Single-column vs two-column resumes: which actually parses better

We ran 200 resumes through five major ATSes in both single-column and two-column layouts. The parse gap is bigger than the design community admits.

Two-column resumes look better on a screen. They fit more in less space, they let you put your skills sidebar somewhere visible, and they signal a small amount of design taste. The trade-off everybody acknowledges in the abstract is that they parse worse — but how much worse, exactly, was never clear to us. So we measured it.

The setup

We took 100 anonymized real resumes from our user base, all originally written in single-column layouts, and produced two-column variants by moving the skills, certifications, and contact info into a left sidebar. Every other piece of content was identical between the two versions. We then ran both versions through the parse APIs of five widely-used ATSes — three publicly accessible, two we have research access to — and scored each on three dimensions: name and contact correctly extracted, work history correctly segmented, skills correctly tagged.

We restricted the test to resumes between one and three pages, written in English, exported as PDFs from either Word, Google Docs, or our own builder. We deliberately did not test resumes built in tools that flatten output to images — those parse terribly regardless of column count.

The numbers

  • Name and contact extracted correctly: 99 percent single-column versus 84 percent two-column. The biggest failure was the candidate's email landing inside the sidebar, which two of the five parsers tagged as a reference or skipped entirely. A 15-point gap on a field this critical is alarming.
  • Work history segmentation: 96 percent single-column versus 71 percent two-column. The most common failure mode was right-column text from one role being concatenated with left-column text from a different role. A recruiter searching the resulting record for python at a senior level would see your python skill attached to your earliest, most junior job.
  • Skills tagging: 91 percent single-column versus 78 percent two-column. Slightly less catastrophic because skills are often rule-matched against a dictionary, so even a scrambled extraction sometimes recovers individual terms.
  • Overall clean-parse rate, defined as all three dimensions intact: 88 percent single-column versus 53 percent two-column.

A 35-point gap in clean-parse rate means roughly one in three two-column resumes gets to a recruiter in a degraded form — name missing, roles fused together, dates dropped — versus roughly one in ten for single-column. If your job application volume is in the dozens, this difference matters a great deal.

Why two-column layouts go wrong

Text extractors generally walk a PDF page in reading order, which they infer from glyph positions on the page. With a single column, reading order is unambiguous — every line is the next line. With two columns, the extractor has to decide: should the second visual line on the page be the second line of the left column, or the first line of the right column? Different libraries make different decisions. Some are correct most of the time. Others are correct rarely. The same PDF can parse perfectly in one ATS and disastrously in another.

Notably, this is not a problem PDF generators can fix from their side. The PDF format does not carry a this-is-two-column hint. The reading order is reconstructed by the consumer of the file, and the consumer in this case is software you do not control. You cannot fix the parser; you can only hand it a layout that does not provoke the bug.

The exception: table-based two-column

There is a narrow class of two-column resumes that parses well. If your two-column layout is built as a table — each row a self-contained record (one role per row, with dates in a narrow first column and content in a wide second column) — the extractor sees a series of unambiguous rows. Reading order is preserved row-by-row, and the parse rate climbs back to near single-column levels. Our ats-two-column template is structured this exact way for this exact reason.

This is different from a free-flowing two-column layout where prose wraps across columns. The free-flowing kind is the parse-killer. The table kind is fine. The visual difference is subtle; the parse difference is enormous.

When the gap does not matter

If you are applying to roles where humans look at every resume — small companies, agencies, executive search firms, design and creative roles where portfolios are the primary screen — the two-column penalty is academic. The human eye handles two columns fine, and the rest is downstream. In those contexts, choosing the design-forward template is the right call.

If you are applying to roles where the first reader is an ATS — large companies, FAANG-adjacent tech, anything posted to LinkedIn Easy Apply, anything filtered to a recruiter queue based on keyword search — the single-column version wins on volume statistics alone. Some of our most successful template lines (ats-classic, in particular) are deliberately the most boring ones, for exactly this reason.

The pragmatic compromise

Keep two versions of your resume. Use the well-designed two-column version for human-first contexts — referrals, networking, your personal site, attaching to a cold email to a hiring manager you know. Use the single-column version for any application that gets submitted through a portal. Our builder lets you switch templates without losing data, which makes maintaining both versions less annoying than keeping two separate documents.

If you want to A/B test this yourself: apply to two similar roles with the same resume in two different layouts and track which one gets a callback. The sample size is small but the result will be informative. Our experience over thousands of users is that the callback-rate gap on portal applications is real, large, and very nearly always in favor of the single-column version.

A note on aesthetic

Resumes should not be ugly. The single-column ats-classic template is not ugly — it is restrained, which is a different thing. Restraint reads as competence in most professional contexts. The two-column impulse often comes from feeling that a one-column document looks under-designed; we would gently push back on that read. A clean single-column document with deliberate typography looks more senior than a busy two-column one in most reader's eyes, including most recruiter's eyes.

The deeper principle: the resume is a working document, not a portfolio piece. Save the visual identity for the portfolio. Submit a parser-friendly resume and link the portfolio from the contact line.

Frequently asked questions

Are there any two-column templates that parse cleanly?
Yes — the trick is using table-based two-column layouts where each row is a self-contained record. Our ats-two-column template uses this pattern. Free-flowing two-column layouts where prose wraps between columns are the parse-killers.
Does this apply to LaTeX resumes too?
Yes. The parse problem is about visual layout, not the tool that produced the PDF. A two-column LaTeX resume has the same reading-order ambiguity as a two-column Word document.
What about resumes designed for a screen, not paper?
If the application asks for a PDF, you are in PDF rules. Some companies now accept a personal site or a structured profile URL; in those cases, design freely. But until the application is clearly screen-first, optimize for the parser.
Will the design industry call me boring for using a single-column resume?
Probably not. The design industry has its own portfolio-first conventions, and your portfolio is where visual identity lives. Recruiters and design directors usually want the resume itself to be readable, which means single-column for the parse and the portfolio link for the personality.
How long should I wait before retesting after switching layouts?
If you are tracking callbacks, give it two to four weeks across at least 10 to 15 applications before drawing conclusions. Application outcomes are noisy at small sample sizes.
Does the recruiter ever see my original PDF?
Sometimes. Many ATS interfaces show the recruiter the extracted structured record by default, with an option to open the original. If your structured record looks broken, most recruiters will not bother to open the original — they will move to the next candidate whose record looks coherent.