Correction, 17 October 2026. Earlier versions of this article called these "tested" patterns from "running thousands of real resumes through a Workday-style parser", and gave a share of the Fortune 500 using Workday, a "75% auto-filtered" figure, "90%" and "95%+" rates, an "average broken resume fails 3 to 4 patterns" figure, and descriptions of how Workday's date parser, keyword matcher, ranking and volume detection work. We have no records of that testing, the figures had no source (the "75% of resumes are rejected by ATS" figure is a widely repeated myth), and we do not know how Workday works internally, so we have removed them. This version explains each pattern by how text extraction works and points to tests of our own parser that we can show.
If you've applied to a large employer through Workday and heard nothing back, the problem may not be your experience. When you upload a resume, Workday pre-fills your candidate profile from the text it extracts. If the file is built in a way that breaks extraction, that profile comes out incomplete or scrambled, and you may not notice. This post covers 12 structural patterns that commonly cause that, why each one breaks when text is extracted, and the fix.
Why this matters
Workday is widely used by large employers, so if you're applying to big companies, there's a good chance some of your applications go through it. The parsed text and the profile fields filled from it are what recruiters search and often read first.
We have not tested Workday itself, and Workday does not publish how its parser works. The patterns below come from how resume text extraction works in general, and from what we see when files go through our own parser. Where we have a real test to point to, we link it. For anonymous counts of how often issues like these turn up in files people scan with us, see the ATS data report (a self-selected sample of files people chose to check).
The 12 failure patterns
Pattern 1: Two-column layout
The setup: sidebar with contact info or skills on the left, main content (experience, education) on the right.
Why it breaks: text extraction reads the text in the order it sits in the file, not the order your eye follows. Many extractors read across the page line by line, which interleaves the two columns: sidebar line, main line, sidebar line, and so on. Others read the whole sidebar first. Either way, skills end up mixed into job descriptions and dates drift away from their roles. In our six-layout benchmark, the two-column version of the same resume was the only one to draw a critical reading-order flag in our parser.
The fix: single column from top to bottom.
Pattern 2: Name styled as decorative graphic
The setup: a large, design-heavy name at the top of the page, perhaps in a custom display font, an unusual colour, or set against a styled background.
Why it breaks: if the design tool flattened the name to an image, the text-extraction layer reads zero characters from it. If the name is in a font that doesn't embed cleanly, the parser may see garbled characters. Either way, the name field can come out blank or wrong.
The fix: name as plain text on line 1. Bold and a few points larger than body text is fine, but use a standard font like Calibri, Arial, Helvetica, or Garamond. More on this in why an ATS can't read your name.
Pattern 3: Contact info in a styled header or footer band
The setup: phone, email, LinkedIn URL placed inside the page header at the top of the page (or in a footer at the bottom).
Why it breaks: some parsers treat page headers and footers as boilerplate and skip them. Others repeat them: in our benchmark, an email placed in a repeated page header was extracted three times. Either way, your contact details are not where a parser expects them.
The fix: contact info goes in the body of the document, on the line immediately below your name. Plain text, separated by pipes or dots: "email@example.com | +1 (555) 555-5555 | San Francisco, CA | linkedin.com/in/yourhandle". The detail is in do ATS read headers and footers?
Pattern 4: Skills section as tables or visual grids
The setup: a Skills section organized as a 2×N or 3×N grid, often with icons or progress bars next to each skill.
Why it breaks: table cells can lose the spaces between them when the text is extracted. It depends on how the file was made. In our benchmark, a simple skills grid parsed cleanly. But in our ChatGPT and Claude resume test, the AI tools' own PDF exports of a resume with a skills table came out with glued cells, and in our Canva-style test a languages table became "EnglishNative". A glued string like "PythonSQLReact" won't match a search for "Python".
The fix: skills as a comma-separated line, not a visual grid. "Python, SQL, React, Docker, AWS, TypeScript, Postgres, GraphQL". Skill bars and star ratings carry no text at all, so drop them.
Pattern 5: Custom icon bullets
The setup: bullet points using custom symbols, chevrons, rounded squares, stars, checkmarks pulled from an icon font.
Why it breaks: icon-font characters can come out as stray symbols in the extracted text, or be dropped. In our Canva-style test, icons turned into stray characters stuck to the contact details. Bullets that don't come through as bullets make it harder for software to tell where one achievement ends and the next begins.
The fix: use a plain hyphen (-) or the standard bullet (•).
Pattern 6: Ambiguous date formats
The setup: dates written as "2020-22", "Jan'20 – Mar'22", or numerically like "03/2020 – 08/2022".
Why it breaks: software has to turn your dates into start and end points to build a work history. Numeric dates are ambiguous between US and international formats. Two-digit years and apostrophes are easy to misread. Year-only ranges ("2020 – 2023") lose the months, so a role can look shorter or longer than it was. If dates are misread, any experience total worked out from them will be wrong too.
The fix: month and year: "March 2020 – August 2022" or "Mar 2020 – Aug 2022". Unambiguous and easy to parse. More in how to format dates on a resume.
Pattern 7: Non-standard section headers
The setup: creative headers like "What I've Built", "My Story", "Where I've Been", "Things I Care About" instead of "Experience", "Summary", "About".
Why it breaks: resume parsers commonly use heading words to work out which section is which. Familiar headings such as "Experience", "Work Experience", "Education", "Skills" and "Summary" are the easiest to recognize. A creative heading can leave the section beneath it misclassified. In our benchmark, swapping in creative headings tripped our missing-standard-headings check.
