Default to a single-column resume for any application a machine reads first. In UnchartedCareer's ATS Autopsy, a test of 48 synthetic resumes read by three local open-source text extractors (pdfplumber, pdfminer.six, python-docx) in June 2026, single-column layouts lost 0% of scored fields, the clean baseline every other layout is measured against. Two columns did nearly as well, losing a single field out of 72. The trouble starts when you add more columns. A three-column layout dropped the most-recent job 33.3% of the time. Read those as a floor on what can go wrong in the reading step, not a verdict on any named ATS.
By UnchartedCareer
Last updated: July 2026
Both hold up well, and single-column holds up perfectly. In the June 2026 test, single-column lost 0 of 72 field reads. Two-column lost 1 of 72, a rate of 1.4%, and the one miss was an education entry. Three-column is where the reading step buckled, losing 7 of 48 field reads, 14.6%. The pattern is steady. The more columns you stack side by side, the more chances the reader has to lose the thread. All of these are a floor on what the reading step can drop, not a verdict on any named ATS.
The verdict for a machine-read application is single-column. It posted the only clean zero, and it removes a risk you otherwise cannot see coming.
A PDF does not store columns. It stores characters at positions on a page, and the extractor decides how to string them into lines. When two blocks of text sit side by side, a reader sweeping left to right across the full width can cross the gap and interleave them, so a word from the right column lands in the middle of a sentence from the left.
The June 2026 test caught this directly. It scrambled the reading order in 2 of 6 three-column reads and in 2 of 9 two-column reads. A job counted as read only when the employer and the title landed within 60 characters of each other, so a scramble that pushed Senior Nurse away from Mercy General counted as a loss even when both phrases survived somewhere on the page. That adjacency rule is the same thing that trips a downstream field-mapper. Again, a floor on the reading step, not a named product.
The most-recent job, which is the worst field to lose. In the June 2026 three-column reads, the most-recent job (work_1) failed 33.3% of the time, 2 of 6, and the education entry failed 50%, 3 of 6. A recruiter scans for your current title first, and in a three-column grid it is among the first things to scramble, because the columns interleave right where that block sits. Treat both figures as a floor on what the reading step can drop, not a verdict on any named ATS.
No, and the data says so plainly. In this test, two columns lost a single field out of 72, and that one miss came from a PDF read, since the DOCX two-column files all came through clean. A well-built two-column resume can read fine. Two columns is not the thing that breaks.
But single-column posted a clean zero, and you do not control which extractor opens your file or how a template placed its sidebar. When the cost of one scramble is your current job going unread, one column is the safe default. A good two-column resume can survive. A single-column resume removes the question.
You can rebuild the layout yourself in a few minutes, and reading the result teaches you what a clean column order looks like.
The copy-paste test shows you reading order and whether the text is selectable. It will not score every field for you, and it will not tell you which sections a downstream mapper failed to label after the text came out. You are reading with your eyes, which catches a bad scramble and misses the quieter field-level misses.
Do the copy-paste test yourself first, because watching your own resume scramble teaches you what a clean column order should read like. When you want the faster read that goes field by field and flags which ones dropped or landed out of order, run your file through the free ATS resume scan at /free-ats-resume-scan. Same check as this page, in about two minutes.
Practice the hard questions now, so the real ones feel easy.
Get me readyNo credible study backs the claim that 75% of resumes are auto-rejected. The trail ends at Preptel, a vendor that folded in 2013. Here is what an ATS really does.
A clean single-column PDF reads as well as Word. In our June 2026 test, layout, not file type, is what broke the machine read. Here is how to check your resume.
In one classic study, graduating professional-school students who negotiated gained 7.4% on average over their first offer, and Pew found about two in three U.S. workers who asked got more than the opening number. Here is how to ask once, well.