How to Check If Your Resume Is ATS Ready: A Seven-Point Self-Audit

You can check most of what an applicant tracking system needs from your resume yourself, in about ten minutes, with no score tool. Seven checks cover it: a working text layer (you can select and copy the text), a single-column layout with no tables or text boxes, standard section headings, contact details in the body of the document rather than the header, every role carrying a title, organization, dates, and bullet points, skills named explicitly, and education spelled out in full. A resume that passes all seven gives a parser everything it is built to extract. Here is each check, what it catches, and what the evidence behind it actually says.

Check 1: the text-layer test

Open your resume in any PDF viewer, select all, copy, and paste into a plain text editor. Three outcomes matter. If the text appears complete and roughly in reading order, you have a healthy text layer and the rest of the audit will go quickly. If lines interleave or sections come out scrambled, a parser walking the same text layer inherits the same scramble. And if little or nothing pastes at all, your resume is effectively a picture of a resume: text embedded in graphics is not reliably read, and parsers generally ignore it.

One nuance keeps this honest. Commercial parsers are more format-tolerant than the folklore suggests. Affinda's parser documentation, to take a primary source, lists PDF, DOC, DOCX, TXT, RTF, HTML, and even image formats like PNG and JPG handled through OCR. So PDF versus DOCX is not a pass-fail question, and a scanned resume is not automatically invisible. But OCR is a fallback, not a plan. A real text layer is the version of your resume you control.

Checks 2 to 4: the structure a parser walks

Parsing runs as a staged pipeline: the software segments the document by standard section headings, categorizes what it finds into typed fields (a date range, a job title, an organization), then indexes those fields for recruiter search. Each structural check below exists because some layout choice interferes with one of those stages.

• One column, top to bottom. In one informal test that ran the same resume through eight ATS platforms, a two-column layout parsed incorrectly in seven of the eight, tables were skipped by five, and sidebar text boxes were dropped by four. Treat that as a risk reading from a small test, not a universal law: modern layout-aware parsers handle columns far better than legacy ones, and the risk is shrinking year over year. The graduated rule the evidence supports: text boxes and sidebars are the most dangerous, tables next, columns a moderate risk.
• Standard section headings. Parsers segment on headings like Work Experience, Education, and Skills. No public source enumerates the exact strings each parser accepts, so the safe move is the plainest conventional name. 'Where I've Been' is a segmentation gamble; 'Work Experience' is not.
• Contact details in the body. Some parsers treat the document header and footer as a separate region and skip it entirely. In that same eight-system test, six of the eight missed contact information placed in a Word header, producing candidate profiles with no name or email attached. Put your name, email, and phone at the top of the body of the page.
• Standard bullet glyphs. Decorative symbols such as checkmarks and arrows can be replaced with junk characters or dropped, merging your bullets into one paragraph. Use the plain round bullet.

Checks 5 to 7: the fields a parser is trying to fill

Parsers extract into a fixed schema. Affinda's documented field list runs past a hundred entries, but the core set is: name and contact, then for each role a job title, organization, location, start and end dates, and a description; a skills list; and for education the institution, degree, and dates. The completeness half of the audit asks whether your resume fills every slot.

• Every role carries all four parts: a title, an organization, a date range, and at least one bullet of description. A role missing dates cannot be placed on a timeline; a role missing an organization may not register as a role at all.
• Dates are explicit and consistent. Parsers are built to recognize ranges like 'January 2018 to April 2022' and to mark a current role. Pick one format and use it everywhere. (Which exact format parses most reliably is genuinely unsettled in the public record; consistency is the defensible rule.)
• Skills appear as a literal list. Recruiter search runs against extracted fields, so a skill that lives only inside a paragraph of prose is easier to miss than one in a dedicated Skills section. Name the tools and competencies you want to be found for, in plain words.
• Education is complete: institution, degree, field, and dates. A partial entry like 'State University' leaves empty fields the schema has slots for.

About free ATS score tools

Run your resume through three free 'ATS score' checkers and you will likely get three different numbers, and none of them is a number any employer sees. Match scoring inside real systems exists but is not uniform: Workday ranks candidates by a match score, Greenhouse relies on a human-graded scorecard rather than an algorithmic rank, and Lever surfaces no candidate-facing score at all. And no published source provides validated weights for an ATS-friendliness score, so every third-party tool is choosing its own. Treat any single score as one tool's opinion, useful for surfacing specific fixable issues, not as a verdict.

A score is also not a rejection forecast. In a small 2025 study of 25 US recruiters (run by a resume-tool vendor, so read the figures as directional rather than statistical), 23 of the 25 said their systems never auto-reject on formatting, content, or missing keywords. The automated rejections that do happen come overwhelmingly from knockout questions about eligibility, such as work authorization and licenses. And the famous claim that 75 percent of resumes are rejected by ATS before a human sees them traces back to a 2012 sales pitch by a startup that folded the following year, with no published methodology behind it.

What a deterministic check can and cannot tell you

A deterministic completeness check, which is what this audit amounts to, can tell you one thing with confidence: whether a parser will come away with your facts intact. Text layer present, one column, standard headings, contact in the body, every role titled and dated, skills explicit, education complete. Those are checkable, and fixing them removes real, documented failure modes.

What it cannot tell you: whether a recruiter will like what they read, how any particular employer has configured their system, or whether you clear the eligibility questions that do most of the automated filtering. No checkable rule covers judgment. So spend the ten minutes, fix what the audit flags, and then put the bulk of your effort where the decision actually gets made: the content itself, written for the human who reads it after the machine has done its sorting.

Sources

• Affinda resume parser API docs (accepted formats, extracted fields) ↗
• QuickCV: informal parse test of one resume across eight ATS platforms ↗
• Enhancv: interviews with 25 US recruiters on ATS auto-rejection ↗
• Jobscan: ATS-friendly formatting guidance ↗