ATS Guide · 2026-06-10

How ATSGrader Scores Your Resume: Methodology & Sources

A resume score is only as honest as the rules behind it. This page documents exactly what ATSGrader checks, the public research each check is based on, and — just as important — what our score cannot tell you. No vendor publishes its ranking algorithm, and anyone who claims to have "cracked" Workday or Greenhouse is selling something.

The short version

ATSGrader is a rule-based engine with 23 checks that runs entirely in your browser. It scores the two things that decide whether your resume surfaces and survives: (1) can screening software parse and find it, and (2) can a human recruiter skim and believe it in a few seconds. The score is a directional diagnostic of best practices — not a simulation of any specific company's system.

What ATS software actually does

An Applicant Tracking System parses your resume into structured fields, stores it, and gives recruiters tools to search, filter and rank the applicant pool. Adoption is near-universal at scale: Jobscan's 2025 analysis detected an ATS at 97.8% of Fortune 500 companies^[2], and Harvard Business School's employer survey found more than 90% of employers use their system to initially filter or rank candidates^[1].

The crucial nuance: in most companies the software doesn't "auto-reject" you. The real mechanism is quieter — recruiters run keyword and filter searches over the pool (LinkedIn documents its Boolean search operators publicly^[4]), and resumes that don't match the search, or that parsed into garbage, are simply never seen. Employers themselves admit this filtering overshoots: 88% agree that qualified candidates are vetted out because they don't match the exact job-description criteria^[1].

The five score categories and their basis

1. Keyword match (heaviest weight)

We extract weighted terms from the job description you paste — required skills, tools, qualifications — and check which appear in your resume, including stem variants ("developing" matches "development"). Basis: recruiter search and ranking tools operate on term matching^[4]; a skill that isn't in your resume can't match anyone's search. This is the closest thing to a "law" in this domain.

2. Structure & contact info

Standard section headers (Experience, Education, Skills), detectable contact block, LinkedIn URL, parseable text. Basis: parsers read linearly and demonstrably scramble or drop content in tables, columns and graphics — Jobscan documented Lever dropping an entire skills section from a two-column layout^[5]. Creative headers ("Battle Scars") can orphan whole sections.

3. Impact & achievements

Share of bullets that contain numbers, and bullets that open with action verbs instead of duty phrases ("responsible for"). Basis: this is for the human pass, not the software — Ladders' eye-tracking study clocked the initial recruiter screen at 7.4 seconds^[3]. Quantified, front-loaded bullets are the standard advice of every serious career service for surviving that skim; we make the convention checkable.

4. Clarity & length

Word count bands, bullet length, run-on detection. Basis: same 7-second reality^[3] plus long-standing recruiter convention (1 page per ~8 years of experience). These are heuristics and we say so.

5. Red flags

Clichés ("team player", "hard-working"), missing contact data, date of birth (a bias-compliance concern in US/UK/EU hiring practice), emoji and exotic glyphs that render as garbage in recruiter views. Basis: hiring-practice convention and parser behavior; these are judgment calls a careful human reviewer would also flag.

The "75%" myth — and why we don't use it

You've seen the claim everywhere: "75% of resumes are rejected by ATS before a human sees them." We traced it to its source so you don't have to. It originates in a 2012 trade-press quote from Preptel, a now-defunct resume-optimization vendor, claiming ATS "kill 75% of candidates' chances"^[6] — no methodology was ever published, and the figure later mutated into the "never seen by a human" version. Recruiters who've investigated it estimate the opposite: the large majority of applications do get human review. We used this stat in our own early copy, found it didn't survive scrutiny, and removed it. What the research does support is quoted above: near-universal automated filtering^[1]^[2] that employers themselves admit vets out qualified people^[1].

What our score can't tell you

It is not any vendor's real algorithm. Workday, Greenhouse, Taleo and iCIMS don't publish theirs; ours is an evidence-based proxy for findability and readability.
A green score is not a job guarantee. Your actual experience, the applicant pool and the recruiter's judgment dominate outcomes.
We only see text. The scan runs on the text in your browser — visual design issues in the original file are checked only insofar as they affect extractable text.
Honesty is on you. The keyword table shows gaps; we tell you to add only terms you can defend in an interview. Stuffing skills you don't have gets you past software and burned by humans.

See the methodology in action — scan your resume

Free scan · no signup · your resume never leaves your browser

Check my resume free →