The competencies get the attention when a team writes an interview rubric, and the levels under each one get whatever time is left. They come out as a ladder of adjectives, from “basic” at 1 to “outstanding” at 5, and those words are all an interviewer reads when choosing a number. A rubric whose 3 says “good” and whose 4 says “very good” leaves each interviewer to decide privately how much better very good is.

Every candidate then gets a number on the same competencies, and each number carries whatever its interviewer had in mind. The form looks consistent, and the debrief still spends its first stretch working out what one interviewer’s 4 was worth before it reaches the candidate.

I’d put the rubric’s effort into the levels. For each competency, I’d write what a 1, a 3, and a 5 sound like in this role’s interviews, working from answers candidates have given, and I’d write the 3 first, because the 3 is the hiring bar. Levels written that way give every interviewer the same words to rate against, and the debrief gets quicker, since the room is left with only the candidate’s answers to argue about.

Write the levels from answers candidates have given.

Levels written from imagination describe the candidate their writer has in mind, while levels built from real answers describe what candidates say. So before you draft anything, collect some answers and sort them with the hiring manager.

Six to ten answers to the question a competency hangs on is enough, taken from notes or recordings of past interviews for this role or the closest one you’ve hired for, and cut to the few sentences that count. Read them side by side and sort them into two piles, hire and no hire, setting aside any you can’t agree on.

Then write down what the hire pile has that the other pile lacks. That difference is the raw material for the 3, and it tends to be concrete: answers in the hire pile often name who made the call, or what changed afterward, where the others describe the project. The answers you set aside are useful too, because they sit close to the line the 3 has to draw.

For a role you’ve never hired for, borrow answers from the nearest role you have, or ask the hiring manager to say aloud, in a candidate’s words, an answer that would get an offer and one that wouldn’t. Both are weaker than real answers, so treat the levels they produce as a first draft and rewrite them after the first round of interviews.

Which competencies belong on the rubric is a decision that comes before this one. Choosing them is the subject of a separate guide to competency-based interviewing.

Start with the 3, because it is the hiring bar.

Every other level is written against the 3, so write it first and give it the most care. On a five-point scale, the 3 is an answer you’d hire on, from someone joining this role at this level, with nothing in it the hiring manager would want a second interview to check.

Then write the 1 and the 5 against it. The 1 describes an answer that shows the gap: something a candidate said that tells you they can’t yet do this part of the job. The 5 describes an answer that would clear the bar for the level above.

I’d leave the 2 and the 4 unwritten. A 2 is an answer with part of the 3 in it, and a 4 is a 3 with part of the 5, so an interviewer can place either one from the levels around it. Three descriptions also keep the rubric short enough to read in the pause after an answer.

Here is that scale for one competency, discovery, on an account executive rubric. The question behind it is “Take me through the first call on a deal you went on to win.”

Read down the column and each written level adds something an interviewer could point to in a transcript. None of it says “good” or “strong,” so an interviewer torn between a 3 and a 4 is asking one question: did the answer carry any of the 5?

For the form those ratings land on, the post on scorecards built for the debrief works through each field.

Delete the adjectives and see what is left.

Before an interviewer sees the rubric, run one check on the level descriptions, including on a rubric you inherited. Copy them out, delete every word that only says how much, such as “strong” or “limited,” and read each competency’s levels again.

If two levels now say the same thing, the scale only ever had one level for that competency. “Communicates clearly” at 3 and “communicates very clearly” at 5 both become “communicates,” and an interviewer choosing between them was choosing how they felt about the candidate.

Rewrite each collapsed level around something the candidate says or leaves out. These substitutes hold up:

• Who made the decision, and whether the candidate says it was them.
• What changed afterward, and how they found out.
• Whose agreement the work needed before it could go ahead.
• What they would do differently now, without being asked.

A level written that way runs a line longer than “strong,” and every word of it can be checked against what the candidate said.

Give each seniority level its own 3.

