Scorecards
How interviewers rate a candidate. Two systems exist, and an interview uses one or the other.
The built-in categories
With no scorecard attached, interviewers score eight fixed categories, each out of five, each with room for a comment:
Communication · Technical skills · Problem solving · Confidence · Teamwork · Adaptability · Cultural fit · Overall
Scores are per interviewer. Two people on the same interview each record their own set; neither overwrites the other, and the report shows both.
This is a reasonable default for a general interview and a poor one for a specific role. "Technical skills: 4" means something different to every person who writes it, and averages across interviewers hide exactly the disagreement worth discussing.
Custom scorecards
Interviews → ScoreCards is where you build your own. A scorecard is a titled list of items, each of which is one of:
| Item type | Interviewer sees |
|---|---|
| Rating | A numeric scale |
| Single choice | One option from a list you define |
| Multiple choice | Any number of options from a list you define |
Each item has a title, an optional description explaining what to look for, a position in the order, and a required flag. Every item can carry a comment.
Write the description. It is the difference between a scorecard and a form: "Rate system design" gets you noise, while "Did they identify the bottleneck before proposing a solution?" gets you an answer two interviewers would agree on.
Scorecards can be prebuilt (available to everyone) or your own company's. Yours are private to your company.
Attaching one
A scorecard reaches an interview through an interview template, which bundles it with the question set for a role.
Decide before the interview, not after. Interviewers fill in whatever the interview was configured with, and switching afterwards does not retrofit answers to items nobody was asked.
Designing one that works
Fewer items than you want. Eight thoughtful ratings beat twenty rushed ones, and an interviewer completing a scorecard from memory an hour later is guessing regardless of how many boxes there are.
Ask about observable behaviour, not impressions. "Explained their reasoning without prompting" is something two people can agree on. "Seemed confident" is not, and it is the kind of criterion that reliably encodes bias — accent, volume, and the candidate's comfort with your particular ritual all read as confidence.
Single choice beats a rating when the levels are real. If what you mean is "did not attempt / attempted / working / working and tested", say that. A 1-to-5 scale with no defined levels is four people using four different scales.
Mark an item required only when a blank is a problem. A required item an interviewer cannot honestly answer gets a middle value, and a middle value invented to satisfy a form is worse than no data.
Reading the results
The report shows each interviewer's scores side by side rather than averaged.
Resist averaging them yourself. Two interviewers who scored a candidate 2 and 5 have not told you the candidate is a 3.5 — they have told you they saw different things, and finding out which one is right is the most valuable half-hour in the process.
Scores are evidence for a human decision. Nothing in the platform converts a scorecard into an outcome, and moving a candidate to Hired or Rejected is always a manual act — see Managing interviews.