Skip to content

ATS & Parsing

Resume Parsing Errors: Why Your Details Go Missing

Resume parsing errors can erase your contact info, dates, and work history before a recruiter ever looks. See what breaks, why it breaks, and how to fix it.

ResumeFlex guidesIdeas for your next move
PublishedUpdated 12 min read9 sections
Share article
Share on XShare on LinkedIn

On this page9 sections

Key takeaways

  • Greenhouse can't parse resumes larger than 2.5 MB, and its own support docs list headers, text boxes, columns, and graphics as causes of failed parses.
  • Textkernel's layout model raised column-separator detection from 60% to 82% and 'well-rendered' CVs from 62% to 90%, yet Adobe, citing Textkernel's own evaluation, still reports roughly 1 in 10 resumes renders poorly.
  • Jobscan's 2026 ATS Usage Report (August 20, 2026) detected an ATS at 97.4% of Fortune 500 companies (487 of 500) with Workday powering hiring at more than 40% of them.
  • Enhancv's 2026 testing (357 runs, 17 templates) found two-column resumes lose almost no text, but correct section order held only 35% of the time under the worst extraction method, and employer-name accuracy fell from 100% to 83%.
  • Jobscan's State of the Job Search 2025 survey found 99.7% of recruiters use keyword filters in their ATS, so unparsed details make you unsearchable, not just incomplete.

Your details go missing because an ATS doesn’t read your resume the way a person does. It runs resume parsing software that converts your file into database fields, and anything the parser can’t recognize never makes it into your candidate record. The usual culprits are layout choices: contact info parked in headers or text boxes, columned layouts, tables, graphics, and quirky date formats.

Every one of those is fixable in an afternoon. Fixing them matters more than most candidates realize, because parsed data (not the file you attached) is what recruiters search first.

What resume parsing actually does

Resume parsing is the process of software reading your resume file and converting it into structured fields (name, email, phone, employers, job titles, dates, education, and skills) inside a company’s recruiting database.

Workday describes it plainly on its own site: as applications arrive, a parsing engine reads document text, extracts details like education, skills, and work history, and organizes them into structured profiles (Workday’s ATS overview). Textkernel, one of the biggest dedicated parsing vendors, says it processes more than 2 billion resumes and job postings a year and is used by over 60% of the global HR tech industry (Textkernel). Odds are that software from one of these two companies touches your application before a human does.

The scale is hard to overstate. Jobscan’s 2026 ATS Usage Report (its sixth annual edition, published August 20, 2026) detected an applicant tracking system at 97.4% of Fortune 500 companies, 487 out of 500, with Workday alone powering hiring at more than 40% of them (Jobscan, August 20, 2026). The parsing step is where your application either becomes a clean, searchable record or a mess.

Here’s the part most job seekers miss. Parsing happens first, and everything after it (keyword filters, match scores, recruiter searches) runs on the parsed fields, not on your attached document. If a field comes out empty or garbled, every downstream step works with the empty or garbled version. For the fuller picture of what these systems capture, see what an applicant tracking system actually reads.

Where parsing fails, field by field

Parsing fails wherever your formatting hides or distorts plain text, and each resume field fails in its own characteristic way. CareerFlow’s help center sums up the causes into four buckets: fancy formatting, funky fonts, inconsistent formatting, and hidden text or special characters (CareerFlow). Here’s how those show up on the page.

Contact info. Greenhouse explicitly lists “name/contact info in headers, footers, or text boxes” among its causes of failed parses, and Adobe’s formatting guide is blunter still: ATS doesn’t read headers and footers at all (Greenhouse Support; Adobe). Park your phone number in a Word header and plenty of systems will happily build a candidate record with no phone number in it.

Dates. A parser uses your dates to calculate tenure, flag employment gaps, and rank how recent your experience is. Seasonal ranges like “Summer 2024,” quarters like “Q3 2022,” and mixing formats across entries all give it trouble. That’s why Novoresume’s October 2026 guide tells you to pick one format (Month Year or MM/YYYY) use “Present” for current roles, and keep the dates on the same line as the job title (Novoresume, October 26, 2026).

Job titles and employers. Greenhouse calls out abbreviated titles like “Sr. Account Exec” instead of “Senior Account Executive,” and company names missing legal identifiers such as “Inc.” or “LLC,” as things that trip up its parser. The employer field is fragile in another way too, more on that in the two-column section below.

Education and skills. Parsers file content under sections by matching the headings, so creative labels like “Where I’ve Worked” or “Toolbox” can leave content homeless, Novoresume keeps a list of the standard labels parsers actually recognize. Skills stored as star ratings, progress bars, or table cells tend to get dropped or garbled on the way in.

