Resume Screening Criteria: A Defensible Framework for Managers

Good resume screening criteria do two jobs: they separate people who can do the work from people who can't, and they leave a written record showing why. The simplest structure is two buckets. First, a short list of pass/fail must-haves tied to the job. Second, a handful of weighted preferences scored on a fixed scale. This guide shows how to build both, run them in about two minutes per resume, and keep the process fair when AI is involved.

TL;DR

• Start from the job, not the resume pile. Every criterion should map to an essential function of the role.
• Use two buckets: pass/fail knock-outs (licenses, work authorization, availability) and weighted preferences scored 1 to 5.
• Sort the first pass into Yes, Maybe and No. Spend your attention on the Maybes.
• Write down the rubric before you open the first resume, and keep your notes. Under federal guidelines, you may need to prove your criteria are job-related.
• If software scores or rejects candidates, keep a human able to review the rejects.

Table of Contents

• Why screening criteria matter more than speed
• Screening is not filtering
• Step 1: Define the job before you read a single resume
• Step 2: Set your knock-out criteria (bucket one)
• Step 3: Score preferences on a 1 to 5 rubric (bucket two)
• Step 4: Sort into Yes, Maybe and No
• A resume screening checklist you can run in two minutes
• Using AI and ATS tools without losing control
• Our perspective: criteria first, software second
• Where Heyhive fits
• Common mistakes to avoid
• FAQ
• Sources

Why screening criteria matter more than speed

Most managers screen the way they read email: fast, by gut, and with whatever mood they are in. That works until you have 80 applicants for four openings, which is common in restaurants, retail and field services.

The volume problem is real. One recent roundup reports that ATS software is used by 98% of Fortune 500 companies, and that 75% to 80% of resumes never reach a human. Treat those as rough industry figures, not guarantees, but the direction is clear. Software or a tired reviewer is making the first cut in many hiring funnels.

The same source says that when a person does look, the scan is short: an average of 7 to 17 seconds. In that window, a reviewer who has no written criteria will fall back on pattern matching: a familiar employer name, a clean layout, a school they like. None of those predict hourly job performance well, and some can lead to unfair outcomes.

Written criteria fix three things at once:

• Consistency. Every applicant is measured against the same yardstick.
• Speed. You decide faster when you know what you are looking for.
• Defensibility. You can explain a decision months later.

Screening is not filtering

Filtering removes people. Screening measures whether someone has what the job requires. The two get confused, especially when a tool offers a one-click "match score."

Testimony submitted to the EEOC on automated hiring tools makes the point that employers who rely on arbitrary automated scores without a documented, job-related validation process risk disparate impact. Put plainly: a keyword count is not a measure of ability.

The legal backdrop is the Uniform Guidelines on Employee Selection Procedures. Under 29 CFR Part 1607, a selection rate for any group that falls below four-fifths (80%) of the rate for the highest group is generally regarded as evidence of adverse impact. If that happens, you may need to show the criteria are job-related and consistent with business necessity. A rubric that ties resume evidence to essential duties is how you show that.

This is general information, not legal advice. If your screening produces lopsided results for a protected group, or you are unsure whether a criterion is lawful, talk to an employment attorney or your HR counsel before you keep using it.

Step 1: Define the job before you read a single resume

Start with what the person will actually do in the first 90 days. For a line cook, that might be station prep, ticket times, food safety, and working closing shifts. For a field technician, it could be driving a company vehicle, using a specific tool set, and arriving on time at customer sites.

Then sort each requirement:

1. Essential: the job cannot be done without it.
2. Trainable: a good hire can learn it within weeks.
3. Nice to have: useful, but never decisive.

The gold standard described in one current checklist is to map every criterion directly to the essential job functions in your job analysis. If you cannot point to a duty that a criterion supports, cut the criterion.

This step is where hourly hiring differs from corporate recruiting. Degrees rarely belong on the list. Availability, reliability signals and specific certifications often do.

Step 2: Set your knock-out criteria (bucket one)

Knock-outs are binary. The candidate has it or does not. Keep the list short, usually three to five items, and make each one verifiable.

