Hiring · Interviews · Templates
Interview scorecard template: how to build and use one
Last updated By T. Krause
An interview scorecard gives every interviewer the same criteria, a rating scale with a description for each score and room for evidence. Here is how to build one in five steps, how to run the debrief, and a filled-in example for an invented support hire.
An interview scorecard template gives every interviewer the same short list of criteria for the role, a rating scale with a description for each score, and space for the evidence behind each score. Each interviewer fills one in alone, straight after the interview. The team compares them in a debrief before anyone decides.
To create an interview scorecard, pick three to five criteria from what the job actually involves and write down what a 1, 2, 3 and 4 look like for each. Then write one main question per criterion and agree the hiring bar before the first interview. After each interview, score alone from your notes, then meet and talk through the scores that differ.
Below, each of those five steps gets its own section. After them comes a complete example for an invented customer support hire, with three interviewers' scores and the debrief. If you'd rather start from a page with the headings in place, copy our free interview scorecard template.
Why use a scorecard at all
In an unstructured interview, each interviewer asks whatever occurs to them and judges the answers on their own terms. The US Office of Personnel Management, the federal government's HR agency and OPM for short, describes the alternative in its September 2008 guide, Structured Interviews: A Practical Guide. In a structured interview all candidates get the same questions in the same order, all are rated on a common scale, and the interviewers agree beforehand on what an acceptable answer is. The scorecard holds the scale and the agreed answers.
The case for structure comes from research on predicting job performance. In 2022, Paul Sackett and three co-authors re-examined earlier meta-analyses of how well selection methods predict job performance, in a paper in the Journal of Applied Psychology. They concluded that the corrections for range restriction in those meta-analyses often overcorrected, so the validity of many methods had been overestimated. Most methods that ranked high before still ranked high, but with mean validity estimates .10 to .20 lower. Structured interviews came out as the top-ranked selection procedure. OPM's summary of structured interview research adds that structured interviews "increase interviewers' agreement on their overall evaluations by limiting the amount of discretion an interviewer is allowed".
Structure doesn't remove bias, and OPM says so. The same summary reports that some differences in interviewer ratings by race have been found, depending on the race of the candidate and the interviewers, how complex the job is and the type of questions asked. Its guide tells interviewers to watch for rating errors such as giving higher ratings to candidates who seem similar to themselves. What a scorecard does is put the evidence next to each score, so the debrief can question a score the evidence doesn't support.
What goes on an interview scorecard
These are the parts of our template, in the order it uses them:
- Interview details. Candidate, role, interviewer, and the round with its length.
- Scale. The scores you can give and what each one means.
- One section per criterion. The score in the heading, and below it the evidence: what the candidate said or did.
- Concerns. Anything that worries you and doesn't fit under a criterion.
- Recommendation. Hire or no hire.
- Questions for the next round. What you'd want a later interviewer or a reference to cover.
One scorecard covers one interviewer and one interview. Three interviewers on a panel means three scorecards.
How to create an interview scorecard in five steps
1. Pick the criteria from the role
Start from the work. Write down what the person will spend most of their week doing and what would go wrong if they couldn't do it. Each of those becomes a criterion. OPM's guide starts with a job analysis and says a structured interview typically assesses between four and six competencies, unless the job is unique or senior. For a small-team hire with a one-hour interview, three to five is enough, and that is the range our template suggests. With more, you run out of time for follow-up questions.
Write each criterion as something you could watch someone do. "Good communicator" is too loose to score. "Writes replies a customer can act on without a follow-up question" is a criterion. A quick test: if you can't picture an answer that earns the top score and one that earns the bottom score, the criterion is still too vague.
Leave out anything a different step tests better. Typing speed belongs in a work sample, and dates of employment belong in a reference check.
2. Write an anchored rating scale and set the bar
A bare 1 to 4 scale means something different to each interviewer. Anchors fix that. An anchor is a sentence for each score that describes what the evidence looks like. OPM's guide says to use one range for all competencies, with at least three levels and ideally five to seven, and to label at least three of them. It also recommends writing example behaviours for each level of each competency, so interviewers have concrete answers to compare against.
Our template uses four levels: 1 is no evidence, 2 is some evidence with gaps, 3 meets the bar and 4 is clearly above the bar. That's fewer than the five to seven OPM aims for. I think it's the right choice for a small team, because without a middle score every rating has to say whether the candidate meets the bar. If you prefer five levels, change the scale before the first interview and keep it for every candidate.
Then write anchors for each criterion. Here is the one for "handles upset customers" from the example further down.
| Score | What the evidence looks like |
|---|---|
| 1 | No example, or the example ends with the customer still angry and nothing resolved |
| 2 | Stays polite and follows the process, but doesn't find out what the customer needs |
| 3 | Acknowledges the problem, finds out what the customer needs, fixes it or says clearly what happens next |
| 4 | Everything in a 3, and also finds the cause and gets it fixed so other customers don't hit it |
Write the hiring bar down at the same time. For example: we hire if the agreed score is 3 or more on every criterion. If you set the bar after the interviews, a candidate everyone liked can move it.