The whole file. A scanned or photographed resume has no text layer, so there’s nothing to parse at all. Adobe recommends the copy-paste test: if you can’t select the text in your PDF, the ATS can’t either (Adobe). And Greenhouse states outright that it can’t parse resumes larger than 2.5 MB, so a graphics-heavy file can upload fine and still parse into an empty record (Greenhouse Support).

Safe, risky, and broken: the formatting verdicts

The safe elements share one trait: they look identical as plain text to how they look on your screen. The breakers either reorder your text or store it somewhere outside the text layer entirely.

Element Verdict Why
Single-column layout Safe Text reads top to bottom in one unambiguous order
Standard headings (“Work Experience,” “Education,” “Skills”) Safe Parsers file content by matching these labels
Month Year dates, plain hyphen, “Present” Safe One consistent, machine-readable pattern
Arial, Calibri, or Times New Roman at 10pt+ Safe Standard character mapping the parser can trust
Two-column layout Risky Text usually survives, but the reading order can scramble
Layout tables Risky A PDF stores characters at positions, not rows and columns; the parser has to rebuild structure that was never saved (Adobe)
Icons, skill bars, photos, word art Breaks On Greenhouse’s published fail list as “graphics, photos and word art”
Contact info in header, footer, or text box Breaks Headers and footers often go unread; text boxes are floating layers
Letter-spaced or decorative fonts Breaks Letter spacing splits words into separate characters, per Greenhouse guidance cited by Adobe
Scanned or image-based file Breaks No text layer to extract
File over 2.5 MB (Greenhouse) Breaks Uploads fine, parses to nothing

Adobe’s evaluation did not identify length, color, bold, italics, or margins under an inch as parsing problems. That finding applies to its test; it does not guarantee that every parser will handle every layout. Start by checking whether your text and sections extract in the right order.

The two-column problem, honestly

Two columns rarely lose your text outright anymore, but they still scramble reading order, and scrambled order scrambles your work history. That’s the real risk, and the newest testing bears it out.

Textkernel, whose parsers sit behind many major ATS platforms, estimated 10–15% of CVs use column layouts. Its shift from hand-written rules to a trained model lifted column-separator detection from 60% to 82%, raised the share of “well-rendered” CVs from 62% to 90%, and improved contact-information fill rates by 4 to 10 percentage points across 12,000+ test resumes (Textkernel). In another test of more than 600 visual column gaps, the new model made the right call 91% of the time versus 82% for the old rules (Textkernel).

Even after parsers got dramatically better, roughly 1 in 10 resumes still renders poorly, per Textkernel’s own evaluation, cited in Adobe’s formatting research (Adobe).

There’s real research money behind this problem. In 2021, Microsoft Research built a 500,000-document dataset called LayoutReader to teach models reading order, and Adobe notes it measurably improved commercial OCR engines (Adobe). Still, the gains land unevenly across systems.

Enhancv ran the most thorough public test, updated October 23, 2026: one resume’s content pushed through 17 templates across 357 controlled runs and seven extraction tools. Total word loss between one- and two-column versions was never more than half a percentage point apart, and email, phone, location, and links came through in all 357 runs (Enhancv, updated October 23, 2026). But under the worst extraction method, correct section order held only 35% of the time with two columns, down from 60% with one, and employer-name accuracy fell from 100% to 83%. One template produced the only run out of 357 where the candidate’s name wasn’t recovered at all.

A two-column resume with selectable text may parse correctly, but check the extracted fields before submitting. If you cannot preview the result, a single-column layout reduces the chance of reading-order errors. See can an ATS read a two-column resume for examples.

What missing details cost you later

Missing details make you unsearchable. Recruiters filter and search parsed fields (not your attached file) so a field that never parsed is a field nobody can find you by.

The behavior is documented on the vendor side. Workday’s own materials describe turning applications into searchable profiles and recruiters searching those profiles with filters like years of experience and required certifications (Workday). Jobscan’s State of the Job Search 2025 report, based on a survey of more than 380 recruiters, found 99.7% of recruiters use keyword filters in their ATS and 76.4% start their candidate search with skills (Jobscan). Those filters see exactly what parsed. Nothing more.

Concretely, when your details don’t parse, here’s what a recruiter can’t do:

  • Run a skills search and land on your profile, even though you have the skill.
  • Verify your tenure or seniority from clean dates, so gap calculations and recency rankings go wrong.
  • Rediscover you months later for a different role, because the searchable record was never complete.