Typical examples for shift-based work:

• Legal authorization to work in the United States
• A required license or certification (food handler card, forklift, CDL, state trade license)
• Minimum age where the role or local law requires it
• Availability for the shifts you actually need to fill (for example, weekends or closing)
• Ability to perform the essential functions, with or without reasonable accommodation

Two cautions. First, do not knock out for things you cannot defend, such as employment gaps or a lack of a degree for a role that does not need one. Second, ask about work authorization the way your counsel recommends, consistently, and the same for everyone.

Knock-outs work best when asked up front, in the application itself, rather than inferred from resume text. Our guide to knockout questions HR can use for legally defensible hiring goes deeper on wording and on how many to use.

Step 3: Score preferences on a 1 to 5 rubric (bucket two)

Once a candidate clears the knock-outs, you rank the rest. Industry practitioners recommend weighted preferred criteria scored on a rubric, typically 1 to 5, so you can compare qualified applicants consistently.

Here is a sample rubric for a restaurant shift lead. Adjust it to your own role.

A few rules keep the rubric honest:

• Write it before you read resumes. Changing weights after you see who applied is how bias sneaks in.
• Score evidence, not impressions. "Led closing for a 120-seat restaurant" earns points. "Seems professional" does not.
• Keep a one-line note per candidate. That note is your record.

What to look for in the experience section

In the same Jobgether roundup, 88% of hiring managers prioritize the work experience section and 75% scan for quantifiable achievements rather than generic duties. Treat that as a signal of what reviewers say they do. For your rubric, the useful question is: does this bullet show what the person did and what changed because of it?

Compare two bullets:

• "Responsible for customer service." (a duty)
• "Handled 60+ covers per shift and cut average wait for tables during Friday rushes." (evidence)

The second earns a higher score because you can picture the work.

What to do about skills sections

Career-center guidance from universities, such as the UIC Engineering resume checklist, advises candidates to leave out soft-skill words like "organized" and to reserve a skills section for verifiable hard skills: software, equipment, languages, certifications. Use the same logic as a reviewer. A list of adjectives earns nothing. A named tool or certification can be checked. Look for soft skills in the bullets, where they show up as actions and outcomes.

Step 4: Sort into Yes, Maybe and No

A ranked list of 200 names looks precise but is not. Experienced hiring managers quoted by Monster suggest a simple Yes/Maybe/No bucket system for the first pass, then spending your human attention on the Maybes, where candidates need a closer look to separate real potential from a keyword match.

Here is how it works in practice:

• Yes: passes all knock-outs and scores strongly on the rubric. Move to a phone screen or interview.
• Maybe: passes knock-outs but has gaps or unclear evidence. Take a second look, or send two or three clarifying questions.
• No: fails a knock-out or scores low across the board. Record the reason in a short note.

If you need to fill roles fast, our playbook for hiring hourly staff quickly without payroll errors covers what to do after the first pass.

A resume screening checklist you can run in two minutes

Use this order for every application. Do not skip steps, and do not reorder them for favorite candidates.

1. Confirm knock-outs. Authorization, license, availability, location or travel radius.
2. Scan the most recent role. Is the work similar to yours?
3. Check for evidence. Look for numbers, scope, tools and outcomes, not just duties.
4. Verify required credentials. Certifications, dates and issuing bodies, where listed.
5. Look at tenure patterns. Short stints are a prompt to ask a question, not a reason to reject.
6. Score the rubric. One score per criterion, each with a brief reason.
7. Assign Yes, Maybe or No. Write one line explaining why.
8. Flag questions for the interview. Anything unclear becomes a question, not an assumption.

What to ignore on purpose: photos, names, graduation years, addresses beyond commute feasibility, and anything else not tied to the job. Removing them lowers the chance that irrelevant details color the score.

Using AI and ATS tools without losing control

Automation is useful for the heavy lifting: parsing, deduplicating and matching against your knock-outs. The risk is letting a number replace judgment.

An academic audit of AI résumé screening, Fairness Is Not Enough, looked at how automated screening can seem neutral while leaning on weak statistical cues, and why relevance scores based on keyword frequency are a poor stand-in for understanding someone's achievements. The practical takeaway: ask any tool to show its reasoning, not just a percentage.

What to demand from any screening software, ours included:

• Explanations. Why did this candidate get this score, in terms of your criteria?
• Reviewable rejects. A manager should be able to open auto-rejected candidates and reverse a call.
• Your criteria, not a hidden model. Knock-outs and weights should be visible and editable.
• Regular checks. Compare pass rates across groups and look for the pattern the four-fifths rule flags.

