TL;DR

  • We rendered one synthetic resume 20 ways (14 PDFs, 6 DOCX) and ran each through the text extraction libraries behind CVPosh's free ATS checker, counting how many of 37 known content items survived.
  • The worst losses came from where text lives, not how it looks. Contact details in a DOCX header or footer, drawn as an image, or hidden behind "Email me" link labels were lost in 4 of 20 files. A resume saved as one page image lost everything (0 of 37).
  • Columns and positioned blocks kept every word but changed reading order in 4 layouts, and the checker still scores three of them 100. A high checker score does not prove a clean parse.

This tests text extraction, not any employer's applicant tracking system. We did not run Workday, Greenhouse, Lever or any other ATS and cannot say how they behave.

Update 2026-11-19: after this test we fixed two things in our own checker, so its scores moved. It now counts Word list items, numbered lists and ➤ or dash markers as bullets (pull requests #123 and #118 in our repository). We re-scored the same files: the DOCX baseline, DOCX table layout and DOCX two-column table go from 90 to 100, the DOCX header and footer files from 70 to 80, and the ➤ PDF from 90 to 100. The table shows the 2026-11-17 score and the score now. What each extractor lost did not change, and we re-ran six of the files through the live checker on 2026-11-19 to confirm.

What we tested

We wrote one fictional resume ("Jordan Sample": 228 words, four sections, nine bullets, ten skills) and generated 20 versions of it. Only the layout changed: single column, two-column sidebar, table layout, different bullet styles, emoji icons, letter-spaced headings, unusual heading names, a ligature-heavy serif font, absolutely positioned blocks, contact details as an image or as link labels, headings as images, and a full-page image.

For each file we extracted text and looked for 37 items: name, email, phone and link; role, company and dates for three jobs; the education line; ten skills; a certification; the summary; and all nine bullets as unbroken sentences. We then ran the CVPosh ATS checker's scoring code on the extracted text.

Results

PDFs were extracted with pdf-parse 1.1.1 and DOCX files with mammoth 1.11.0, the libraries the checker uses. "Order" counts how many of 10 landmarks in the resume (summary, employers, bullets, education, skills) appear in the expected sequence.

Layout Contact found Sections found Items found Order Score 11-17 / now
PDF baseline, single column all 4/4 37/37 10/10 100 / 100
PDF two-column sidebar all 4/4 36/37 8/10 100 / 100
PDF table layout all 4/4 37/37 10/10 100 / 100
PDF arrow-glyph bullets (➤) all 4/4 37/37 10/10 90 / 100
PDF CSS-drawn bullets all 4/4 37/37 7/10 90 / 90
PDF emoji icons all 4/4 36/37 10/10 100 / 100
PDF contact as image none 4/4 34/37 10/10 80 / 80
PDF letter-spaced headings all 1/4 37/37 10/10 80 / 80
PDF creative heading names all 1/4 37/37 10/10 80 / 80
PDF ligature serif font all 4/4 37/37 10/10 100 / 100
PDF positioned blocks, shuffled order all 4/4 37/37 7/10 100 / 100
PDF contact behind link labels none 4/4 34/37 10/10 80 / 80
PDF headings as images all 1/4 37/37 10/10 80 / 80
PDF full page as image none 0/4 0/37 n/a 30 / 30
DOCX baseline (Word list bullets) all 4/4 37/37 10/10 90 / 100
DOCX baseline (typed "•" bullets) all 4/4 37/37 10/10 100 / 100
DOCX contact in header none 4/4 33/37 10/10 70 / 80
DOCX contact in footer none 4/4 34/37 10/10 70 / 80
DOCX table layout all 4/4 37/37 10/10 90 / 100
DOCX two-column table all 4/4 37/37 8/10 90 / 100

Raw numbers for both extractors are in the results CSV.

Finding 1: contact details are the most fragile field

Four files lost email, phone and link together: the contact line as an image, the contact line as link labels ("Email me", "Call me") with the real addresses only in link targets, and contact details in a DOCX header or in a DOCX footer.

In the link-label PDF the address exists only as a link annotation, and neither pdf-parse nor poppler's pdftotext returned it. For the DOCX files, mammoth returned the body only; in the header file even the name was gone because we had put it in the header too. In the checker this cost 20 points each (15 for no email, 5 for no phone). A contact line in the page body, as plain text, was extracted by every method we tried.

Also: in the two-column PDF the LinkedIn URL wrapped at its hyphen inside the narrow sidebar. pdf-parse returned jordan- and sample on separate lines and pdftotext dropped the hyphen (jordansample). The checker still saw a link, but not a working URL.

Finding 2: headings need words the parser recognises

The checker looks for experience, education, skills and summary or profile. Three PDFs fell to 1 of 4 sections:

  • Headings as images. The text was not in the text layer.
  • Creative names ("About me", "Where I've worked", "Studies", "Toolbox"). All content was extracted, but the headings were not words a checklist knows. The one section still found came from "years of experience" in the summary.
  • Letter-spaced uppercase headings. pdf-parse returned E X P E R I E N C E; pdftotext returned EXPERIENCE. The same file scored 80 or 100 depending on the extractor.

