If you have been sending applications into the void and blaming the AI screener, you are probably wrong about which machine killed you. A public ATS checker just published its Aug 30 to Oct 6, 2026 sample, and the majority of resumes never made it to the scoring step at all. They died earlier, at the PDF-to-text parser, in silence.
The headline number: 61% failed before any scoring happened
Between August 30 and October 6, 2026, 3,382 people uploaded a resume for the first time to Resume Optimizer Pro's free ATS checker. Before any keyword model ran, 61% of those uploads extracted either with merged lines from side-by-side columns (59%) or produced no readable text at all (1.7%). That is 2,065 resumes out of 3,382 that the downstream AI never got to judge on content, because the content was already scrambled or absent.
The mechanism is boring and mechanical. When an ATS converts your document to plain text, it reads top to bottom, left to right. A two-column layout gets serialized as row 1 column 1, row 1 column 2, row 2 column 1, row 2 column 2. The visual structure you see (read column one entirely, then column two) is lost. Your "Skills" sidebar now runs through the middle of your "Experience" bullets as glued text.
3,382 resumes analyzed by one public checker between Aug 30 and Oct 6, 2026.
The 1.7% that produced no readable text at all is a smaller but more embarrassing category: the file was effectively an image. Canva's "Download as PNG" and iPhone-scanned PDFs are the two main culprits. If you exported your resume as a picture, the parser has nothing to read, and no amount of keyword tuning will save you.
What "parsing failure" actually looks like (it is not auto-reject)
Parsing failure is not the same as rejection. 98% of Fortune 500 companies use an ATS, but 92% of them do not auto-reject on format: they rank and sort. Your name, dates, and skills usually still extract. What is lost is structure.
The parser can no longer tell that "Senior Analyst" belongs above three specific bullets, or that "SQL, Python, dbt" is a skills list rather than part of a sentence. Keyword scoring still fires, which is why candidates convince themselves the resume is "fine." But the recruiter view, the thing a human actually opens, shows garbled job titles and skills bleeding into company names. Two seconds of eye contact, next.
This is why the fix is undersold on both sides. The optimists say ATS parsing has been solved and modern Workday handles columns. The doomsayers say any decorative element is an auto-reject. Both are wrong. The truth is quieter: your resume gets through, but as a jumbled transcript the human reviewer silently discounts.
The controlled test: single column scored 100, two column scored 85
A controlled six-layout benchmark from ATS Verification found that the clean single-column control parsed at a perfect 100/100, every field extracted, zero issues. The two-column version was the only layout of the six to draw a critical flag (a reading-order scramble) and the only one to lose points, falling to 85/100. In a larger self-selected sample of 10,420 real documents, the single most common failure, a skills table flattened into glued text, appeared in 42.9% of them.
Here is the dataset condensed:
| Segment | Figure | What it tells you |
|---|---|---|
| First-time uploads failing at parser (Aug 30 to Oct 6, 2026) | 2,065 of 3,382 (61%) | Majority never reach scoring |
| Column-merge vs no-text-at-all ratio | ~35x (2,008 vs ~57) | Columns are the dominant failure, not PNGs |
| Skills table flattened into glued text (10,420-doc sample) | 42.9% | Most common single defect |
| Two-column layouts, estimated parsing failure rate | 31.2% | Average score impact: -14 points |
| Design/Creative resumes flagging a format issue | 71% | vs 48% for Software Engineering |
| Formatting issues as share of all ATS parsing failures | ~23% | Cheapest category to fix |
Vendor-reported numbers from one parsing service push the gap wider: two-column Canva resumes failed Workday and Greenhouse text extraction in 78.4% of submissions across their analyzed set, and candidates who switched to single-column reported a 12.7% interview callback rate versus 2.1% for the unparsed two-column crowd. Treat the 6x gap as directional rather than peer-reviewed, but the direction is the same across every data source.
Designers get penalized twice, engineers barely at all
Designers are the group most likely to use a two-column template and also the group measured at the highest format-flag rate, which stacks the deck against them twice. Software engineers mostly dodge it because the default engineering resume (Jake's Resume on Overleaf) is already single column.
In Refolk's index of professional profiles, there are roughly 353,968 current Software Engineers in the US and roughly 114,810 Product and Graphic Designers. Designers are about one third the size of the SWE pool, but they carry a 71% format-flag rate against engineers' 48%. The parsing penalty is concentrated exactly where portfolios are supposed to be doing the talking.
The audience in the middle (roughly 153,538 US Marketing Managers plus Data Analysts in the Refolk index) is where "designed" resumes are common but not dominant, and where switching a sidebar template to a single-column one tends to move interview rates the most per hour of work.
Your resume is not blank. It is just quietly jumbled, and the recruiter is reading the jumble.
Which ATS you are applying through actually matters
The right fix depends on the parser on the other end, not a blanket rule. Modern parsers (Greenhouse, Lever, Ashby, newer Workday) handle a two-column layout rendered via Word's native "Columns" feature. Legacy systems (Taleo, older iCIMS) still butcher it.
A decision rule by employer type:
- Early and mid-stage startups on Greenhouse, Lever, or Ashby: two columns usually survive if built with native Word Columns, not text boxes. Still not optimal. Still costs you points on the human read.