Oversight design is its own discipline. If you want a deeper look at how teams structure review of AI output, this piece on human-in-the-loop oversight models lays out several approaches you can adapt to hiring.

Keep the resume format from hurting good candidates

Parsing failures cost you people. University career centers advise candidates to use a simple format, since ATS software can misread columns, tables, headers, footers and graphics. If your pipeline drops or garbles resumes that fail to parse, spot-check them. A strong applicant with a designed two-column resume should not disappear because a parser could not read it. Sample the rejected pile every week.

Our perspective: criteria first, software second

We build workforce software, so we see what happens after the hire: schedules, shifts, time clocks, payroll hours. That vantage point shapes how we think about screening.

The cheapest bad hire to prevent is the one you never should have advanced. A criterion that does not connect to the job, whether it is a degree for a role that does not need one or a keyword that proves nothing, costs you good people and adds legal risk.

Here is what we believe:

• AI should draft, people should decide. It is the same principle behind our scheduling: the system generates, and you approve before anything is published. Hiring deserves the same control.
• Knock-outs belong at the front. Clear, job-related questions asked early save everyone time.
• A rubric is a kindness. To candidates, it means a fair read. To managers, it means fewer second-guesses.
• Hiring and scheduling are one problem. Availability, certifications and overtime limits matter at the resume stage because they decide whether you can actually staff the shifts you have.

If a tool cannot tell you why it ranked someone, treat that as a warning sign. You are always in control of who moves forward.

— Heyhive

Where Heyhive fits

If you hire hourly and field staff, screening is only the first step. Heyhive brings AI-assisted hiring, AI scheduling, GPS-verified time clocks and payroll-ready hours into one interface, with managers approving before anything goes live. That means the criteria you use to hire, such as certifications and availability, carry through to how you schedule. Start with the platform overview and see how it works for your team.

Common mistakes to avoid

• Changing criteria midstream. If the rubric moves after you see candidates, you cannot show consistency.
• Over-weighting pedigree. Brand-name employers and schools are poor predictors in most hourly roles.
• Treating gaps as red flags. Ask about them. Do not assume.
• Using an unexplained score as the decision. Scores can inform a decision, but a person should own it.
• Not keeping notes. If you cannot say why you rejected someone, you cannot defend it.
• Skipping the audit. Compare results across groups regularly, and escalate to counsel if you see a gap.

FAQ

What should be the average ATS score of a resume?

There is no standard number. Scores differ by vendor and are based on each tool's own method, so a 70 in one system means nothing in another. Use scores as a sorting aid at most, and judge candidates against your own knock-outs and rubric.

How do resumes get screened?

Usually in two passes. Software parses and filters on your knock-outs and keywords, then a person scans what remains. Industry data suggests recruiters spend roughly 7 to 17 seconds on an initial look, so a written rubric helps you use that time well.

What is the gold standard for resumes?

From a screener's view, it is a clear, single-column, text-based document with standard headings and concrete achievements in the experience bullets. Career-center guidance, such as the UIC checklist, favors this because it is easy for both software and people to read.

Do all resumes need a skills section?

No. A skills section is most useful when it lists hard, verifiable items such as software, equipment, certifications or languages. If a candidate leaves it out but shows the same skills in their job bullets, score the evidence, not the section.

Is a Yes/Maybe/No sort enough for a big applicant pool?

For a first pass, yes. It is faster than ranking everyone and keeps your focus on the Maybes. Apply the rubric in more detail once you have a short list to compare.

Sources

• How Recruiters Evaluate Remote Job Applications in 2025, Jobgether
• Resume Screening Checklist for Recruiters, ResumeWorld
• 29 CFR Part 1607, Uniform Guidelines on Employee Selection Procedures, eCFR
• Testimony of Matthew Scherer, U.S. Equal Employment Opportunity Commission
• How to Review a Resume in 10 Steps, Monster
• UIC Engineering Resume Checklist
• Fairness Is Not Enough: Auditing Competence and Intersectional Bias in AI-powered Résumé Screening, arXiv

Recommended

• 1–3 Knockout Questions HR Should Use for Legally Defensible Hiring