AI resume screening is software that reads your application and helps an employer decide whom to look at first. In most setups it does four things: it parses your resume into a profile, applies the employer's knockout questions, compares your qualifications with the job and gives you a grade or category, and puts you in a sorted list for a recruiter. The vendors' own documentation says people make the final call. In practice, a low grade on a long list can mean nobody opens your resume. This guide walks through each step, the rules that apply to it, and what you can change.
How common it is
In a Harvard Business School and Accenture survey of more than 2,250 executives in the US, UK and Germany, more than 90% of employers said they use their recruiting systems to initially filter or rank candidates: 94% for middle-skills roles and 92% for high-skills roles. Not all of that is AI. Much of it is rules and keyword filters. Today large applicant tracking system (ATS) vendors document AI matching features of their own; two are described below.
Step 1: Parsing turns your resume into fields
When you upload a resume, the ATS extracts your name, contact details, jobs, dates, education and skills into a candidate profile. That step is called resume parsing. Everything after it works on the parsed profile, so a bad parse affects every later step.
Greenhouse's help page on unsuccessful resume parses lists what goes wrong: files over 2.5MB, graphics and photos, resumes uploaded as images, tables, headers and footers, contact details in a header or text box, columned layouts, missing sections and incomplete job titles. A failed parse means the employer has to enter details by hand; a partial one needs correcting.
What you control: the file. A single-column, text-based .docx or PDF parses cleanly. Our ATS friendly resume template is built for this step.
Step 2: Knockout questions filter by your answers
Many applications include questions with fixed answers: work authorization, willingness to relocate, a license, a degree. Employers can tie rules to them. Greenhouse's auto-reject rule "uses custom application questions to screen and reject candidates based on their responses," and its auto-advance rule can move a candidate straight to a later stage.
This step is not AI at all. It is a rule the employer writes. It is also one of the few steps that can reject you without a person involved.
What you control: your answers. Read every question, answer truthfully, and skip roles where a hard requirement rules you out.
Step 3: AI matching grades you against the job
This is the step most people mean by AI resume screening. The software compares your parsed profile with the job's requirements and gives you a grade or label. Two examples from vendor documentation:
- Workday HiredScore. Workday's documentation describes candidate grades from A to D, based on "how well their resumes match the skills and qualifications specified in the job description." An A meets all required qualifications and most preferred ones. A B meets all required qualifications. A C meets "most but not all basic qualifications." A D does not meet most required qualifications.
- Greenhouse Talent Matching. Greenhouse's Talent Matching uses criteria the employer sets from the job description and, if available, the interview scorecard. It sorts candidates into "Strong match", "Good match", "Partial match" and "Limited match". Its AI principles say Greenhouse does not assign a single numerical score and shows categories "with explanations rather than composite fit scores."
Both work from what is written on your resume. Neither can credit a skill you have but did not write down, or a requirement you meet under a different name.
What you control: whether the requirements you meet are visible. Put each one in plain words, in the posting's own terms where they honestly fit. Our tailored resume guide shows how.
Step 4: A recruiter works through a sorted list
The vendors describe these grades as help for people. Greenhouse says Talent Matching "does not automatically advance or reject candidates" and that recruiters and hiring managers make those decisions. Workday's bias audit page describes HiredScore Spotlight's output as match grades, with A as the closest match.
That human step is where volume matters. In Greenhouse's 2026 candidate survey, 81% of US candidates said they had been rejected from a job in the past year without ever speaking to anyone. In 25 interviews with US recruiters in late 2025, Enhancv found that 11 had an AI match score in their system: 9 used it as a guide and 2 as a hard decision.
What you control: timing and fit. Apply early to roles you match well, and if you have a contact at the company, a referral puts your application in front of a person directly.
Step 5: Interviews, sometimes also AI-assisted
Some employers add AI to later stages too, such as one-way video interviews or game-based assessments. Greenhouse, for example, says its AI Interviewer feature undergoes monthly bias audits by an outside firm, alongside Talent Matching. If a disability affects how such a test measures you, the Americans with Disabilities Act lets you ask for an accommodation (more on that below).
One application through the five steps
To make this concrete, follow one invented applicant, Maya, applying for a payroll specialist role at a company using an ATS with AI matching.
- Parsing. Maya uploads a two-column resume made in a design app, with her contact details in a sidebar. The parse picks up her jobs but misses her email and mixes her skills into her first job. Her profile is incomplete before anyone looks at it.
- Knockout questions. The form asks whether she has a payroll certification. She has one, answers yes, and passes this step.
- Matching. The job's required qualifications include multi-state payroll and a named payroll system. Maya has done both, but her resume says "processed pay for staff in several locations" and lists the system only in the scrambled skills section. The match tool can only credit what it reads, so she lands in a lower category than her experience deserves.
- Recruiter review. The recruiter sorts by category and works down the list. A shortlist is full before she is reached.
- Interview. She never gets there.
Now the same application with three changes: a single-column resume with contact details in the body, a bullet that says "Ran multi-state payroll for 450 employees in 6 states", and the payroll system named in that bullet and at the top of the skills list. Every fact is the same. The parse is clean, the match tool can see both requirements, and her application sits higher on the list the recruiter works through.
What AI screening cannot see
- Skills you have but did not write down.
- Experience described in your old employer's internal vocabulary.
- The reason behind a gap or a career change, unless you state it.
