How to Write Hiring Criteria That Actually Rank Candidates
8 min read
You posted a job for a sales rep. 147 applications arrived in 72 hours. Half look identical, most claim "5+ years closing enterprise deals," and you have no systematic way to tell who actually fits. So you skim the first 20, pick three names that feel right, and hope.\n\nThe problem is not the volume. The problem is you are reading resumes without a rubric. A hiring criterion is a specific, measurable requirement with a weight that tells you how much it matters. When you define criteria before you read resumes, you turn a pile of claims into a ranked shortlist with evidence behind every decision. This guide shows you how to build that rubric, avoid the traps that waste your time, and score candidates so the best ones surface first.
Turn your job description into a rubric with 4-7 criteria
Most job descriptions are marketing copy written to attract applicants. A rubric is the inverse: a checklist you use to evaluate them. Start with your job description and extract the 4-7 things that actually predict success in the role.
Good criteria are specific, measurable, and non-overlapping. "Strong communication skills" is vague. "Conducted 20+ sales demos per month in a previous role" is a criterion you can score from a resume. "Experience with CRM software" and "Experience with Salesforce" overlap; pick one or combine them.
For a customer support role, your rubric might include: prior support experience (measured in months or ticket volume), written communication quality (assessed from cover letter or emails), familiarity with your product category, availability for evening shifts, and problem-solving examples. For a bookkeeper: years using QuickBooks, monthly transaction volume handled, payroll experience, and references from previous clients.
Each criterion needs a clear definition so you (and anyone else reviewing resumes) score the same candidate the same way. "3+ years QuickBooks, handling 200+ monthly transactions" is clear. "Accounting experience" leaves room for one reviewer to count an intro course and another to require a CPA. Write the detail you would give someone covering your desk for a week.
Separate must-haves from nice-to-haves with weights
Not all criteria matter equally, and treating them as if they do produces bad rankings. A software engineer who writes clean code but has never touched your specific framework is often better than someone who knows the framework but writes unmaintainable code. Weighting forces you to decide which matters more.
Assign each criterion a weight from 0 to 10, where 10 means "this is non-negotiable" and 0 means "ignore this entirely." A criterion weighted at 10 has twice the ranking impact of one weighted at 5. If prior industry experience is critical, give it a 9 or 10. If familiarity with a specific tool is a nice-to-have that you can train, give it a 3.
Must-haves get weights of 8-10. Nice-to-haves get 3-5. Anything below 3 probably should not be a criterion at all; if it does not move your decision, delete it. For a marketing manager role, "5+ years running paid campaigns" might be a 10, "experience with our specific ad platform" might be a 4, and "graphic design skills" might be a 2 or removed entirely if you have a designer on staff.
Weights also let you adapt one rubric to different hiring contexts. If you are filling a role in a new market, you might weight industry experience at 10. If you are backfilling a role on an experienced team, you might drop it to 5 and weight culture fit or learning speed higher. The same base criteria, re-weighted, become a different filter.
Score each criterion with quoted evidence, not gut feel
Once you have a weighted rubric, score each candidate on each criterion using a 0-10 scale. The score is not a guess. It is a measurement of how well the resume demonstrates that specific requirement, backed by a direct quote or concrete observation.
For "Conducted 20+ sales demos per month," a candidate who writes "Led 30+ product demos monthly, closing 18% of opportunities" scores a 9 or 10, and your evidence is that quoted sentence. A candidate who says "Presented to clients regularly" with no numbers scores a 3 or 4, evidence: vague claim, no volume. A candidate with no mention of demos scores 0-1.
This evidence discipline does two things. First, it makes your scores defensible. If someone asks why Candidate A ranked higher than Candidate B, you point to the rubric and the quoted lines, not a feeling. Second, it surfaces resume fluff fast. A candidate who claims "expert-level Python" but whose resume shows one 6-week bootcamp project scores low on "3+ years Python in production," and the evidence is the gap between the claim and the work history.
When the evidence is missing or contradictory, score it low and flag it. "Says 5 years of sales experience, but work history shows 18 months in role, prior jobs unrelated" is a 2-3 with a concern note. You will ask about it in the interview, but it does not get benefit-of-the-doubt points in the ranking.
Calculate the overall score as a weighted average
Each candidate's overall score is the weighted average of their per-criterion scores, scaled to 0-100. This is not subjective. It is arithmetic.
If your rubric has three criteria (A: weight 10, B: weight 5, C: weight 3), and a candidate scores 8 on A, 6 on B, and 9 on C, the overall score is ((8×10) + (6×5) + (9×3)) / (10+5+3) × 10 = 72.8, rounds to 73. A candidate who scores 10, 10, 0 on the same rubric gets ((10×10) + (10×5) + (0×3)) / 18 × 10 = 83. The second candidate ranks higher even though they completely lack criterion C, because you weighted A and B as more important.
