What an ATS parser does
An applicant tracking system (ATS) stores every application in a database recruiters search. Before it can do that, a parser has to read your file in four steps:
- Extract the text. A DOCX stores text in order. A PDF only stores pieces of text at positions on the page, so the parser has to rebuild the lines.
- Split it into sections by headings such as Experience, Education and Skills.
- Pull out fields: name, email, phone, job titles, employers, dates, skills.
- Index it so a recruiter's keyword search can find you.
If step 1 scrambles the text, every later step inherits the mistake. A parsing problem rarely rejects you outright, but it can leave your skills or contact details missing from your profile, where a recruiter's search can't find them.
Common parsing failures (and what each looks like)
- Two columns. Many parsers read straight across the page, so a sidebar gets glued onto your main column line by line:
EXPERIENCE SKILLS Data Analyst, Northwind SQL, Tableau Jan 2022 – Present Python, Excel
Your job title now ends in "SQL, Tableau", and the Skills heading is lost. Use one column. - Tables. The text comes out cell by cell, so a job title can end up far from its date or company. Move table content into normal lines.
- Text boxes. In Word, a text box sits outside the main text. Some parsers skip it; others read it in a different place. Keep everything in normal paragraphs.
- Headers and footers. Some systems ignore them. If your email and phone are only in the page header, your profile can end up without contact details. Put them in the first lines of the body.
- Icons, symbols and skill bars. Phone and mail icons, star ratings and progress bars are dropped or turned into junk characters. Write "Phone:" or simply the number.
- Fonts with no Unicode mapping (garbled letters). Some PDFs, often from design tools or unusual fonts, draw letters that have no text code behind them. They look fine on screen, but a parser gets gaps or random characters, so "Python" can become "Pyt�on". Save the PDF again from Word or Google Docs with standard fonts, or send the DOCX.
- Overprinted text. Some templates fake bold by drawing the same word twice in the same spot. A simple parser may read "EExxppeerriieennccee". Use real bold instead.
- Images and scanned pages. Text inside a picture or a scanned page has no text layer, so the parser gets almost nothing.
- Creative headings. "Where I've been" is harder to map than "Experience".
How to run a parsing test on your resume
- Upload the exact file you send to employers (PDF or DOCX) in the tool above. Pasting text skips the parsing step, so it can't show layout problems.
- Open What the ATS reads. Read it top to bottom as if you were the parser.
- Check the essentials: your name, email and phone appear near the top; each job title sits next to its employer and dates; Experience, Education and Skills show up in the recognized sections.
- Look at the fixes list. Each problem shows what it costs and how many points you get back by fixing it.
- Fix the file, upload it again and compare. It's free to repeat, so test every version you send.
Want the full picture, with bullet feedback and your keyword gap for a job? Use the ATS resume checker.
What the parsed-text view shows (and what it can't)
- Plain text in reading order. PDFs are read row by row, left to right, the way a simple parser does. Merged lines are the sign of a column problem.
- Recognized sections as chips, for example Contact, Experience, Education, Skills.
- Characters that can't be read are shown as "�", so you can see which words a parser would garble.
- Layout warnings: two columns (PDF), tables and contact details only in the Word header or footer (DOCX), images, icons, tab-aligned text and scanned pages.
What it can't show: the exact rules of a specific employer's system. Vendors and employer settings differ and aren't public. If your file reads cleanly here, a simple parser can read it; a stricter one may still differ. Every rule we use is listed on how the checks work.
For a quick look at the text without the explanation, use the resume scanner. To see how these problems add up to a number, see the ATS score checker.
Looking for a resume parser for recruiting?
This page is for job seekers testing their own resume. ATSGlass does not offer a resume parsing API or software for recruiters. If you hire, ask your ATS vendor how its parser handles columns, tables and headers, and test it with a few real files.
FAQ
What is an ATS parser?
It is the part of an applicant tracking system that turns your resume file into plain text and fields such as name, email, job titles, dates and skills, so recruiters can search and filter applications.
How do I test how an ATS parses my resume?
Upload the PDF or DOCX you send to employers above and open the parsed-text view. Check that your contact details are at the top, each job title sits next to its dates, and your main sections are recognized.
Why does my resume text look scrambled or have strange characters?
Scrambled lines usually mean two columns or a table read straight across the page. Strange characters, shown here as "�", mean the PDF uses a font without a Unicode mapping. Use one column and save the PDF again from Word or Google Docs with standard fonts.
Can an ATS read PDF resumes?
Usually, if the PDF has real text and a simple one-column layout. A scanned or image-only PDF has no text layer, so a parser gets almost nothing. If in doubt, test both your PDF and your DOCX here.
Is this the same parser employers use?
No. Each ATS vendor has its own parser and settings, and they aren't public. This test reads your file the way a simple parser does and flags the layout problems that commonly break parsing.
Is my resume uploaded or stored?
No. Your file is read in your browser and this prototype runs every check on your device. Your resume is not uploaded, stored or used to train models.
Does it cost anything?
Checks, your score, the top 5 fixes and the parsed-text view cost nothing and need no account. Pro features are marked with a lock.
Related tools
Last updated 2026-11-03. How we check.