Structured Interview Scorecard: Template, Evidence and Ratings

The difference between a form that records impressions and one that records evidence comes down to three columns. Each criterion needs a written description of the behavior that would earn each rating, a space for what the candidate actually said, and a space for what could not be confirmed in this interview and should be checked later.

The OPM guide recommends asking applicants for the same job a consistent set of questions, applying a common rating scale, and documenting the observations behind each rating official guide. It was published in 2008, but it describes design principles rather than a dated product, so it still serves as a reference. It is not a legal requirement for private employers, and your organization's own hiring policy takes precedence.

The order matters: job analysis, then questions, then the rating scale, then individual notes, then review. Questions written without a job analysis get interpreted differently by each panelist, and notes taken without a scale never connect to a rating.

Start with recurring work and less frequent but critical situations in the role. Ask people who do and supervise the work what behavior is needed, then compare that account with the job description. Replace broad labels such as “communication” with observable behavior, such as explaining a requirement change and its impact to affected teams.

1. List recurring tasks and important exceptions in the role.
2. For each task, write one or two sentences describing the behavior that separates good outcomes from poor ones.
3. Keep only what an interview can observe. Certifications belong on the application review, and code quality belongs in a work sample.
4. Select criteria you can assess within the interview time and define each one in a sentence.

If the process includes a recorded or AI-assisted round, check which criteria that method can actually assess. The role requirements should stay consistent, but each stage may gather different evidence. Candidates can use the AI interview screening guide to understand those formats.

Choose past-behavior and situational questions that fit the role. The first asks what a candidate did in an actual experience; the second asks what they would do under stated conditions. These formats provide different evidence, so choose them based on the job requirements and the experience expected of applicants.

The following is a hypothetical example for illustration. For a criterion called 'prioritizing under conflicting deadlines', the past-behavior question could be: 'Tell me about a time two deadlines overlapped and you could not finish both. What did you do first, and what happened to the other one?' The situational version could be: 'In your first week, two teams ask you for different deliverables due the same day. What do you do?'

Plan the follow-ups in advance and write them on the question sheet. If an answer drifts into what the team did, the follow-up is 'Which part did you personally handle?' If the outcome is missing, it is 'What happened after that?' Keeping follow-ups on a list means every candidate gets the same chance to fill gaps, and it stops the interview from turning into an unscripted conversation for some candidates and not others.

A Situation, Task, Action, Result note area next to each question speeds up note-taking. The MIT career center describes STAR as a way to get candidates to describe their own actions and role specifically and truthfully official guide. It is a note structure, not a scoring rule; an answer does not need all four parts to earn the top rating.

This three-level example covers one criterion: adjusting priorities when deadlines conflict. Preterview created it for illustration; it is not a validated assessment model. Adapt the anchors to the job and distinguish fluent delivery from evidence of the relevant behavior.

A level 1 on one criterion is not a rejection. A candidate with little work history may show something different on the situational question, so rate each criterion separately and leave the overall decision for the debrief. Before the first interview, have the panel rate the same written sample answer independently and compare; disagreements at that stage are cheaper to fix than after the candidates have been seen.

One row per criterion. The evidence column holds what the candidate said, as close to verbatim as your notes allow. Interpretation goes only in the rating column. The follow-up column records what this interview could not confirm and where it should be checked, such as a work sample, a second-round interview, or a reference call.

Do not repeat the candidate's name, school, or contact details on the scorecard; link it by application number. How long the notes are kept and who may read them depends on your organization's data-handling policy, so check with the responsible team. This article is not legal advice.

At the top of the form, list three rules for every interviewer: keep the question order, separate observation from interpretation, and do not ask anything that is not on the sheet. Reserve a fixed slot at the end for the candidate's own questions, and note anything relevant to a criterion that comes up there. What candidates tend to ask in that slot is described in questions to ask at the end of an interview.

The following is a hypothetical example for illustration, not a real candidate's answer. The criterion is 'prioritizing under conflicting deadlines'.

Before: 'Seems proactive and responsible. Would probably prioritize well. 3.' Another panelist reading this cannot tell what earned the rating, and there is no way to compare this candidate with the next one.

After: “When two deadlines overlapped, agreed with the manager to move one by two days and sent interim results to that task’s owner. Said the delayed task affected fewer people.” Under this example scale, record a provisional 2. Follow-up: “Final completion and impact of the delay were not established; ask about them.”

The test for a good note is whether a panelist who was not in the room could read it and arrive at the same rating. 'Seems responsible' cannot be reviewed; a record of what was said and what was missing can be discussed in the debrief exactly as written. Candidates who have rehearsed will often produce well-shaped answers, and that is fine; the note still records what was said, and the follow-up column records what was not. The candidate side of rehearsal is described in the mock interview practice guide.

A scoring process may use averages or totals, but the evidence still needs review. When ratings differ, compare the recorded answer with the defined anchors. A difference alone does not tell you which rater is wrong, and aggregation rules should not change from candidate to candidate.

1. Compare the evidence and anchors for criteria where ratings differ.
2. Ask for missing evidence. If it cannot be recovered, follow the organization’s procedure for incomplete assessment records.
3. Hand the follow-up column to whoever runs the next stage, with the specific items to check.
4. If a question failed to elicit anything about its criterion for most candidates, record the change to make to the question sheet.

The authorized decision-makers should use the predefined job requirements, assessment evidence and the organization’s hiring procedure. Do not turn this illustrative scale into a universal passing score. Any formal scoring policy takes precedence over the example.