EasyHire.me
Log in
Try it for free
Log inTry it for free

Interview Scorecard Template: Criteria, Anchors and a Filled Example

Two reviewers watch the same answer. One writes 3, the other writes 5, and both are right by their own reading — because nobody ever wrote down what a 3 means. A scorecard is the list of things you score; the anchors are what make the numbers comparable. Here is the template, the anchored scale that does the actual work, and a filled example you can copy.

Updated

Most teams that decide to “score interviews properly” get as far as a shared spreadsheet with the candidates down the side and four criteria across the top. Then the scores come in, two reviewers disagree by two points on the same recorded answer, and the debrief goes back to being a conversation about impressions.

The missing piece is almost never the spreadsheet. It is the sentence that says what a 3 looks like. This page gives you the scorecard structure, the anchored scale that makes the numbers mean the same thing to different people, a filled example, and the procedure for using it without the scores drifting. Everything here is meant to live in a table you own — there is nothing to download and nothing to request.

Scorecard, rubric, scoring sheet: what each word is doing

The three terms get used interchangeably, which hides the part teams skip. A scorecard is the list of criteria you score and the place the score goes. A rubric is the definition of what each score means for each criterion. A scoring sheet is the filled-in artifact for one candidate, which is what you actually keep.

Teams reliably build the first and skip the second. A scorecard without a rubric still forces everyone to rate the same four things, which is worth something, but it does not make two reviewers' 4s the same 4. The anchors are the cheap part and the part that carries the whole method.

The template: what goes in each column

Six fields are enough. Anything else is reporting you will not do twice.

FieldWhat it holdsExample
CriterionOne skill or behaviour the role genuinely requires, written as something a person doesDefuses an angry customer without escalating to a manager
WeightHow much the criterion counts. Two levels are plenty: critical and standard2× (critical)
Evidence sourceThe question, exercise or work sample that produces evidence for this criterionQ3: a time a customer was angry and what you did next
AnchorsWhat a 1, a 3 and a 5 look like for this criterion, in observable behaviourSee the scale in the next section
ScoreThe number, on one scale used across every criterion4
Evidence noteThe quote or observation the score rests on — the field that makes the score reviewable laterNamed the SLA, offered the refund before being asked, did not blame the shipping team
The scorecard structure. Copy the fields into a spreadsheet, one row per criterion, one sheet per candidate.

Three rules about the criteria themselves:

  • Four to six, not twelve. Every criterion you add costs attention on all the others, and a reviewer scoring twelve rows starts filling the bottom half from the top half.
  • Every criterion maps to at least one question. If nothing in the interview produces evidence for a row, the row gets filled from impression. That is the test that removes “culture fit” and “passion” from most scorecards.
  • Written before the role goes live, with the hiring manager, not after the first batch of candidates has been seen. A scorecard written afterwards describes the people you already liked.

Writing the anchored scale (the part that gets skipped)

An anchor is a description of observable behaviour at a given score. Not a quality adjective — “excellent” and “adequate” are just numbers with letters. The test of a good anchor is whether two people who have never spoken could read it and sort the same three answers the same way.

ScoreWhat it means in generalAnchor for this criterion
5Clear evidence above the bar for the roleDescribes a specific incident, what the customer actually wanted, the trade-off they made, and what they changed afterwards so it recurred less often
4Between the anchors for 3 and 5
3Meets the barDescribes a real incident and a reasonable response, with the outcome stated; no reflection on what caused it
2Between the anchors for 1 and 3
1Evidence againstAnswers in general terms with no incident, or the incident ends in blaming the customer or a colleague
An anchored five-point scale for one criterion: “Defuses an angry customer without escalating”. Write anchors for 1, 3 and 5 only.

Writing three anchors per criterion rather than five is deliberate. Five anchors per row across six criteria is thirty paragraphs of definition, which is the point at which teams abandon the rubric altogether. The even numbers are explicitly “between the two described”, and reviewers use them exactly that way.

Set the passing bar before the interviews start, as a number: for example, every critical criterion at 3 or above, and a weighted total of 18 out of 30. A bar set after the fact gets set wherever the candidate you liked happened to land, which is the failure the whole artifact exists to prevent.