This formula is the entire ranking engine. Candidates with overall scores of 70+ are your shortlist. Scores of 45-69 are maybes: they have gaps, but the gaps might be fixable or the role might be hard to fill. Below 45 is a reject unless you have information the resume does not show.
The power of the weighted average is that you can re-weight the rubric and re-rank instantly without re-reading resumes. If you initially weighted "prior industry experience" at 10 and got zero shortlist candidates, drop it to 5, and candidates who score high on other criteria will rise. If you weighted "availability for weekend shifts" at 3 and realize it is actually non-negotiable, raise it to 10, and the ranking reorders to surface candidates who mentioned weekend availability.
Avoid these four mistakes that break ranking
Mistake one: too many criteria. If your rubric has 15 criteria, you are either double-counting things or including requirements that do not actually matter. More criteria do not make the ranking more accurate; they make every candidate's score regress to the mean, and your shortlist becomes a tie. Stick to 4-7.
Mistake two: equal weights by default. If you do not assign weights, you are implicitly saying every criterion matters the same amount, which is never true. "Can start in two weeks" is not as important as "has managed a team of five." Weight them accordingly, or your ranking will prioritize the wrong candidates.
Mistake three: scoring candidates against each other instead of against the rubric. A candidate who scores 6 on "3+ years managing projects" gets a 6 whether they are the best in the pile or the worst. Comparative scoring ("this one is better than that one") hides whether anyone actually meets the bar. Score against the criteria, then rank by the overall.
Mistake four: changing the rubric mid-review. If you read 30 resumes, decide a criterion does not matter, delete it, and keep going, the first 30 and the last 30 are scored on different scales and cannot be ranked together. If you need to change the rubric, re-score everyone, or at minimum re-score anyone near the shortlist cutoff. Tools like SiftFirst handle re-weighting automatically without re-scoring, which is why changing weights is safe but changing criteria definitions is not.
Use the rubric in interviews and track what predicts success
The rubric does not stop at resume screening. Bring it into the interview. If a candidate scored low on "experience with high-volume support tickets" because their resume did not mention ticket volume, ask them directly: "What was your average daily ticket volume in your last role?" If they score high on paper but you want to verify, ask for specifics: "You mentioned closing 18% of demos. Walk me through your process for a typical demo."
Candidates who score well across the board but have one red flag (a concern note from the evidence review) are the ones where interviews add the most value. The rubric tells you what to probe. A candidate with a 9 on technical skills and a 3 on communication because their cover letter was two sentences should be interviewed with an eye on how they explain complex ideas, not on rehashing their technical background.
After you hire, track which criteria actually predicted success. If you weighted "prior industry experience" at 10, and six months later your best performer is someone who scored a 4 on that criterion but a 10 on problem-solving, adjust the rubric for the next hire. Your rubric is a hypothesis about what matters. Real performance data is the test.
Key takeaways
- ✓A hiring rubric is 4-7 specific, measurable, non-overlapping criteria extracted from your job description.
- ✓Assign each criterion a weight from 0-10 based on how critical it is; must-haves get 8-10, nice-to-haves get 3-5.
- ✓Score each candidate 0-10 on each criterion with a direct quote or observation as evidence, not gut feel.
- ✓Calculate the overall score as a weighted average of per-criterion scores, scaled to 0-100; 70+ is shortlist, 45-69 is maybe, below 45 is reject.
- ✓Avoid too many criteria, equal weights, scoring candidates against each other, and changing the rubric mid-review.
- ✓Use the rubric in interviews to probe gaps and verify high scores, then track which criteria predicted success to refine future rubrics.
Screen your applicant pile in minutes
SiftFirst ranks every applicant against criteria you set, with a quote from each resume behind every score. You set the criteria; you decide every hire. Free to try, no signup.
Try a free screening →FAQ
How many criteria should a hiring rubric have?
4-7 criteria is the sweet spot. Fewer than four and you are probably missing something that matters. More than seven and you are either double-counting ("communication skills" and "client-facing experience" often measure the same thing) or including requirements that do not actually move your decision. Each additional criterion dilutes the ranking, so keep only the criteria that predict success in the role.
What if no candidates score above 70 on my rubric?
Lower the weights on nice-to-haves or relax the definitions on criteria where you can train. If "prior industry experience" is weighted at 10 and no one has it, but several candidates score high on everything else, drop industry experience to 5 or 3 and see who rises. If everyone scores low across the board, your job description might be attracting the wrong audience, or your expectations for the role (and the salary) might be misaligned.
Can I use the same rubric for multiple hires in the same role?
Yes, and you should. A rubric that worked once is a hypothesis you can test and refine. If you hired three customer support reps using the same weighted criteria, and one is now your top performer, compare their original screening scores to the others. The criteria where the top performer scored highest are the ones to weight more heavily next time. Over several hires, your rubric becomes a tested model of what works, not a one-time guess.