The loudest interviewer in the room hired his clone, and the scorecard was blank when the candidate walked in. I have watched this enough times that I do not treat it as a personality flaw. I treat it as what happens when the only shared document is the résumé.
He liked the candidate because the candidate talked the way he talks. Fast, certain, a little amused by the current team. The other two interviewers had doubts. They mentioned them. He had more words. The offer went out on Friday. Six months later the team had two people who dismissed the same kinds of problems, and nobody who would sit in the boring parts of the job.
A scorecard you agree before the interview is the brake. Not a culture speech. Not a reminder to "stay objective." A page that exists before anyone is charming.
Agree it while nobody is in the waiting room
Do this in a half hour, before the first screen, with the people who will interview. Four lines. What this person must be able to do in the first six months. What a weak answer sounds like, in a sentence you would actually say out loud. What you will not hire for, even if you enjoy the conversation.
Write the "will not" line. It is the one that saves you. "We will not hire someone who cannot explain a decision they got wrong." Or, for a support lead, "We will not hire someone who has not done a messy handoff." The loud interviewer can still like a candidate. He cannot pretend the line was not there.
If you write the scorecard after a favourite emerges, you will describe the favourite. I have seen a "must have" appear that matched one person's last job and nobody else's. That is not a standard. That is a costume.
Same page, filled before you talk to each other
Each interviewer scores before the debrief, not during it. During it, you will bargain. The loud voice sets the anchor. Everyone else edits toward him because disagreement feels like a scene.
A score is not a vibe from 1 to 5 on "overall." It is a mark on each of the four lines, plus one sentence of what you heard. "Said the migration slipped because the other team was political. Could not name what he would do differently." That sentence is harder to steamroll than "I wasn't sure."
Then the debrief has a shape. Where do we disagree on the lines? If the loud interviewer loved the candidate and the lines say no, the lines win. If you are going to override a line, say so in writing and say why. "We are ignoring the handoff line because we will pair them with Asha for a quarter." Sometimes that is a real reason. Often, saying it out loud kills it. Good.
Clones are a hiring outcome, not a mystery
People hire people who feel easy. Easy often means familiar. Familiar, in a small team, means you already have this person. The job was supposed to add the skill you lack, and instead you added a second opinion you already hold.
The scorecard is how you hire the person who is worse at performing the interview and better at the work. They will lose a charisma contest. They should not lose a page you wrote when nobody was performing.
AI can draft the first four lines from the job post so you are not inventing criteria at midnight. It can also lay the interview notes against those lines so you see the gap. It cannot attend the pre-meeting and force the argument. That half hour is the actual control. Skip it and the tool just helps you document the clone faster.
If you only do one thing on the next role
Book thirty minutes with every interviewer before any candidate is scheduled. Leave with four lines and one "will not." Put it where the notes must go. Refuse to debrief without the marks filled in.
You will still have a loud voice in the room. You want that voice. You do not want it to be the process.