A filled scorecard, end to end

One candidate, a customer support role, four criteria with the critical ones double-weighted. This is what a reviewer hands to the hiring manager.

CriterionWeightScoreEvidence note
Defuses an angry customer without escalating2×4Chargeback case. Named the SLA, offered the refund before being asked, did not blame shipping. No follow-up change mentioned, so not a 5.
Writes a clear reply without a template2×3Written exercise answer was accurate and polite, but buried the actual answer in the third paragraph.
Finds the answer in documentation before asking1×5Walked through how they search, including what they do when the docs are wrong: files the gap, then answers.
Handles a queue without dropping the slow tickets1×3Has a triage habit and described it; no example of a ticket that sat for a week.
Weighted total 22 of 30; every critical criterion at 3 or above; bar was 18 and all criticals at 3. Recommendation: advance.

The evidence notes are the reason to keep the sheet. Six weeks later, when the hire is going well or badly, the notes are what let you check whether the criteria predicted anything. Scores on their own cannot be audited, and a scorecard you cannot audit never gets better.

Using it without the scores drifting

The artifact is half the method. The procedure around it is the other half, and it is four habits.

Score independently first

Every reviewer scores in writing before the group discusses anything. The first opinion spoken in a debrief tends to become the decision, and a score typed afterwards is a vote for whoever spoke first.

Calibrate on two or three

Before scoring the pool, have everyone score the same two or three candidates and compare. Where you disagree by two points or more, the anchor is ambiguous — fix the wording, not the reviewer.

Review in batches

Scoring ten answers in one sitting keeps your mental benchmark steady. Scoring them across ten days means the tenth candidate is judged against a standard that has quietly moved.

Score while watching

Fill the evidence note as the answer plays, not from memory afterwards. Memory keeps the impression and loses the quote, and the quote is the part that holds up.

One more rule worth stating explicitly: nobody changes a score in the debrief to match the group. If a discussion genuinely changes your reading, add a dated second score and a line saying why. The disagreement is information about the criteria; overwriting it is how a structured process quietly becomes a consensus process.

Scoring async video answers

A one-way video interview is the easiest place to use a scorecard well, because the conditions are identical for everyone: the same questions, the same time limits, the same number of retakes. The variable that normally wrecks comparability — the interviewer who asked a sharper follow-up of one candidate than another — is simply absent.

Two adjustments to the template help in that format. Map each criterion to a specific question number, so a reviewer scoring row three knows exactly which answer to watch. And allow a rewatch before scoring rather than relying on a first impression, which is the one advantage recorded answers have over a live call.

On EasyHire.me the question set is configured once with per-question time limits and retake rules, so every candidate answers under the same conditions, and the AI analysis gives a consistent first pass across the whole pool before a human scores it. See one-way video interview software for how the format works, or how to set up video interview software for the setup itself.

Adapting the scorecard by role

The structure stays the same across levels. The criteria and the anchors move.

Junior roles

Anchor on potential indicators rather than track record: how the candidate structures an unfamiliar problem, how clearly they explain something they do know, what they did when they got stuck last. Demanding evidence of outcomes from someone who has not had the chance to produce any just scores them on opportunity.

Senior roles

Anchor on demonstrated outcomes and the decisions behind them. The common mistake is scoring credentials — title, employer — which the anchors should make impossible by requiring a specific result and the candidate's own part in it.

Manager roles

The hardest case is the first-time manager, where evidence of managing does not exist yet, so the anchors have to describe the underlying behaviours: directness with people they like, influence without authority, a realistic picture of what the job costs. For internal promotions specifically, the scoring template and the questions that go with it are in interview questions for internal candidates.

Five things that make a scorecard useless

Adjectives where the anchors should be
“Excellent / good / fair / poor” is the original problem with different labels. If the row does not describe a behaviour, the reviewer supplies their own definition.
More criteria than questions
Any criterion with no evidence source gets scored on impression, and it drags the total with it. Cut the row or add the question.
The bar set after the interviews
Deciding what counts as a pass once you have seen the scores turns the scorecard into documentation of a decision already made.
Narrative-only reviews
“Strong candidate, good communicator” with no scores and no quotes cannot be compared to anything, including the same reviewer's note on the next candidate.
Filled in after the decision
A scorecard completed to justify a choice is worse than none: it reads as evidence while containing none, which is exactly the thing you would want to rely on if a rejection is ever questioned.

