Two-column resumes: what an ATS actually sees
The best-looking resume templates are two-column. That is exactly the layout an applicant tracking system reads worst — and you can verify it yourself in about a minute, without taking anyone's word for it.
Updated
An ATS does not read your resume
The first misunderstanding is the word „read”. You look at a page and see a layout: sidebar here, heading there, a date alongside. Software sees no layout. It gets a file and tries to map fragments of it onto database fields — job title, company, dates, school, skills.
Anything that makes that mapping ambiguous costs you. And it is not about taste: the layout loses because it is ambiguous, not because it is ugly.
A PDF has no columns — it has coordinates
A PDF file does not record „this is the left column”. It records a list of text fragments, each with a position on the page. The document structure is simply not in there; it has to be reconstructed from the numbers alone.
The reconstruction works like this: fragments sitting at a similar height get joined into one line, and lines into paragraphs. With a single column that works fine. With two, the line at height 300px contains both a fragment from the sidebar and one from the main column, because they sit side by side.
The result looks roughly like this:
- „SKILLS Marketing Manager, ACME Ltd.”
- „Google Ads 01.2021 – 03.2023”
- „SQL Grew campaign revenue by 42%”
A skill name is now glued to a job title, a proficiency level to an employment date. The parser takes that as text and tries to extract a company name from it. It will not.
Word files have the same problem in a different shape
DOCX records structure explicitly — a paragraph is a paragraph, a list item carries a list marker. Columns in a resume are usually built with a table: one row, two cells.
That sometimes reads better, because a cell can be read whole and the columns unwrap themselves. But not every parser does this, and some systems take tables apart row by row — which brings back exactly the same scrambled text as the PDF. A layout table is a bet on which system you land in front of.
Why unscrambling the columns is not the fix
Tools that accept an uploaded resume can usually unwrap columns — read one side, then the other. Ours does too, because otherwise the analysis would run on scrambled text and the whole report would be worthless.
But that creates a trap worth knowing about: an unwrapped resume then looks single-column. Stopping there would score a two-column file better than it deserves, and tell you „layout is fine” about a document that genuinely falls apart.
What it actually costs
The format score is made of twelve checks with different weights. Layout is the third heaviest — only standard sections and complete contact details outrank it.
- Standard sections (experience, education, skills) — weight 0.20
- Contact details inside the document body — 0.16
- Single-column layout — 0.14
- Full employment dates, with months — 0.12
- No icons, emoji or decorative glyphs — 0.10
The remaining seven checks — length, bullets, walls of text, consent clause, sensitive data, unusual headings, leftover page numbering — share what is left.
Check your own file in a minute, no tool needed
Open your resume in any PDF viewer, select everything (Ctrl+A), copy it, and paste it into a plain text editor.
What you see is very close to what a parser gets. If it reads in order, you are fine. If company names interleave with skill names and dates sit in the middle of sentences, that is the problem.
The same test catches two other things along the way: a resume saved as an image (nothing selects — it is a scan or text converted to outlines, and no system will read it) and icons used instead of labels (phone and address turn into empty boxes or vanish).
What to do instead
- One column, top to bottom. Boring and effective. A recruiter reads top-down anyway.
- Skills as a section, not a sidebar. Same content, same place in the document, none of the parsing risk.
- Contact details in the body, not in the page header or footer — those are often skipped during extraction.
- Standard section names. „Work experience”, not „My journey”. Parsers match headings against a known list.
- Dates with month and year, in one format throughout.
If a two-column template matters to you, that is your call and it can be a reasonable one — for instance when applying directly to a small company with no system in between. It is just worth making that choice knowing the price, rather than because the template was first in the list.