Screening Rubric Builder

Convert a JD into a screening rubric specific enough that two reviewers reading the same resume reach similar advance/reject decisions. This is the pre-interview stage only: it decides who gets an interview, not how interviews are scored.

Workflow

1. Derive 5-8 screenable competencies from the JD. Extract them from the role's outcomes and must-haves. Each must be observable from an application (resume, cover letter, portfolio, work samples) and job-relevant - "ships schemas that survive scale," not "smart." Drop anything you cannot tie to a stated JD requirement. Never use "culture fit"; if culture matters, decompose it into named, assessable behaviors.
2. Weight each competency by impact on success. Assign weights summing to 100% that reflect how much each predicts performance in this specific role. Core role skills outweigh nice-to-haves. Write a one-line rationale per weight so it is defensible across the whole applicant slate. Never weight a proxy for a protected trait.
3. Write an anchored scale. Use a 1-4 scale with written behavioral anchors per competency: what a 1 application shows, what a 4 shows, with concrete evidence examples drawn from what is visible in an application. Anchors are what turn a vibe into a rating.
4. Require an evidence note per score. For every score the reviewer must cite the specific line, project, or artifact in the application that justifies it. No cited evidence, no score.
5. Set the advance rule up front. Define the weighted threshold to advance to interviews before reviewing any applicant, plus minimum scores on each must-have competency. Pre-committing the bar prevents post-hoc rationalization. A low total with one standout area is a reviewer conversation, not an automatic pass or reject.
6. Apply the rubric identically to every applicant for the role, and keep the scored records for consistency review.

Quality Bar

• Every competency traces to a specific JD requirement and is judgeable from the application alone.
• Weights sum to 100% and each carries a written rationale.
• Each competency has anchors for both ends of the scale, not just a label.
• The advance threshold and per-must-have minimums are written before the first applicant is read.
• A second reviewer scoring the same application from the same rubric lands within one point.

Do NOT

• Do NOT score the interview loop or design interviewer scorecards - that is hiring-scorecard's job; stop at the advance/reject decision.
• Do NOT auto-advance or auto-reject anyone; the rubric assists human judgment, it does not replace it.
• Do NOT include or infer age, gender, race, disability, family status, or background, or any proxy for them.
• Do NOT assess the same competency in both the screen and a later interview stage with no plan - measure it once, in the stage where it is most visible.
• Do NOT keep any competency you cannot defend as job-related; flag and cut it.