Each cost 20 points. The bigger risk is any system that groups your history by section: it has to recognise the header first.

Finding 3: columns and positioned blocks keep the words, not the order

The two-column sidebar, the DOCX two-column table, the positioned-blocks PDF and the CSS-drawn-bullets PDF kept every word, but pdf-parse changed the order:

  • In both column layouts, skills and education came before summary and experience, because the left column is read first.
  • In the CSS-bullets PDF, all nine bullets came after the three job headings, detached from their employers. Our dots were small positioned shapes, which changed the order the text was written to the page.
  • In the positioned-blocks PDF, which looks like a normal one-page resume, skills came first and the name last.

pdftotext re-sorted the CSS-bullet and positioned-block files into correct visual order (10 of 10). So the problem is partly the file and partly the extractor, and we cannot tell you which one an employer's system resembles. The score did not move: the checker has no reading-order rule, so the two-column PDF and the positioned-blocks PDF both scored 100. A checklist can show that data is present, not that it is in the right place.

What did not break anything

A table layout (dates in one column, roles in the other) extracted in order as PDF and DOCX. A Palatino font with ligatures enabled returned normal text, including "Airflow" and "Certified". Emoji next to contact details left the values intact, although they came out as private-use characters in pdf-parse. A ➤ bullet was extracted as a character.

Our own checker undercounted bullets (fixed after the test)

We found this while testing. The checker counted a bullet only when a line started with •, - or *. A DOCX with real Word list bullets scored 90 and the same file with typed "•" scored 100, because mammoth does not put list markers in its raw text. The ➤ PDF also scored 90. So the checker deducted 10 points from resumes whose bullets read fine. We fixed both after the test: the same files now count 9 bullets and score 100. Two things remain. The CSS-drawn-dots PDF still scores 90 because its bullets are shapes, not characters, so the checker sees none. And the score can still disagree with a real parse in the other direction: contact details in a Word header or footer still cost 20 points (80 against 100 for the body version), and the checker still has no reading-order rule.

What to do with this

  1. Put your name and contact line in the document body as plain text, not in a Word header or footer, an image, or link labels.
  2. Use the section names a reader expects: Summary, Experience, Education, Skills.
  3. Keep headings as real text, without letter-spacing.
  4. If you use columns, paste the PDF into a text editor, or run it through the checker, and read the order that comes out.
  5. Keep long URLs out of narrow columns.
  6. Export a text-based PDF. If you can't select the text, nothing else works.

FAQ

Does an ATS reject resumes with columns?
We did not test any employer ATS, so we can't say. In our extraction every word of the two-column files survived, but the reading order changed.

Is DOCX safer than PDF?
Not in our test. DOCX avoided the ordering problems but was the only format where header and footer content disappeared.

Do icons hurt?
Icons next to text did not remove the text. Text drawn as an image did.

Does a high checker score mean my resume will parse correctly?
No. The checker starts at 100 and subtracts points when a rule fires, and it has no reading-order rule. Three layouts with changed order still score 100.

Is this representative of real templates?
No. It is one synthetic resume laid out by us in Chrome and a DOCX library. Word, Google Docs, Canva or LaTeX output can behave differently. Treat the results as a list of things to check, not as rates.

Methodology and data

  • Date run: 2026-11-17. One fictional resume, no real person.
  • Variants: 14 PDFs from Google Chrome 157.0.8103.79 (headless print-to-PDF, Letter) and 6 DOCX files from docx 8.6.0. Image variants are screenshots of the same HTML.
  • Extraction: pdf-parse 1.1.1 and mammoth 1.11.0, as in the checker. PDFs were also run through poppler pdftotext 26.07.0 (default mode) as a cross-check; both sets of numbers are in the CSV.
  • Scoring: ats-scorer.ts from the CVPosh repository at origin/main commit 184d4b4. We also uploaded five of the files (PDF baseline, contact as image, letter-spaced headings; DOCX baseline and header) to the live checker API at cvposh.com; all five returned the same score as our local run. The other 15 were not uploaded. Update 2026-11-19: the files were re-scored with the current checker (Word list items via parseDocxWithListBullets, bullet rule of PR #118); numbers are in results-after-bullet-fix.csv and results-docx-list-bullets.csv in the same folder, and six files (PDF ➤ bullets, CSS-drawn bullets, DOCX baseline, footer, table layout, two-column table) were run through the live checker, which returned 100, 90, 100, 80, 100 and 100.
  • Items: 37; an item is found when its text appears after whitespace and case normalisation, bullets as complete sentences.
  • Limits: one resume, one run (the pipeline is deterministic), no employer ATS, no OCR. Chrome's native list markers were not extracted by either extractor, so the baseline draws bullets as text ("•"). "Order" is a landmark test, not a full reading-order evaluation.
  • Reproduce: scripts and CSV at cvposh.com/research/resume-layout-test.

CVPosh is our product. The checker is free and needs no sign-up. How its score is calculated.