ATS Resume Format: What Software Actually Reads

Why format matters more than design

A resume has to work twice. First a parser reads it and turns it into database fields; then a human reads it and decides whether to call you. Most formatting advice optimises only for the second reader, which is how people end up with beautiful documents that arrive as scrambled text.

The good news is that the two readers want mostly the same thing: clear structure, obvious sections, and no visual cleverness that obscures the content. A resume formatted for reliable parsing is also, almost always, an easy resume to skim.

Layout: one column, top to bottom

This is the highest-impact decision on the page. Parsers read a document in its underlying content order, which is not always the order your eye follows on screen.

A two-column layout with skills in a sidebar looks organised to you. To a parser working through the content stream, it can interleave the sidebar with the main column, producing a job title followed by an unrelated skill followed by half a date range. The output is not obviously broken enough to be caught, just wrong enough to be useless.

Single column avoids the problem entirely. Your sections appear in the order you intended, every time, in every system.

The four elements that break parsing

Headers and footers

The single most damaging habit. Many parsers ignore the header and footer regions of a document, which means a resume with the name, phone number, and email placed in the header can be stored with no contact information at all. Put contact details in the first lines of the body.

Tables

Tables are read cell by cell, and the reading order depends on the parser. A two-column table pairing dates with job titles may come out as every date followed by every title, severing the connection between them. If you want aligned dates, use tabs or a right-aligned text position rather than a table.

Text boxes and shapes

Content inside a text box sits outside the main content flow and is commonly dropped. The same applies to anything placed in a shape or a floating frame. If it is not ordinary body text, treat it as at risk.

Images and icons

An image contains no text. A skills chart showing five-star ratings communicates nothing to a parser and, incidentally, communicates very little to a recruiter either — nobody knows what four stars in Excel means. Icons beside contact details are usually harmless, but never let an icon replace the word it represents.

Section headings

Parsers identify sections by matching against expected labels. Use the conventional ones:

• "Work Experience", "Professional Experience", or "Employment History"
• "Education"
• "Skills" or "Technical Skills"
• "Certifications"
• "Professional Summary"

Avoid inventive alternatives — "My Journey", "What I Bring", "Career Highlights Reel". They gain you nothing and risk an entire block of your resume being filed as uncategorised text.

Dates: the quiet failure

Date formatting causes more silent parsing errors than font choice ever will, and almost nobody checks it.

Write dates as "Mar 2022 – Aug 2024" or "March 2022 – August 2024". Include the month; a year alone makes a three-month gap and a twenty-three-month gap look identical.

Avoid purely numeric formats. "03/04/2022" is March 4th in the US and 3rd April almost everywhere else, and a parser has no way to know which you meant. For a current role, "Mar 2022 – Present" is understood universally.

Keep the format identical in every entry. Mixed conventions within one document are a common cause of one role parsing correctly and the next one not.

Fonts, sizes, and spacing

This gets far more attention than it deserves. Any widely available font is fine — Calibri, Arial, Helvetica, Georgia, Garamond, Times New Roman. The parser reads characters, not typefaces.

What matters is legibility for the human: 10 to 12 point body text, headings clearly larger than body text, and enough white space that the page does not read as a wall. Shrinking to 8 point to force everything onto one page is self-defeating — better to cut a weak bullet.

Use ordinary bullet characters. Decorative symbols, emoji, and unusual glyphs occasionally render as replacement characters in extracted text.

File type and length

Which file type

A text-based PDF is the sensible default: it preserves your layout and is read reliably by current systems. Send .docx when the posting asks for it specifically. Never send a scan, a photo, or a screenshot — verify by opening your file and trying to select a word.

How long

One page for students, recent graduates, and career changers with limited relevant history. Two pages once you have enough relevant experience to fill them without padding. Three pages is justified in a small number of fields — academic, medical, and senior technical roles with publication or project lists — and almost nowhere else.

A two-minute format check

Before sending, run your finished PDF through this:

1. Select all and copy. Open the PDF, select the entire document, and copy it.
2. Paste into a plain text editor. Notes, TextEdit in plain text mode, or any blank email body.
3. Read what came out. This is roughly what a parser sees. Are the sections in order? Is each job title still next to its own dates? Is your phone number present?
4. Fix whatever is scrambled. Anything out of order in the pasted text is a formatting problem in the original — almost always a column, a table, or a header.

This test takes two minutes and catches nearly every serious formatting failure. It is the single most useful thing in this guide.