Skip to content

ATS & Parsing

Can an ATS Read a Two-Column Resume?

Can an ATS read a two-column resume? See what 2026 parser tests found, where columns break, and when to keep or ditch the two-column layout.

ResumeFlex guidesIdeas for your next move
PublishedUpdated 12 min read11 sections
Share article
Share on XShare on LinkedIn

On this page11 sections

Key takeaways

  • Textkernel's column-detection upgrade lifted correct column-separator decisions from 60% to 82% and well-rendered column CVs from 62% to 90%.
  • Enhancv's 357-run extraction test (updated October 2026) found two-column resumes lose almost no words, but section structure held only 35% of the time versus 60% for single-column on older extraction methods.
  • ATS Verification's scanner logged 8,220 real resumes in October 2026: 5.1% had two-column reading-order errors, and Canva exports failed at 13.9%.
  • Greenhouse's own support docs list columned layouts among the causes of failed or partial parses, alongside a 2.5 MB parse limit.
  • W3C's PDF3 technique shows screen readers hit the same wall: untagged two-column PDFs get read straight across both columns.

Yes, most modern applicant tracking systems can read a two-column resume; the words almost always come through. The catch is what happens after extraction: parsers can pull that text out of order, which scrambles the fields recruiters actually search. So the answer isn’t a flat yes or no. It’s “usually, at a risk you don’t fully control.”

Most advice on this debate sits at two useless extremes: “columns always break the ATS” or “modern AI reads anything.” The 2026 test data supports neither. Here’s what it actually shows, and how to decide for your own application.

What happens when an ATS reads a two-column resume?

An ATS converts your file into structured fields (name, employers, dates, skills) and a two-column resume forces it to guess the reading order while it does that. Guess right, and everything maps cleanly. Guess wrong, and your sidebar skills end up wedged inside a job description.

A resume parser doesn’t see your page; it reconstructs it. As Adobe’s ATS formatting research puts it, “a PDF records individual characters at specific positions on the page, not words and sentences.” The parser has to infer where sentences start, which lines belong together, and what order they were meant to be read in.

With one column, that inference is easy, text flows top to bottom. With two, the parser has to decide whether a visual gap is a column separator or just spacing, and different engines make that call differently. Jobscan’s September 2026 guide on tables and columns notes that “most ATS parsers read linearly (left to right),” and their demo parser grabbed the work-experience section of a designed resume while ignoring contact info, skills, and everything else.

Three things break most often:

  • Reading order mixes. The parser sweeps across the page row by row, interleaving sidebar text into the middle of your experience bullets. A deeper look at parsing errors shows how this turns clean bullets into fragments.
  • Sidebar content drops. Skills or certifications parked in a narrow right column sometimes vanish from the parsed profile entirely, because text boxes float outside the document’s text flow.
  • Headers and footers get skipped. Adobe’s research is blunt: “ATS does not read headers and footers,” which is why contact details parked up top produce candidate records with no phone number.

What do 2026 parser tests actually show?

The data says two things at once: two-column resumes rarely lose words anymore, but they lose structure far more often than single-column ones. That distinction is the whole story.

Enhancv ran the most rigorous public test available, published in August 2026 and updated that October. They pushed one resume through all 17 of their templates, three resume lengths, and seven PDF extraction engines, 357 total runs. The results:

  • Text survives. “Total word recovery barely moved between one column and two, never more than half a percentage point apart,” and contact details came through correctly in all 357 runs.
  • Structure suffers, on older methods. Sections stayed intact only 35% of the time with two columns, down from 60% with one, under geometry-based extraction. Newer reading-order methods shrank that gap to about a point.
  • AI parsing narrows it further, with one exception. Once an AI model reads the extracted text, the structural loss shrinks to roughly three points, but employer names still dropped from 100% to 83% correct, which matters because recruiters search by company name.

ATS Verification approached it from the other direction, scanning real-world files rather than lab templates. Their controlled test of six layouts (updated October 24, 2026) scored the single-column control a perfect 100/100, while the two-column version (the only layout of the six to earn a critical flag) fell to 85/100. Their production scanner data is more telling:

Across 8,220 real resumes scanned in October 2026, 5.1% had two columns read out of order, and Canva files were the worst offenders at 13.9%.

Meanwhile, the parsing vendors themselves have been fixing this. Textkernel, which powers parsing for platforms like Bullhorn, estimates that at least 15% of CV documents use a column layout and rebuilt its engine around that reality. On a test set of column-separator gaps, correct decisions jumped from 60% to 82% (more than half the errors gone) and in a blind evaluation of roughly 700 CVs, well-rendered column CVs rose from 62% to 90%.