The fix: use the standard words. "Experience" not "What I've Built". "Education" not "My Academic Story". "Skills" not "Things I'm Good At". Your creative voice can live in the bullet content; the section headers should be boring.
Pattern 8: Text inside text boxes or shapes
The setup: a callout box around a key achievement, a shaped container for an "Objective" statement, or a coloured card around a Certifications block.
Why it breaks: text in a Word text box or shape is stored separately from the document's main text flow. Depending on the parser, it can come out in the wrong place or be missed.
The fix: all content in the main text flow. No text boxes, no callouts, no decorative containers.
Pattern 9: Smart quotes and em-dashes (low risk)
The setup: typographic punctuation from Word's auto-correct or from copy-pasting: curly quotes, em-dashes (—), en-dashes (–), and curly apostrophes.
Why it's on the list: this one is often overstated. In our benchmark, em-dashes and curly quotes parsed cleanly, and our scanner flags them as information only, not as a problem. Character mix-ups are possible when text passes between systems with different encodings, so plain punctuation is the cautious choice, but this is the last thing to fix, not the first.
The fix: if you want to be safe, use straight quotes and regular hyphens. Spend your effort on patterns 1 to 8 first.
Pattern 10: Image-based PDFs
The setup: a PDF that's essentially a photograph of a resume, either scanned from a printed copy or exported with settings that flatten everything to an image.
Why it breaks: there is no text layer to read. Unless a system runs OCR on the image, nothing is extracted and the profile stays empty.
The fix: always export as text-based PDF from Word (File → Save As → PDF) or Google Docs (File → Download → PDF Document). To verify: open the PDF in a browser and try to highlight a line of text. If a text cursor appears, it's text-based. If clicking selects the whole area like an image, it's image-based and needs to be re-exported.
Pattern 11: Skills only inside experience bullets, with no Skills section
The setup: "Built an ML pipeline using Python, TensorFlow, and AWS SageMaker" mentioned inside a bullet, with no separate Skills section listing those technologies.
Why it matters: this isn't a parsing failure; the words still extract. But a dedicated Skills section gives parsers an obvious place to fill a skills field from, and gives a recruiter scanning or searching your profile a quick list to check against the posting. Skills mentioned only in passing are easier to miss.
The fix: include a dedicated Skills section listing your technologies, tools, and methods, and keep using them in context in your bullets too.
Pattern 12: Resume too long for the level
The setup: a 3-page resume from a candidate with 4 years of experience. Or a 4-page resume from anyone.
Why it matters: this isn't a parsing failure either; long resumes extract fine. It's a reader problem. Recruiters skim, and a long resume for an early-career role buries the strongest points and suggests you haven't prioritized.
The fix: as a rule of thumb, 1 page early in your career, 2 pages for most experienced professionals, more only at senior executive level or where the employer or country expects it. Check yours with our free length checker if you're not sure.
Problems tend to come in clusters
Patterns rarely arrive alone, because they come from the same design choices. A designed template with a sidebar often brings a table, icons and contact details in a styled band at the same time. When we ran a Canva-style sidebar resume through our parser, three problems showed up in one file: the sidebar was read before the work history, icons became stray characters, and a table glued into single words. That file still scored 80/100 in our scanner, because every word survived; what it lost was order and clean fields.
This is why patching one issue at a time is slow. Changing the template usually fixes several patterns in one go.
How to fix most of them at once
The fastest way to clear the structural patterns: rebuild in Word or Google Docs with these defaults:
- Single column, top-to-bottom flow
- Calibri or Arial 11pt body, 14pt name
- Standard section headers: Summary, Experience, Education, Skills
- Plain bullets (- or •)
- Month and year dates: "March 2020 – August 2022"
- Comma-separated skills, not a visual grid
- Contact details in the body, not the page header
- Export from Word or Google Docs as PDF, never print-as-image
None of this guarantees how any one employer's system will handle your file, and it can't make up for a weak match to the role. It removes the structural reasons a file comes out wrong, and those rules are the same whichever ATS the employer uses.
Check your own resume against these patterns
You don't need to guess which patterns your resume has. ATS Verification runs your file through our own parser (not Workday's) and shows what it extracts. It flags the structural ones from this list, including two-column scrambles, glued table text, icon bullets, contact details in headers, missing standard headings, ambiguous dates and a name it can't find, with the fix for each. Free, in seconds, no signup. How the checks work is on the methodology page.
For more on Workday specifically:
- Will Workday Read My PDF Resume? What Breaks and Why
- Greenhouse vs Workday: what differs and one format for both
- Are Canva resume templates ATS-friendly?
- Workday ATS guide
Quick action steps
- Scan your current resume free at atsverification.com and see which structural issues show up
- If there are one or two, fix them manually in Word
- If there are several, the easier path is a new single-column file, by hand or with our $5 ATS-safe Rebuild, which keeps your content and produces a clean .docx
- Re-scan the new version and read the extracted text before applying
- Use the clean version for online applications. Keep a designed version for people (networking, interviews, email)
One thing this list doesn't cover is how many applications you send, because that's about behaviour, not the file. If you're weighing auto-apply tools, see our Jobright AI review.
Want to see the data behind several of these patterns? Our 2026 ATS parsing benchmark ran one resume in six layouts through our parser and showed which layouts actually broke, and which did not.
