How an ATS Reads Your Resume: Parsing Explained (2026)

An applicant tracking system does not "read" your resume the way a recruiter does. It runs it through a parser — software designed to extract structured data from an unstructured document. Understanding what happens at each step of that pipeline is the most direct path to fixing the real problems with your resume, rather than chasing folklore.

Step 1: Upload and file conversion

When you submit your resume through an application portal, the system receives the file and begins conversion. For a PDF, it extracts the text layer (or, in the case of a scanned image PDF, attempts OCR). For a .docx file, it reads the XML structure underneath.

The key point at this stage: if your PDF was created by scanning a physical document or saving from a design tool that embeds text as paths rather than characters, the extracted text may be empty or garbled. Always create your resume from a word processor or text-based PDF export — never from a photo or design file.

Step 2: Section detection

The parser looks for section headings — words like "Experience", "Education", "Skills", "Certifications" — and uses them to divide your resume into labeled regions. Content following the "Education" heading is tagged as education data; content under "Skills" is tagged as skills.

This is why non-standard headings cause problems. A heading like "Where I've Grown" or "My Arsenal" may not be recognized, and the content beneath it gets attributed to the wrong category — or dropped entirely.

Step 3: Field extraction

Within each section, the parser extracts specific fields. In the Experience section, it tries to identify: job title, employer name, location, start date, end date, and the description bullets. In Education: degree type, field of study, institution, and graduation date.

Multi-column layouts are a common source of extraction failures at this step. When two columns of text are present, many parsers read left-to-right line by line across the full width, merging content from unrelated sections. A job title from the left column may end up concatenated with a skill from the right column, making both useless.

What commonly breaks field extraction

• Tables and text boxes — content inside these elements is frequently skipped or extracted out of order.
• Headers and footers — contact information placed in the document header or footer is often not read. Put your name, phone, and email in the main body.
• Non-standard date formats — dates written as "Spring 2021" or "Q3 2020" may not parse. Use "May 2021" or "2021".
• Icons and symbols for contact info — a phone icon before your number or a LinkedIn icon before your URL can break extraction of those fields.
• Merged cells in tables used as layout — a common design trick that reliably causes parsing errors.
• Graphic elements overlapping text — visual skill bars, rating charts, or decorative dividers that sit on top of text cause the underlying text to be missed.

Step 4: Recruiter keyword search and filtering

Once your record is in the database, it surfaces (or does not) based on what recruiters search for. A recruiter working a software engineering role might search for "Python" and "AWS" and "senior engineer" — and only candidates whose parsed records contain those terms appear in the results.

The practical implication: even a perfectly formatted resume will not appear in search results if it uses different terminology than the recruiter searches. "Agile project management" and "Scrum master" overlap significantly in meaning but may not overlap in a keyword search. Mirroring the job description's own language is the most reliable fix.

Step 5: Optional ranking

Many systems offer a rank or match score based on keyword overlap between the job description and parsed resume fields. This score is shown to the recruiter as a sorting tool, not an automatic gate. Recruiters can sort by score, ignore it, or configure their own weighting. There is no universal published threshold that determines "pass" or "fail" — that decision is always made by a human.

Frequently asked questions

Can an ATS read a PDF resume?

Yes, as long as the PDF contains a real text layer — which it will if you saved it from Microsoft Word, Google Docs, or any standard word processor. The risk is layout complexity (columns, text boxes, tables) that trips up the parser, not the file format itself. A clean single-column PDF parses well in the vast majority of modern systems.

Does the ATS read my cover letter?

Many systems accept a cover letter separately and store it as a text field, but recruiters vary widely in how much weight they give it. The resume is the primary parsed document for keyword matching and field extraction. Focus your optimization effort there.

What happens to my resume if the parser fails?

A failed or partial parse typically means your record appears in the system with empty or incorrect fields. You may still be technically "in the database", but searches for your skills, title, or experience level will not surface you. From a practical standpoint, a bad parse has the same effect as not applying.

How do I know if my resume parsed correctly?

The only reliable way is to test it. Tools like ATSGrader simulate the parse step and show you how your resume's content is being extracted — all in your browser, with nothing stored or uploaded.

Related: what is an ATS and how does it work · ATS-friendly resume format · why resumes get rejected by ATS