A well-written rubric can still lose its meaning when one role is hired at several levels. Use one rubric for every level, and the senior candidate scores a 5 by default while the junior one scores a 2, so the rating ends up reporting seniority instead of the competency.

Keep the competency and the question the same across levels, and write a separate 3 per level. For discovery, a junior account executive’s 3 might be the single-buyer answer in the table above, while a senior account executive’s 3 covers the whole buying group and the order the candidate brought each person in.

The junior 5 will often read close to the senior 3. That overlap is fine as long as each rubric says which level it was written for, because a hiring manager reading a scorecard can then tell which bar the number was measured against.

Place the answers you set aside before the role opens.

The answers you couldn’t agree on in the first step are the best test the new levels will get, because they sit closest to the line the 3 draws. Before any interview is scheduled, go back through them with the hiring manager and place each one against the levels.

Each answer should land on one side of the 3 on the strength of a phrase the 3 names, such as the buyer’s own words or the people who had to sign off. An answer you still split on means the 3 hasn’t drawn its line yet, so find the detail that would settle it and write that detail into the 3.

Watch which way the set-aside answers fall, too. If every one of them lands below the 3, the bar may be asking for more than the role needs, so read it again against the hire pile before any candidate is rated on it.

Where Metaview fits in interview rubrics.

For the first step, I’d put the competency’s question to Reports and have it summarize, from your team’s interview data, how past candidates for the role answered, then sort those summaries with the hiring manager.

1

2

1. 1The section’s headline, which frames the table as details taken from conversations.
2. 2The line on summarizing what each candidate has said about a skill, the use the first step of this method needs.

The page’s skill line is close to what the first step asks for: one competency, read across every candidate who answered it.

EvenUp’s sales hiring ran into the gap this method closes. “There were six items hiring managers were supposed to assess for,” says LC Dyas, senior talent partner at EvenUp, “but they all had their own questions. We’d never calibrated on what a good answer looked like.” Once its rubrics were clear, the team used Reports to check whether interviewers were assessing for key attributes, and found that hiring managers weren’t asking about coachability.

I’d copy the order: agree on the levels first, then check whether interviewers ask about each competency, because that check tells you little while a good answer still means one thing to this interviewer and another to the next.

Once the levels are in use, the check on each rating comes from the interview itself. The notice candidates see says being recorded isn’t required and won’t affect their outcome, and that if they and the interviewer both consent, a “Metaview Notetaker” appears as a participant during the interview. When the Notetaker joins, it holds a recording of the call and a transcript of it, and each section of its structured notes opens onto the stretch of transcript behind it.

1

2

1. 1A one-line summary under the first question, with none of the candidate’s own sentence in it.
2. 2The note on the company’s values, which says the candidate liked them without saying which one, so no level could be checked against it.

Before choosing between a 3 and a 4, I’d follow a line like either of those to the candidate’s sentence behind it.

Metaview captures every spoken word. For a 3 on discovery, that means the buyer’s problem in the buyer’s words is either there in the transcript, or the call never reached it.

From that recording the Notetaker also drafts a scorecard, and the interviewer holds that draft against the written levels before rating it and submitting it.

Read the first round of scorecards against the levels.

The first scorecards submitted on the new rubric show where its wording held and where it gave way. Lay each rating beside its evidence line and the level it claims, and read for these four mismatches:

• A 3 whose evidence names none of what the 3 lists. That level still reads as a degree word to the people using it, so rewrite it from the hire pile.
• Most ratings sitting at 3. The 3 is drawn too wide, so move the detail that separated your two piles into it.
• A 5 whose evidence never reaches what the 5 names. The 5 describes the level above too loosely, so tie it to the strongest answer you sorted.
• Senior and junior scorecards at the same number on one competency. One 3 is serving two levels, so split it.

Whatever turns up, start the next rubric there. Begin that role’s sorting with the answers this one couldn’t place and the 3s that drew the most mismatches, so its levels start where these went wrong.