

Competency      | Must-have? | [Candidate A]             | [Candidate B]             | [Candidate C]
----------------|------------|---------------------------|---------------------------|--------------
[C1]            | Yes        | [score]: "[evidence]"     | [score]: "[evidence]"     | [score]: ...
[C2]            | Yes        |                           |                           |
[C3]            | No         |                           |                           |
[C4]            | No         |                           |                           |
Must-have check |            | Clears / Needs a plan / Out | ...                     | ...
Open questions  |            | [what nobody asked]       |                           |
Risks           |            | [job-related risk]        |                           |
Pay / start     |            | [ask], [date]             |                           |

If you also want a weighted total, add it as a row below the must-have check and use the formulas in the weighted interview scorecard. Keep the individual scores visible next to it.

A filled example: three finalists

Invented candidates for an invented role, Customer Success Manager for mid-market accounts. Candidates are labeled A, B and C on purpose: names invite associations that have nothing to do with the job.

The matrix does not make the decision, but it shows where to look: A and B both clear the must-haves with different strengths, and the open questions say exactly what a reference call or a short follow-up should settle. The debrief that works through a table like this is described in the interview debrief template.

Version 3: the client shortlist matrix for agencies

Clients compare candidates whether you give them a matrix or not. Giving them one means they compare on the brief rather than on the resume formatting. Keep it to one screen, and send it with the individual submittals, not instead of them.

SHORTLIST: [role] for [client]          Sent: [date]     Recruiter: [name]

                       | [Candidate 1]          | [Candidate 2]          | [Candidate 3]
-----------------------|------------------------|------------------------|---------------
Headline               | [one line]             |                        |
[Must-have 1]          | [evidence, their words]|                        |
[Must-have 2]          |                        |                        |
[Must-have 3]          |                        |                        |
Pay expectation        | [figure, base/total]   |                        |
Notice / earliest start|                        |                        |
Interview availability |                        |                        |
Other processes        | [stage elsewhere]      |                        |
Recruiter's view       | [strength + risk]      |                        |

Order the columns by your recommendation and say so. The submittals behind it follow the candidate submittal template, and how to present candidates to clients covers the call that goes with it.

Version 4: the panel matrix for one candidate

Before a debrief, put each interviewer's scores for one candidate in a grid. The spread column shows where the discussion should go: a row where everyone agrees needs a minute, a row with a 4 and a 1 needs the evidence read aloud.

PANEL MATRIX: [candidate], [role]

Competency | Interviewer 1 | Interviewer 2 | Interviewer 3 | Spread (max - min) | Discuss?
-----------|---------------|---------------|---------------|--------------------|---------
[C1]       | 3             | 3             | 4             | 1                  | No
[C2]       | 4             | 1             |               | 3                  | Yes
[C3]       | 2             | 2             | 2             | 0                  | No

Rules: scores entered before the debrief; blank = not assessed by
that interviewer, never a guess; do not average a spread of 2 or more,
resolve it from the evidence.

What keeps a comparison matrix fair?

• Same columns for everyone, fixed before the first interview. A column added after meeting a favorite tends to favor them.
• Blank is not zero. A competency nobody asked about stays blank and goes into open questions.
• Compare across rows, not down columns. Read one competency for all candidates at a time; it limits the halo from one strong answer.
• Evidence in every cell. A score without a line of evidence is an opinion.
• Nothing personal. No photos, ages, family details, accents or "culture fit" columns. What not to write in interview notes applies to the matrix too.
• Keep "none of them" possible. A matrix makes someone look best; it does not make them good enough.

Can AI fill in the matrix for you?

It can do the copying, and you do the checking. Give an approved AI tool the completed scorecards (not your impressions) and ask it to lay them out:

Using only the scorecards below, build a comparison table with one row
per competency and one column per candidate. In each cell put the score
and the single strongest piece of evidence, quoted exactly as it
appears in the scorecard. If a competency has no score, leave the cell
blank and add it to an "Open questions" row. Do not add any
information that is not in the scorecards. Do not rank the candidates.

[paste scorecards]

Then check every cell against the scorecard it came from. Anything quoted should exist word for word, and any cell the AI filled where the scorecard was blank must be emptied. Some interview tools can produce the raw material directly: Interview Signal, for example, exports each scorecard as a CSV row, so several candidates can be pasted into one sheet. For building the scorecards themselves with AI, see how to build an interview scorecard with AI.