The stakes compound fast. Jobscan’s analysis of nearly a million job searches found resumes containing the job title drew 10.6 times more interview invitations (Jobscan), if your title parsed as garbage, that signal never existed. Filters can also be brutally literal: Harvard Business School and Accenture’s 2021 Hidden Workers report found 88% of employers admitting qualified candidates get vetted out by exact job-description criteria (94% for middle-skill roles), figures Adobe’s guide cites directly (Adobe).

A failed parse is worse than a weak resume. A weak resume still gets read; Adobe’s June 2024 survey put average resume review at 8.48 minutes, with 15% read thoroughly. A failed parse gets read by nobody, because the search that would surface you never returns your name in the first place.

The review screen is your safety net

After you upload, most systems show you the parsed fields before you submit; that screen is where you catch missing details, and it’s often your only chance. Treat it as part of the application, not a formality.

Like most systems, Workday won’t auto-fill fields it can’t reliably detect, so blank fields on the review screen are a signal the parser gave up on that item. The widely repeated best practice is to review and correct every populated field (especially dates and company names) before clicking submit. iCIMS behaves the same way: your file becomes a structured, searchable profile, and the standing advice is to check that your profile fields match your resume rather than trusting autofill (Enhancv on iCIMS).

When a parse fails completely in Greenhouse, recruiters see a message that the resume “has only been attached to the candidate,” and someone has to enter the details by hand (Greenhouse Support). Don’t count on that someone. Recruiters are busy, and the parsed record is what their searches see regardless.

Once an application is submitted, most systems lock it. Some portals let you update your candidate profile for future applications, but the submitted version typically stays as it was. If a big employer’s application looks broken, a cleaned-up reapplication or a direct note to the recruiter beats hoping someone fixes the record.

How to prevent resume parsing errors

One rule covers almost everything: keep all content in the document body as plain, consistent text. The checklist turns that rule into habits.

  1. Put your name and contact info in the body of the document, never in a header, footer, or text box.
  2. Use one column unless you have a specific reason not to.
  3. Stick to standard section headings: Work Experience, Education, Skills, Certifications.
  4. Pick one date format (Month Year or MM/YYYY), use “Present” for current roles, and keep dates on the same line as the title.
  5. Spell out job titles, and include company identifiers like “Inc.” or “LLC.”
  6. Skip skill bars, star ratings, icons, photos, and charts; use plain text bullets.
  7. Use a standard font (Arial, Calibri, Times New Roman) at 10pt or larger, with no letter spacing.
  8. Submit a text-based PDF or a .docx, and run the copy-paste test first. The two formats are compared in should you upload your resume as a PDF or Word file.
  9. Keep the file lean, comfortably under 2.5 MB, since that’s Greenhouse’s hard parse ceiling.
  10. Fill in profile fields yourself even when autofill works, and verify every field on the review screen before you submit.

Then test before you submit, not after. Open your PDF and try to select the text; if you can’t highlight it, neither can a parser. Free checkers go deeper, Jobscan’s scanner compares your resume against a specific job description and shows you what its parser picks up, and Teal’s resume checker runs 15-plus formatting and structure checks with a 0–100 score (Teal; Jobscan). If the extracted text looks scrambled to you, it looks scrambled to the ATS.

Do these ten things and parsing stops being a lottery. It becomes boring and predictable, the same fields, filled the same way, every time. That’s exactly what you want.

FAQ

What is resume parsing? Resume parsing is software converting your resume file into structured database fields (contact info, employers, titles, dates, education, skills) so an ATS can store, filter, and search you. It happens the moment you upload, before any human sees your application.

Why do applications ask me to retype my details after I upload my resume? Because the parsed fields, not your attached document, are what recruiters search. Employers ask you to verify or complete those fields precisely because parsing is imperfect.

Why does my application say my resume was attached but the fields are empty? That’s a failed parse; Greenhouse’s exact message is that the resume “has only been attached to the candidate.” Common causes include files over 2.5 MB, image-based uploads, and contact info in headers, footers, or text boxes.

Can an ATS read a two-column resume? Usually the text comes through, but the reading order can scramble; Enhancv’s 2026 testing found correct section order held only 35% of the time under the worst extraction method. The full breakdown is in our two-column guide.

What file format parses most reliably? A text-based PDF or a .docx file; both give the parser a clean text layer. Scanned or image-based files parse to nothing.

What date format should I use on my resume? One format, everywhere: Month Year (Jun 2022, Present) or MM/YYYY, with “Present” for current roles and dates on the same line as the job title. Consistency matters more than which format you pick.

Can I fix parsing errors after I’ve already applied? Usually not within the submitted application, most systems lock it once it’s in. Update your candidate profile for future applications, and for the live one, a cleaned-up reapplication or a direct note to the recruiter is your best move.

Sources

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

Filed under

Back to guidesBack to top