Everyone in hiring is talking about competency-based interviewing right now. A few years ago, hardly anyone was.

So what changed? AI.

The idea was always sound: instead of hiring on a hunch, you score each candidate on how well their answers match what the role actually needs. Doing it properly used to mean writing rubrics by hand and taking notes through every interview, which few teams had time for.

Now AI does that part in the background, scoring answers against your competencies while you stay in the conversation. That one change is why almost any team can run it now, and Metaview is the tool most of them use to do it.

How to hire better using competency-based interviewing

Done well, it comes down to three things:

• Define the competencies the role needs, before you start interviewing.
• Ask questions that get candidates to show real evidence, not just talk well.
• Score every candidate against the same rubric, based on that evidence.

The common thread is simple. You reward evidence, not confidence.

Most people get hired because they interview well. But interviewing well and doing the job well aren’t the same thing, and the gap between them is where bad hires come from.

There’s hard data behind this. Our 2026 AI & Hiring Alignment Report, which surveyed 505 recruiting leaders and hiring managers across North America and EMEA, looked at how aligned teams are on what they want, and what that does to results:

The pattern is hard to miss. Teams that agree on what they’re looking for hit their goals far more often, and defining competencies is how you get that agreement before anyone walks into a room.

Step 1: Define the competencies for the role

It starts before anyone interviews.

Once you know the role, write down the competencies someone needs to do it well, ideally in the intake call while the role is still fresh. Metaview captures the intake call and pulls out the role’s must-haves, so what you’re hiring for is written down before the first interview.

Those must-haves become the competencies you interview against, so the intake call is where competency-based interviewing really begins.

The right competencies depend on the job. For an engineering role, you might look at:

• How they break down an unfamiliar problem
• The quality and clarity of the code they write
• How they design systems and weigh tradeoffs
• How they work with people outside engineering

A sales or finance role would look nothing alike. As a rule of thumb, four to six per role is the sweet spot: enough to cover what matters, few enough to hold in your head.

Then comes the part that decides everything, which is how you write each one down. A competency has to be a behavior you can watch for, not a vague trait:

The version on the right tells every interviewer exactly what to listen for. The version on the left tells them nothing.

This is harder than it looks, and it’s most of the work. As Metaview’s co-founder Siadhal Magos puts it:

Once your competencies are set, you can build them into the interview itself, so every round is structured around the ones it should assess.

In Metaview, you pick the interview type, and the round is structured around your competencies, so interviewers know exactly what they’re there to assess. Teams use it to keep every loop consistent without rewriting a guide each time.

It also pays to define three levels for each competency, so everyone knows what a weak, a solid, and an exceptional answer sounds like. Then spread them across the loop, so each interviewer owns two or three and nothing slips through.

Step 2: Ask questions that get real evidence

With the competencies defined, the questions almost write themselves.

For each one, ask for a specific, real example. Then keep following up until you have evidence, not a rehearsed summary.

Two kinds of question do most of the work:

• Behavioral questions ask about the past: “walk me through a time you...”
• Situational questions ask about a hypothetical: “how would you approach...”

But the real signal is in the follow-ups. What did you personally do? and what happened next? separate the people who did the work from the people who were standing nearby. If you’d rather not start from scratch, our interview questions library has sets for common competencies.

The hard part has always been capturing all of this while staying present. That’s the job AI quietly takes off your hands: Notetaker captures the full interview and writes structured notes against each competency as the candidate talks.

You stay in the conversation. The evidence is organized by the time the call ends.

It also helps to give each interviewer one competency to own, rather than the whole candidate. Five people each going deep on one thing beat five people trading vague impressions, every time.

Step 3: Score candidates against the same rubric

Scoring is where it all pays off, or quietly falls apart.

Rate each competency against the interview rubric you wrote, tie every rating to specific evidence, and use the same scale across the whole loop.

And do it now, while the answer is fresh. A scorecard filled in two days later is a memory test, and memory drifts straight back toward the gut feel you were trying to replace.

Once the evidence is captured, most of the scoring is already done. Metaview drafts the scorecard straight from the interview, each rating tied to what the candidate actually said. You review and adjust instead of starting from a blank page. Scoring takes minutes.

The payoff shows up when you compare. Because everyone is scored the same way, the stronger candidate is obvious:

Same competencies, same scale. On the evidence, Candidate B is the stronger hire, and you can show exactly why, which is the part a hunch never could.

Common competency-based interviewing mistakes

Most teams get the concept fast. The mistakes are all in the execution, and they’re easy to avoid once you’ve seen them.

Generating a generic rubric and stopping there. It’s tempting to paste a job description into ChatGPT, take the tidy list it hands back, and move on. But a generic rubric measures generic things. It won’t tell you what predicts success in your roles, and that’s the entire point.

Letting the rubric drift. A few weeks in, people are scoring from a hazy memory of what a 3 used to mean. Regular calibration fixes it, and Metaview makes that drift easy to spot instead of leaving you to guess. Reports show how each competency is being scored across the team, so drift is easy to catch before it skews a hire.

Scoring the whole candidate at once. One strong answer colors everything else, which is the exact interviewer bias the rubric exists to remove. Score each competency on its own evidence, and that pull mostly disappears.

Where competency-based hiring is heading

Step back, and the direction is obvious.

Hiring is moving from something a handful of great interviewers do well to something any team can run consistently. AI is what makes that possible.

It’s already working. Brex rebuilt their values interview on Metaview, turning six different interpretations into one shared rubric, so every engineer is scored the same way across eight offices.

And the results hold up. In our 2025 survey of 380 talent acquisition professionals, 92% said they made better hiring decisions once every interview was captured and scored the same way.

The best interviewers have always worked this way. The difference now is that the rest of the team can too.