How to Set Up Greenhouse Scorecards for Structured Hiring
A Greenhouse scorecard is the evaluation form an interviewer fills after a session, built from the job’s interview plan and its focus attributes. You create them under Jobs → [Job] → Job Setup → Interview Plan, then attach scorecards per interview stage.
Get this right and debriefs run on evidence. Get it wrong and you end up with five free-text essays that mention different things, and a hiring manager asking “so is this person good or not?” for twenty minutes.
How does the Greenhouse scorecard structure work?
Greenhouse nests it three levels deep:
- Interview Plan — the ordered stages for the job.
- Interview Kit — the package for one stage: what to ask, what to evaluate.
- Scorecard — the form attached to that kit, made of focus attributes and a recommendation.
Focus attributes are the scored items. They come from a shared attribute library, so Technical Depth on a backend role is the same object as Technical Depth on a data role. That is what makes reporting across jobs meaningful.
Greenhouse’s default recommendation scale is four points: Strong No, No, Yes, Strong Yes. There is also a question type for free text and one for a numeric rating, but the four-point scale is the one to standardise on.
How do you build a scorecard step by step?
- Open the job, go to Job Setup → Interview Plan.
- Pick a stage and click Add Interview Kit (or edit the existing one).
- Under Scorecard, add focus attributes — three or four, from the library.
- For each attribute, set whether the field is required. Make attributes required when a blank answer would block your decision.
- Add key questions to the kit. Greenhouse shows these to the interviewer in the packet.
- Repeat per stage, then publish the job.
Set the required-field policy deliberately
Required fields sound good until a scheduled interview happens with the wrong panel and the interviewer cannot answer half of them. A pragmatic split: mark the recommendation required, mark one or two core attributes required, leave the rest optional. That way submission is never blocked hard enough that people abandon the form and email feedback instead.
How do focus attributes drive reporting?
Every submitted scorecard feeds the candidate’s aggregate view and the hiring team’s reports. Greenhouse slices by attribute across candidates, which lets you answer questions like “do our final-round candidates score lower on communication than on coding?” — a signal about your sourcing, not your candidates.
One caveat: attribute-level reporting only works if attributes are reused. If every job creates new attribute names, reports fragment and you cannot compare anything. Audit the attribute library twice a year; merge near-duplicates and archive the dead ones.
What should interviewers actually write?
Greenhouse’s own guidance, and general structured-interview practice, points the same way: describe behaviour, do not diagnose character.
| Weak note | Better note |
|---|---|
| “Great communicator.” | “Explained the tradeoff between read replicas and caching in under 90 seconds, with a concrete failure case.” |
| “Not a culture fit.” | “Pushed back on the team’s migration plan without asking what constraint drove it.” |
| “Seems smart.” | “Derived the O(n log n) bound unprompted, then questioned whether it mattered at this data size.” |
The right-hand column is auditable. If a hiring manager disagrees, they can point at the observation rather than the verdict.
What breaks scorecard discipline?
- Scorecards submitted days later. Memory reconstruction. Greenhouse timestamps submissions — check the delta during your next hiring retro.
- Interviewers reading each other’s scorecards first. Greenhouse shows submitted scorecards to other interviewers by default in many configurations. Consider restricting visibility until the debrief to preserve independence.
- No debrief owner. Someone needs to read every scorecard before the call. That person is usually the recruiter.
- Hiring manager pressure. When the manager has already decided, scorecards get retrofitted. Hard to fix with software; worth naming in your process doc.
How does this compare with a lighter tool?
If you are a team of eight hiring two engineers a year, the full interview plan and attribute library is a lot of admin. Greenhouse is priced per employee per year and built around the assumption that recruiting is continuous. A shared doc and a calendar event gets you 70% of the benefit at zero cost. The remaining 30% — cross-candidate comparison and audit trails — only starts to matter at volume.
Key Takeaways
- Greenhouse scorecards are built from the job’s interview plan and its shared focus attributes.
- Use the four-point scale (Strong No to Strong Yes) consistently across all jobs.
- Make the recommendation required, but keep most attributes optional so submission never blocks.
- Reuse attributes from the shared library or reporting fragments and comparisons break.
- Write behavioural evidence, not verdicts — the note should be auditable.