3. Write the questions for each criterion
OPM's guide describes two kinds of question. A behavioural question asks about something the candidate did in the past. A situational question gives them a realistic scenario from the job and asks what they would do. You can mix both in one interview.
Write one main question per criterion and two or three follow-up probes. Ask every candidate the same main questions in the same order. The probes can vary, because they depend on the answer. For "handles upset customers" that could be:
- Behavioural: "Tell me about the last time a customer was angry with you. What had happened, what did you do, and how did it end?"
- Situational: "A school owner writes that she has been charged twice and wants it fixed today. What do you do in the next hour?"
- Probes: "What did you say first?" "Who else did you tell?" "What would you do differently now?"
The probe "Who else did you tell?" is there because it's the only way to find out whether the candidate went after the cause, which is what separates a 3 from a 4 on the scale above.
4. Score alone, straight after the interview
OPM's guide says each panel member should review their notes and rate the candidate immediately after the interview, without talking to the other panel members. It also asks for ratings supported by actual examples of the candidate's answers.
So write the evidence first and pick the score second. Evidence is what the candidate said or did. "Confident" isn't evidence. "Named the two checks she runs before sending a reply" is. If you can't find evidence for a criterion in your interview notes, the score is 1 on our scale, however good the conversation felt.
Don't open anyone else's scorecard and don't talk about the candidate until every interviewer has written theirs. Once you've heard a colleague's verdict, you can't score as if you hadn't.
5. Run the debrief
Collect all the scorecards in one table before the meeting, with a row per criterion and a column per interviewer. Then:
- Start with the criteria where scores are furthest apart. Where everyone gave the same score, a sentence is enough.
- For each one, every interviewer reads out their evidence before saying their score.
- Change a score only because of evidence, meaning something the candidate said or did that you missed or weighed differently. Rank and confidence in the room don't count.
- Agree one score per criterion and check the agreed scores against the bar from step 2.
- Write down the decision and the reason.
OPM's guide describes the same order for interview panels. Members rate independently, then compare notes, ratings and observations, work through the discrepancies and reach a consensus.
Use the average of each row to choose what to discuss. Don't use it to make the decision. In the example below, two criteria average 3.00. One of them is three interviewers who all gave a 3. The other is a 2, a 3 and a 4, and that needs a conversation.
The person making the hire should speak last. If the debrief opens with their verdict, everyone else's evidence starts to sound like agreement or dissent. Once you've decided, add the hire and its reason to your decision log, so the reasoning is still there when someone asks about it a year later.
A filled-in example: customer support specialist
This example is invented. The company, the candidate and the interviewers don't exist.
Brightwell is a 12-person company that sells booking software to dance schools, and it is hiring its second customer support specialist. Hana Kovač had one 60-minute panel interview with three people: Jonas Weber, the support lead, Elif Arslan, the product manager, and Tom Reid, one of the co-founders. Before the first candidate, they agreed four criteria and the bar.
- Written replies. Writes answers a customer can act on without a follow-up question.
- Troubleshooting. Finds the cause of a problem by asking the right questions before guessing.
- Handling upset customers. Anchored as in the table in step 2.
- Managing the queue. Decides what to answer first when there are more tickets than time.
The bar: hire if the agreed score is 3 or more on every criterion.
Jonas's scorecard, written before the debrief
Interview. Candidate: Hana Kovač. Role: customer support specialist. Interviewer: Jonas Weber. Round: panel, 60 minutes.
Scale. 1 no evidence, 2 some evidence with gaps, 3 meets the bar, 4 clearly above the bar.
Written replies: 4
- Rewrote our sample reply about a failed card payment in five lines, with the one link the customer needed
- Cut the opening apology paragraph and said why: the customer wants the fix first
Troubleshooting: 3
- Asked which browser the teacher used and whether other teachers at the school saw the same error before suggesting anything
- Needed a prompt to check the school's plan limits
Handling upset customers: 3
- Example from her current job: a parent was charged twice for a term. She refunded the duplicate the same day and told the parent when the money would arrive
- I didn't note her telling anyone about the cause, so no 4
Managing the queue: 2
- Said she answers the oldest ticket first
- No example of changing the order when something urgent comes in
Concerns. Hasn't used saved replies in a help desk tool. Queue handling matters most in the first week of term.
Recommendation. No hire, because managing the queue is below the bar.
Questions for the next round. If there is one, show her last September's first-day queue and ask what she answers first.