- Transferable work from freelance, military, academic or volunteer roles, unless it is described in the posting's terms.
The Harvard Business School report found 88% of employers agreed that qualified high-skills candidates are "vetted out of the process because they do not match the exact criteria established by the job description." Rigid matching is a known weakness, and the fix on your side is to make the match explicit.
Each gap has a matching fix on the page. Write skills into the bullets where you used them, not only in a skills list. Translate internal titles and team names into common terms. Give a gap or a change of direction one plain line, with dates. And describe freelance, military or volunteer work with the same structure as paid roles: title, organization, dates and results.
The rules that apply
AI screening is legal in the US, but employers stay responsible for what their tools do.
- New York City. Local Law 144 covers a tool that "substantially assists or replaces" discretionary decision-making, and the city's FAQ says screening counts. Employers need a bias audit before use, must publish its summary, and must give notice to candidates who live in the city at least 10 business days before use.
- Colorado. SB 26-189, signed on May 14, 2026, covers automated decision-making technology that materially influences employment decisions. It gives people notice, a plain-language description of the system's role after an adverse outcome, and a right to request meaningful human review.
- California. Civil Rights Council regulations in effect since October 1, 2025 apply anti-discrimination law to automated-decision systems used in hiring.
- Disability. Under Title I of the ADA, employers must make reasonable accommodations for qualified applicants and give tests in a way that measures skills rather than an impairment.
- Courts. In Mobley v. Workday, a federal court in Northern California on May 16, 2025 granted preliminary certification of a collective of applicants aged 40 and over in an age discrimination claim about Workday's AI recommendation system. The order is an early procedural step, not a finding that the system discriminates.
If you want a person to review your application instead, read should I opt out of AI resume screening.
What to do as a candidate
- Use a plain, single-column resume so the parse is clean.
- Read knockout questions carefully and answer truthfully.
- Put each required qualification you meet in plain words in the top half of page one.
- Use the posting's terms where they honestly describe your work.
- Explain gaps and career changes in one line rather than leaving them blank.
- Apply early, and use referrals where you have them.
- Never add hidden text or skills you lack. Hiring managers report catching both, as our post on whether ATS can detect AI generated resumes explains.
For the twelve most common ways applications fail inside these systems, see common ATS resume rejection reasons. For wider numbers on applications and hiring, see our job search statistics.
For the bigger picture, read what an AI career agent does, and how AI fits into the rest of an AI job search.
Finding your next role
Meet Elip, an AI career agent. Add your resume and say what you want. Each morning it brings you a few jobs that fit, each with the reason. It rewrites your resume for each role, with your approval. It is free for candidates.
Sources
- Fuller, Raman, Sage-Gavin, Hines, "Hidden Workers: Untapped Talent", Harvard Business School Project on Managing the Future of Work and Accenture (September 2021): https://pubwww.hbs.edu/managing-the-future-of-work/Documents/research/Hidden%20Workers%20Executive%20Summary.pdf
- Greenhouse Support, "Unsuccessful resume parse": https://support.greenhouse.io/hc/en-us/articles/200989175-Unsuccessful-resume-parse
- Greenhouse Support, "Application Rules Overview": https://support.greenhouse.io/hc/en-us/articles/203105595-Application-Rules-Overview
- Greenhouse Support, "Talent Matching": https://support.greenhouse.io/hc/en-us/articles/41396009937307-Talent-Matching
- Greenhouse, "Our AI principles": https://www.greenhouse.com/ai-principles
- Workday HiredScore documentation, "Candidate grades": https://doc.workday.com/hiredscore/en-us/workday-hiredscore/recruiter-productivity-/reference--candidate-grades.html
- Workday, "Responsible AI and Bias Mitigation for HiredScore Spotlight": https://www.workday.com/en-us/legal/responsible-ai-and-bias-mitigation.html
- Greenhouse, 2026 candidate survey release (October 20, 2026): https://www.greenhouse.com/newsroom/6-in-10-job-seekers-are-rethinking-how-they-write-to-avoid-sounding-like-ai-even-as-most-use-it-in-their-search
- Enhancv, recruiter interviews on ATS rejection (25 US recruiters, September to October 2025): https://enhancv.com/blog/does-ats-reject-resumes/
- NYC Department of Consumer and Worker Protection, "Automated Employment Decision Tools: Frequently Asked Questions": https://www.nyc.gov/assets/dca/downloads/pdf/about/DCWP-AEDT-FAQ.pdf
- Colorado General Assembly, SB 26-189, "Automated Decision-Making Technology": https://leg.colorado.gov/bills/sb26-189
- California Civil Rights Department, Civil Rights Council regulations on automated-decision systems (June 30, 2025): https://calcivilrights.ca.gov/2025/06/30/civil-rights-council-secures-approval-for-regulations-to-protect-against-employment-discrimination-related-to-artificial-intelligence/
- Americans with Disabilities Act, Title I, 42 U.S.C. 12112(b) (ADA.gov): https://www.ada.gov/law-and-regs/ada/
- Mobley v. Workday, Inc., No. 23-cv-00770-RFL (N.D. Cal.), Order Granting Preliminary Collective Certification, May 16, 2025: https://www.akingump.com/a/web/j136HnNQbGP5id5rwiJBvE/aqeuu5/3_23-cv-770_128-order.pdf