Where this fits

A scorecard is one stage of screening, not the whole of it. For the process around it — criteria, resume review, which evaluation to run and in what order — see candidate screening best practices. If the problem you actually have is that screening takes too long rather than that it is inconsistent, how to screen candidates faster is the better starting point. And for the questions that produce the evidence a scorecard scores, there are 40 video interview questions grouped by what each one measures.

Frequently Asked Questions

What should an interview scorecard include?
Six fields cover it: the criterion, written as a behaviour the role requires; its weight, for which two levels are enough (critical and standard); the evidence source, meaning the question or exercise that produces something to score; the anchors describing what a 1, a 3 and a 5 look like for that criterion; the score itself; and an evidence note holding the quote or observation the score rests on. Keep the criteria to four to six, and make sure every one of them maps to at least one question — a criterion with no evidence source gets filled from impression, which is what the scorecard exists to prevent.
What scale should an interview scorecard use?
A five-point scale with written anchors for 1, 3 and 5 is the practical choice. The number of points matters far less than whether the points are defined: a 1–3 scale with clear anchors beats a 1–10 scale without them every time. Write anchors for the odd numbers only and treat 2 and 4 as explicitly "between the two described" — writing five anchors for each of six criteria is thirty paragraphs of definition, and that is the point at which teams abandon the rubric. Use the same scale for every criterion so the totals mean something.
Should interviewers see each other’s scores before scoring?
No. Every reviewer should score independently and in writing before the group discusses anything, because the first opinion spoken in a debrief tends to become the decision. Once the scores are in, compare them: a disagreement of two points or more usually means the anchor is ambiguous rather than that one reviewer is wrong, so the fix is the wording. If a discussion genuinely changes someone’s reading, add a dated second score with a line explaining why instead of overwriting the first — the disagreement is information about the criteria.
Can you use the same interview scorecard for every role?
The structure carries across roles; the criteria and anchors do not. For junior roles, anchor on potential indicators — how someone structures an unfamiliar problem, how clearly they explain what they do know, what they did when they last got stuck — because demanding evidence of outcomes from someone who has not had the chance to produce any scores them on opportunity. For senior roles, anchor on specific outcomes and the decisions behind them, which also stops the scorecard rewarding titles and employer names. For a first-time manager, evidence of managing does not exist yet, so the anchors have to describe the underlying behaviours instead.
Should AI score the interview for you?
AI is useful as a consistent first pass over a whole pool — transcripts, summaries and a preliminary read applied identically to every candidate — and that is genuinely hard to do by hand when the pool is large. It does not remove the need for human scoring against the criteria, and the hiring decision should not rest on an automated score alone. Under GDPR Article 22, candidates in the EU have rights around decisions made solely by automated means, which is a practical reason as well as a fairness one to keep a person scoring the criteria that decide the outcome.

Key Takeaways

The scorecard is the easy half. The anchors — a written description of what a 1, a 3 and a 5 look like — are what make two reviewers' numbers comparable, and they are the half that gets skipped.
Six fields per criterion are enough: criterion, weight, evidence source, anchors, score, evidence note.
Four to six criteria, and every one of them mapped to a question. Any criterion with no evidence source gets scored on impression.
Write anchors for 1, 3 and 5 only, and let 2 and 4 be “between the two described”. Thirty paragraphs of definition is why rubrics get abandoned.
Set the passing bar as a number before the interviews start. A bar set afterwards lands wherever the candidate you liked did.
Score independently and in writing first, calibrate on two or three candidates, review in batches, and fill the evidence note while the answer is playing.
Keep the evidence notes. Scores alone cannot be audited against how the hire actually worked out, and a scorecard you cannot audit never improves.

Give every candidate the same answers to score

Same questions, same time limits, same conditions — so your anchors have something comparable to score. The EasyHire.me free plan includes AI analysis — 5 interviews/month, no credit card.

Last updated: November 2026. → Candidate Screening Best Practices · How to Screen Candidates Faster · Interview Questions for Internal Candidates · Video Interview Questions to Ask Candidates · One-Way Video Interview Software

logo
  • Pricing
  • Integrations
  • Blog
  • Demo
  • About us
  • Job board