That last number cuts both ways. Ninety percent is a genuine improvement. It also means roughly one in ten column resumes still renders badly even inside one of the best parsers on the market.

Where do two-column layouts still fail?

Two-column resumes break most often on older systems, image-based files, company-name searches, and heavily designed exports. If your application hits any of those paths, the modern-parser statistics stop applying to you.

  • Legacy ATS configurations. Older deployments of systems like Taleo run extraction that predates reading-order models, and plenty of large enterprises still run them. A parser that reads strictly left to right across a two-column page produces exactly the word salad the older guides warn about.
  • OCR and image paths. When a file has no text layer, the parser either fails outright or OCRs it, and OCR errors compound layout confusion. Greenhouse’s own support docs list resumes “uploaded as an image, rather than a document” among its parse-failure causes.
  • Company-name searches. Even with AI parsing, Enhancv measured employer-name accuracy falling from 100% to 83% on two-column layouts. If a recruiter filters by “did this person work at [company],” a scrambled employer field takes you out of the running silently.
  • Design-tool exports. Canva files failed at 13.9% in ATS Verification’s October 2026 scan data (nearly triple the overall rate) because design tools often produce layered text boxes rather than a clean text flow.

What do ATS vendors themselves say about columns?

Vendors treat columns as a known, documented risk rather than a guaranteed failure. That’s a meaningful shift from the “columns always break” era, but it’s not a clean bill of health.

Greenhouse’s troubleshooting documentation explicitly lists “columned layouts” among the formatting issues that cause unsuccessful or partial parses, alongside tables, headers, footers, text boxes, and graphics, and notes that resumes larger than 2.5 MB won’t parse at all. Partial parses, it says, must be “manually corrected and verified,” usually by a recruiter retyping your details.

Workday’s documented behavior is similar. A June 2026 guide to Workday’s parser (cross-referenced against its Recruiting documentation) explains that “Workday parses the document in reading order, so one column maps cleanly to the candidate profile,” while a two-column file means “the parser interleaves the sidebar into the middle of your experience text, so good bullets read back as fragments.” Workday also strips headers and footers on extraction.

Adobe’s research page splits the difference nicely: “Columns don’t get you rejected, but they raise the odds that specific fields will be imported as a jumbled mess.” That’s the most accurate one-line summary of the 2026 evidence.

Single-column vs. two-column resume: the tradeoffs

Single-column wins on risk, two-column wins on visual appeal, and the right answer depends on who reads the file first. Here’s the side-by-side:

Factor Single-column resume Two-column resume
Parsing risk Lowest (text flows top to bottom in one pass Higher) correct extraction depends on the parser and how the file was built
What typically breaks Almost nothing structural Reading order mixes, dropped sidebars, detached job dates
Vendor documentation No special warnings Greenhouse lists columns among parse-failure causes
Accessibility Screen readers follow it naturally Often read across both columns unless properly tagged (W3C PDF3)
Recruiter readability Familiar and fast to skim Can look sharper; opinions split on preference
Effort required Minimal Test before submitting, keep a single-column backup
Best used for Online applications through any ATS Human-first submission: email, LinkedIn, networking

If that table reads as “single column for portals, two columns for humans,” that’s because it is. The context piece in what an applicant tracking system actually reads explains why the portal path is so unforgiving.

Do two-column resumes create accessibility problems too?

Yes, and this is the part almost every resume article skips. The same reading-order ambiguity that trips parsers trips screen readers, and the W3C says so in exactly those terms.

WCAG technique PDF3 warns that in an untagged two-column PDF, “a screen reader may read the document from top to bottom, across both columns, interpreting them as one column.” The underlying success criterion, 1.3.2 Meaningful Sequence, requires that a correct reading order “can be programmatically determined”, column by column, not row by row across the page.

That matters for two reasons. First, blind and low-vision recruiters and colleagues may read your resume with assistive tech, and a two-column layout is hardest for exactly the tools they rely on. Second, it reframes the ATS question: a parser and a screen reader face the same structural problem, so a layout that reads cleanly to one tends to read cleanly to the other. A single-column resume satisfies that requirement for free, with no tagging work.

When is a two-column resume the right choice?

A two-column resume is the right choice when a human is the only reader. No parser sits between you and the recipient, so the extraction risk drops to zero while the design advantage stays.

That describes more situations than you might think:

  • Email applications to a hiring manager or founder who asked you to write directly.
  • LinkedIn Easy Apply and direct messages, where the file often lands in a human’s inbox or a simple attachment viewer.
  • Networking handoffs: conferences, coffee chats, portfolio PDFs where the design is part of the point.
  • Creative and design roles, where recruiters expect visual craft and a plain document would undersell you.