- Large enterprises and government on Taleo or older iCIMS: single column, no exceptions. The parser is from a decade ago and it will mash your sidebar into your bullets.
- Workday at a Fortune 500: single column. Workday itself has improved, but the configuration a huge employer runs is often five years behind the vendor's current build.
- Anything you cannot identify: single column. The default assumption should be the oldest parser in the chain.
The companies hiring at scale in these roles in the Refolk index (Google, Microsoft, LinkedIn, Figma, Ashby, Glean, Palantir, Discord, Runway) skew modern on ATS, but their enterprise peers do not. If you apply broadly, the only portable format is the one that works on the worst parser you will touch.
The 10-minute self-diagnostic
Open your current resume PDF and run this test before you send another application. It takes less time than tweaking a bullet.
- Copy-paste it into a plain text file. Select all, copy, paste into Notepad or TextEdit in plain text mode. This is roughly what the parser sees.
- Read it top to bottom. If your skills section is now interleaved with job titles, or a sidebar word is wedged inside a bullet, you have the column-merge problem.
- Check for missing text. If nothing pastes, or only a few words paste, your file is an image. Rebuild it as a text PDF.
- Look at your dates. If "2023 to Present" shows up next to the wrong employer, your reading order is scrambled even if the words are all there.
- Confirm your headings. "Experience" should be its own line, not glued to the last word of your summary.
- Open it on your phone. Weird line breaks in mobile preview often mirror what the parser does.
If any of those tests fail, do not touch keywords yet. Fix the format first, because the format is why your keywords are not landing.
The format fix, matched to your role
The right template depends on how visual your field expects a resume to be. Three defaults that actually work in 2026:
- Engineers, analysts, operators: Jake's Resume on Overleaf. Single column, standard section headings (Education, Experience, Projects, Technical Skills), no icons, no columns, no graphics. Compiles to a PDF that passes every major ATS scanner. Boring on purpose.
- Marketers, PMs, generalists: a single-column Word or Google Docs template with left-aligned section headings. Skip the Friggeri CV on Overleaf (it is explicitly tagged "High ATS Risk, two-column, graphic sidebar" in the same gallery that recommends Jake's).
- Designers: a single-column application resume, with a one-line portfolio URL at the top. Keep the designed version for the portfolio site and the in-person interview. The ATS is not the audience that evaluates visual craft.
Then, and only then, tailor the content. Formatting causes roughly 23% of parsing failures, and that one is solved in five minutes. The bigger problem sits above it: 99% of resumes, as submitted, have experience gaps against the specific job they target. That is the single biggest reason resumes fail screening, ahead of formatting, and no amount of switching to single-column PDF will fix bullets that do not mention what the posting asked for.
Rewriting bullets for a specific posting is the slow, repetitive part most candidates skip after they send five. It is also what Refolk does on every application: paste the posting, get your own resume back rewritten for it in a parser-safe single-column layout, with the cover letter drafted and a fit score that tells you whether the gap between your history and the job is small enough to be worth the apply.
The single most common parsing defect across modern ATS platforms.
What to do today
Fix the format first, because it is cheap and the failure is invisible. Then fix the content, because that is where interviews are actually won or lost. In order:
- Rebuild your resume as a single-column text PDF, no sidebars, no icons, no text boxes.
- Run the plain-text paste test above.
- Confirm it opens cleanly on a phone.
- For every application, rewrite the top third of the resume (summary, top-of-page skills, most recent role bullets) to mirror the exact phrasing of the posting.
- Only then look at the ATS score the employer's own tooling returns.
The 61% number is scary but useful. It tells you that most of your competition is losing before the race starts, and that the first fix is a template swap, not a career pivot.
FAQ
Will a two-column resume get auto-rejected?
Usually not. 92% of Fortune 500 ATS instances do not auto-reject on format, they rank and sort. Your resume still enters the pile. What you lose is structure in the recruiter view: titles glued to skills, bullets attributed to the wrong employer, section headings swallowed into paragraphs. The human read is where you lose the slot, not the automated one.
How do I tell which ATS a company uses?
Look at the URL on their careers page. "boards.greenhouse.io" is Greenhouse, "jobs.lever.co" is Lever, "jobs.ashbyhq.com" is Ashby, "myworkdayjobs.com" is Workday, "taleo.net" is Taleo, "icims.com" is iCIMS. Greenhouse, Lever, and Ashby are modern and forgiving. Taleo and older iCIMS are the ones where a two-column layout genuinely costs you.
My resume is pretty. Do I really have to give that up?
For the application version, yes. Keep the designed version for your portfolio site, LinkedIn, and the interview itself, where a human is actually looking at it. The resume you upload to a careers page is a data file, not a design artifact, and the parser is the first reader. A single-column text PDF plus a one-line portfolio URL at the top gets both audiences what they need.
If I fix the format and still get no responses, what next?
Format is roughly 23% of parsing failures. The bigger issue is content-to-posting fit: 99% of resumes have experience gaps against the specific job they target. Rewrite the summary, top skills, and most recent bullets to match the posting's exact language, and apply only where the overlap is real.