All three scorecards side by side
Elif and Tom wrote theirs the same way. Averages are rounded to two decimal places.
| Criterion | Jonas | Elif | Tom | Average | Spread |
|---|---|---|---|---|---|
| Written replies | 4 | 3 | 4 | 3.67 | 1 |
| Troubleshooting | 3 | 3 | 3 | 3.00 | 0 |
| Handling upset customers | 3 | 4 | 2 | 3.00 | 2 |
| Managing the queue | 2 | 3 | 3 | 2.67 | 1 |
| Total, out of 16 | 12 | 13 | 12 |
The debrief
Tom is the hiring manager, so he spoke last on each criterion. They took the rows in order of spread, largest first.
Handling upset customers, scored 3, 4 and 2. Tom gave a 2 because Hana offered the refund before asking the parent what they wanted. Elif gave a 4 because her notes had Hana saying she told the billing team about the double charge. Jonas had missed that line while writing up the previous answer. They agreed that a parent charged twice wants the money back, so asking first would have added nothing, and Tom moved up. They also agreed that telling billing isn't the same as finding the cause, so Elif moved down. Agreed score: 3.
Written replies, scored 4, 3 and 4. Elif had given a 3 because the reply didn't name the customer's plan. Jonas pointed out that the plan made no difference to the fix. Elif agreed the rewrite matched the anchor for a 4. Agreed score: 4.
Managing the queue, scored 2, 3 and 3. Elif and Tom both had notes on Hana's answer to the situational question: during an outage she would tag every ticket about it, answer them with one reply, then go back to oldest first. Jonas hadn't written that down. With the answer read out, he agreed it met the bar. Agreed score: 3, which is higher than the 2.67 average. Nobody rounded anything. The score changed because of evidence.
Troubleshooting, scored 3, 3 and 3. No discussion needed. Agreed score: 3.
The agreed scores were 4, 3, 3 and 3, a total of 13 out of 16. Every criterion was at 3 or more, so Hana met the bar and Brightwell made an offer. Jonas changed his recommendation and said why in a comment on his scorecard. His concern about saved replies went into her onboarding plan, and Tom added the hire and the reason to the team's decision log.
Mistakes that make scorecards useless
Criteria nobody can observe. "Culture fit" and "passion" give each interviewer licence to score their own impression. Replace them with something the candidate can show you in an hour.
Scoring days later. By then you're scoring your memory of the conversation, and the debrief has often started in the corridor. Block 15 minutes after each interview in the calendar.
Changing the criteria between candidates. If the third candidate is scored on something the first two weren't asked about, the scores can't be compared. Change the scorecard between hiring rounds, not during one.
Treating the total as the decision. A high total can hide a 1 on the criterion that matters most. Check each agreed score against the bar.
Using the interview scorecard template in Prompt Notebook
Prompt Notebook is ours, so weigh this section with that in mind. The interview scorecard template is one page per interview. It opens with a purple callout telling you to fill it in alone, right after the interview and before the debrief. Below that are headings for the interview details, the 1 to 4 scale, four competencies with a score and evidence for each, concerns, a hire or no hire checklist and questions for the next round. It comes filled in with a sample scorecard for an invented backend engineer, so you can see how much evidence to write.
Before the first interview, rename the competency headings to your criteria and add your anchors under the scale. Each interviewer uses the template to make their own page, and nobody shares theirs until everyone has written one. Then type @ inside the candidate's card in the hiring pipeline template and pick each scorecard, so they're all one click away from the candidate. Each scorecard then lists the card under Linked from.
On the template page, Draft a variant with AI writes a version for a different role from a prompt. Check the competencies it picks against step 1 before you use it.
Open the interview scorecard template and change the headings to the criteria for your next hire.
Frequently asked questions
What should an interview scorecard include?
The candidate, role, interviewer and interview round, the rating scale with a description for each score, and one section per criterion with the score and the evidence behind it. Add a place for concerns, a hire or no hire recommendation and questions for the next round.
How many criteria should an interview scorecard have?
Three to five for a one-hour interview. OPM's structured interview guide says a structured interview typically assesses four to six competencies, unless the job is unique or senior. Fewer criteria leave time for follow-up questions on each one.
Should I use a 1 to 4 or a 1 to 5 rating scale?
Either works if every score has a written description and every interviewer uses the same scale for every candidate. OPM's guide suggests five to seven levels. A 1 to 4 scale has no middle score, so each rating says whether the candidate meets the bar.
Should interviewers see each other's scores before the debrief?
No. Each interviewer scores alone from their own notes straight after the interview, and the scores are compared in the debrief. OPM's guide describes the same order for interview panels.
Do you average the scores to make the hiring decision?
No. Use the averages and the spread to decide what to discuss first. The decision comes from one agreed score per criterion, checked against a bar set before the first interview. In the example above, managing the queue averaged 2.67 and was agreed at 3 after the debrief.