The behavior data backs this up: two out of three resumes downloaded from Enhancv use a multi-column layout, which tells you plenty of job seekers are choosing these designs deliberately, not naively.

How do you reduce the risk if you keep two columns?

If you’re set on the two-column design, you can shrink the parsing risk substantially with about fifteen minutes of work. None of these steps guarantees a perfect parse on every system, but each one removes a known failure mode.

  1. Run the plain-text test. Open your PDF, select all, and paste into Notepad. Jobscan recommends the same check: “If your columns merge into one scrambled line, or sections jump out of order,” the ATS will misread it too.
  2. Keep your name and contact info full-width and in the body. Not in a header, footer, or text box; Greenhouse flags all three as parse-breakers, and a candidate record with no phone number is the worst-case outcome.
  3. Build columns with real text flow, not text boxes or tables. Text boxes float outside the reading order, and tables get read cell by cell until dates detach from titles.
  4. Export as a text-based PDF or DOCX. The file-type choice changes what the parser sees; the tradeoffs are covered in PDF vs. Word for resumes. Never submit an image-based export.
  5. Keep the design modest. A narrow sidebar with skills is a different risk class than a graphic-heavy template, remember that Canva exports failed at 13.9% in ATS Verification’s scan data.
  6. Use standard section headings: “Work Experience,” “Education,” “Skills”, so the parser can anchor sections even if the order wobbles.
  7. Review the parsed profile before you submit. Workday, Greenhouse, and similar portals show you the extracted fields. If dates landed in the wrong rows, fix them by hand before clicking submit.
  8. Keep a single-column version ready. If you can’t verify the parse, or the role goes through a big-company portal, swap files. It costs you nothing.

The verdict

The internet’s flat “never use two columns” advice is outdated, but it’s not wrong enough to retire. The parsers got better, and the test data proves it, yet every serious dataset from 2026 still shows columns as the highest-risk structural choice you can make, and the failure is silent: no rejection email, just fields quietly misfiled. That asymmetry settles it. A two-column resume saves you maybe an hour of design work and risks the one search (by company name, by keyword) that gets you found. Use the good-looking version where a human reads first. Use the boring version where a machine does. That’s not cowardice; it’s matching your format to your reader.

FAQ

Do all ATS platforms read two-column resumes? No, capability varies by platform and by which parsing engine it licenses. Modern engines handle columns far better than the rule-based parsers of a few years ago, but vendors like Greenhouse still document columns as a parse risk, and legacy configurations lag well behind.

Will a two-column resume get automatically rejected? Rarely. ATS Verification’s 2026 testing found two-column resumes are seldom auto-rejected but often “quietly downranked” when fields misparse. The realistic damage is scrambled or missing data, not a rejection notice.

Is any two-column resume format ATS-safe? A modest, text-based two-column layout with a full-width header and no text boxes can parse cleanly on modern systems, Enhancv’s testing showed word recovery near-identical to single-column. “Safe” is doing heavy lifting there, though; it’s safe until it meets an older parser or an OCR path.

How can I test my resume’s parseability myself? Paste your entire resume into a plain-text editor like Notepad and read it top to bottom. If the text reads in the right order with sections intact, a parser will likely reconstruct it correctly; if it’s jumbled, you’ve reproduced the failure at home.

Are Canva resume templates bad for ATS parsing? They’re the worst-performing common export in the 2026 scan data; 13.9% of Canva files had columns read out of order versus 5.1% overall, per ATS Verification. The problem isn’t Canva’s templates per se; it’s that design tools frequently layer text in boxes instead of a clean flow.

Do two-column resumes hurt screen-reader accessibility? They can, badly, if the PDF isn’t tagged. W3C’s PDF3 technique documents screen readers reading “from top to bottom, across both columns, interpreting them as one column” in untagged files, the same reading-order failure parsers hit.

Should I use one column or two for online applications? Use one column for anything that goes through an application portal, and reserve two columns for email, LinkedIn, and in-person submissions. The single-column format eliminates the reading-order risk entirely, and no 2026 test data shows a parsing penalty for it.

What file type works best for a two-column resume? A text-based PDF or a DOCX, with the PDF verified for selectable text. What matters most is that the file has a real text layer in a logical order; an image-based PDF is unreadable to parsers and screen readers alike.

Sources

Written by ResumeFlex. Questions or corrections? Contact ResumeFlex. Read our source and correction standards.

Filed under

Back to guidesBack to top