Most advice about ATS formatting is guesswork repeated until it sounded true. This page is the opposite: the eight rules below are about whether a parser can extract text from your file, which is a mechanical property you can test rather than a matter of opinion.
An ATS-friendly resume format is a single-column document saved as a PDF with a real text layer, using standard section headings, ordinary fonts, no text inside images or text boxes, and dates written the same way throughout — so that applicant tracking software can extract every word in the order a human would read them.
Each one is a thing software either can or cannot do with your file. They are ordered by how often they are the actual problem.
A parser reads the text layer inside the PDF. If your resume was exported as an image, or scanned, there is no text layer and the software extracts nothing at all — no name, no skills, no dates.
Open your PDF and press Ctrl+F, then search for your own surname. If nothing highlights, an ATS sees a blank page.
A parser reads a page as a single stream. In a two-column layout that stream interleaves — a line of your sidebar, a line of your experience, a line of your sidebar — and job titles end up glued to skills.
Sidebars, side-by-side skill panels, and any layout where two blocks sit at the same height on the page.
Software looks for a known set of words to decide where each section begins. "Work Experience" is recognised everywhere; "My Journey" and "Where I have made an impact" are recognised nowhere.
Creative headings do not just fail to help — they cause the section beneath them to be filed under whatever heading came before it.
Text placed in a graphic, a text box, or a shape often sits outside the document's reading order, so it is either extracted in the wrong place or dropped entirely.
Skill bars, infographic-style rating dots, and headers built as a single image are the usual culprits. Rating dots also convey nothing a parser can read: "4 out of 5 dots" is not a level.
An ATS computes your years of experience from the dates it finds. Mixing "Jan 2024", "01/2024" and "2024-01" in one document produces gaps and overlaps that were never in your career.
Pick one format and use it in every role, including the current one — and write the present role as "Present", not a blank.
Fonts that are not embedded in the file can be substituted at parse time, which mangles characters. Decorative and icon fonts extract as symbols or as nothing.
Icon fonts used for contact details are the common trap: a phone glyph extracts as an empty box, so the phone number loses its label.
Some parsers skip the page header and footer entirely, because those regions usually hold page numbers rather than content. Put your email and phone in the first lines of the document itself.
A resume whose only phone number lives in a page header can be parsed perfectly and still arrive with no way to contact you.
Table cells are extracted cell by cell, which is correct for data and wrong for prose. A layout table wrapping your whole resume shreds the reading order.
If you cannot select a full sentence across a row with your mouse, a parser cannot read it as a sentence either.
Two-column resumes are not banned and they are not ruined. They are measurably harder to parse, and the honest way to say it is with a number rather than a warning.
A single column parses cleanly in essentially every case. A two-column layout parses correctly most of the time and fails in a specific way when it fails: your sidebar content interleaves with your experience. If you are applying somewhere that matters and you cannot test the file first, use one column.
PDF, unless the application form explicitly asks for .doc or .docx — in which case, obviously, give it what it asks for.
A PDF with a real text layer is read reliably and looks identical everywhere. A .docx can reflow differently on the reader's machine, so what they see is not always what you sent. The old advice to avoid PDFs dates from parsers that genuinely could not read them, and it has been out of date for years.
The trap is not the extension — it is what is inside. A PDF exported as an image fails completely, while a plain .docx parses fine. That is why rule one comes first.
Two checks, in order. The first takes five seconds and catches the failure that matters most.
Resumes exported from this builder carry a real, invisible text layer, so a parser reads the whole document while a human sees the design. You do not have to take that on trust: export one and run it through the checker on this site.
A single-column PDF with a real text layer, standard section headings ("Work Experience", "Education", "Skills"), consistent dates, ordinary embedded fonts, and no text inside images or text boxes. That combination is readable by every major applicant tracking system, and each part of it is testable on your own file.
No — they parse them, and most of the time they parse them correctly. The risk is specific rather than fatal: in a two-column layout the extracted text can interleave, so sidebar content lands in the middle of a job description. Our own twelve templates measure 7-18% left-edge clustering when single-column and 22-45% when two-column, which is the same signal the checker uses on an uploaded file.
PDF, unless the form asks for Word. A PDF with a text layer parses reliably and renders identically for the reader. What actually breaks parsing is not the file extension but whether the file contains real text — an image-only PDF fails completely while a plain .docx is fine.
Any ordinary text font that is embedded in the file. The parser reads characters, not typefaces, so the choice matters far less than people are told. What does break is decorative and icon fonts, which extract as symbols or as nothing at all — an icon used to label your phone number can leave the number with no label.
Some do and some skip them, which is reason enough not to rely on them. Keep your name, email, phone and location in the first lines of the document body. A resume whose only phone number sits in a page header can be parsed perfectly and still leave the recruiter with no way to reach you.
Test it rather than trusting a checklist. Open the PDF and search for your own surname — if nothing highlights, there is no text layer and nothing else on this page matters yet. Then run the file through a parser that reports structure, so you can see the reading order the software gets rather than the layout you designed.
Every template here is labelled single-column or two-column before you pick it, and every export carries a real text layer. Building is free and there is no sign-up to start.
Build my resume