Interview Scorecard Template: Free Rubric With Anchors

An interview scorecard template needs five things: job-related criteria, the question you asked, a rating scale with written anchors, an evidence column for what the candidate said or did, and a decision signed by a named person. Copy the free template below into Word, Excel or Google Docs. It works on paper, with no tool.

A person fills in this card, from evidence, and a person decides. We make an AI notetaker, and further down we say what ours does with interviews, including the part we would not rely on. You do not need it to use this template.

Fill in one card per candidate, per interviewer, per round. Select the table, copy it and paste it into Word, Google Docs, Excel or Sheets. Write the criteria and the questions before the first interview, then leave them alone until the role closes. At the top of the card, add the candidate, role, round, interviewer, date and the time you finished scoring.

A 1-5 scale with no definitions lets each interviewer use a private scale. Anchors fix that. The US Office of Personnel Management describes highly structured interviews as scored against benchmarks of proficiency, and links more structure to better validity and reliability. Start with these generic anchors.

Then write job-specific anchors for your two or three most important criteria. Describe a 1, a 3 and a 5 in behavior you could hear in an answer. Here is what that looks like for a customer support lead, an invented example.

The same role, one interviewer, one round. The candidate and the answers are invented.

1. Take the criteria from the job, not from a list of traits. Four to six is plenty for one interview (our rule of thumb, not a study).
2. Ask every candidate the same questions. In the most structured format OPM describes, all applicants get the same lead questions and the same follow-up probes.
3. Write evidence during the interview, in their words where you can. Score right after it, before you talk to anyone on the panel.
4. Keep what the job does not need off the card: age, family, health, origin, religion and the like.
5. Drop an undefined "culture fit" row, because it collects impressions. If a value matters for the job, write it as a behavior you can ask about, such as raising a problem early.
6. Rate alone, then compare. OPM describes panels where each member rates individually and the group then resolves large gaps.

Give each criterion one owner, so two interviewers do not ask the same question and nobody skips the hard one.

Collect every card before the meeting, then start with the biggest gaps, not the average. Someone who scored a 2 and someone who scored a 4 heard different things, and the evidence column shows what.

End the sheet with four lines: the decision, who decided, the reason in terms of the criteria, and the date. A named person signs it. The average is one input to that decision, not the decision.

A recording helps with one column: evidence. You stop quoting from memory and check the words instead. It also adds duties, so it is optional.

A recorder with no bot shows the candidate nothing, so a written notice before the interview is your job. Several US states, California among them, require every party to agree before a call is recorded, and in the EU you need a lawful basis under the GDPR and should say why you record. Rules differ by country and state, so check the rules where everyone on the call is.

Here is how Reline fits. It captures your microphone and the computer's audio locally on Mac, Windows, Linux (beta) or in the browser, and no bot joins. Transcription, AI and storage run in the cloud. The recording is kept, so you can press play on any transcript line before you paste a quote into the Evidence column. Summaries cite nothing, so check them against the transcript. Chat answers can cite a note or a moment.

• Speakers: labels are Me vs Other, plus beta voice numbering you name yourself. Names are not automatic. In a room, everyone arrives through one microphone, so check who spoke before you quote.
• ATS: no connector to any ATS. You paste the card in by hand.
• Retention: no auto-delete schedule, so your retention period is a rule your team follows. Delete permanently in the trash removes the note, its recordings and its transcript at once. Anything left in the trash is purged at 30 days.
• Access: notes are private by default, and a workspace role alone grants nothing.
• Paperwork: a DPA is available. There is no SOC 2 report and no BAA.
• Price: $15 per seat a month. Free is 10 hours of transcription and 1 hour per recording.

It can draft one, but we would not let it score for you. AI can help you find what was said. A person should set each score, and a person must make the decision. A score or a recommendation also draws more legal attention than a plain transcript, which is where these rules come in.

Stated plainly: when you run it on a recorded interview, the Lens reads the transcript and drafts a score from 1 to 5 on six fixed dimensions, one of them Culture Fit, with an evidence line for each. It adds a total out of 30, strengths, concerns and a hiring recommendation from Strong Yes to Strong No. It is told to score only what was discussed and to mark the rest N/A.

That output is a draft from an AI model, and a person must decide. We would not copy its numbers into your card or treat its recommendation as a hiring signal. If you only want help with the Evidence column, choose "Add to my workspace" on the Lens, which makes your own copy, and edit that copy: remove the scores, the Culture Fit row, the total and the recommendation, and list your own criteria. Then press play on the line behind each quote and score it yourself.

The template is yours either way. If you want the exact words in front of you when you fill in the Evidence column, tell the candidate first, then record.