How ATS Parses Your Resume

When you upload a resume, the ATS extracts plain text from your file, splits it into sections, maps content to database fields (name, experience, education, skills), and indexes keywords for recruiter search. Formatting that disrupts reading order—columns, tables, graphics—breaks this pipeline and leaves your profile incomplete.

Parsing is not judgment—it is data entry. The system tries to build a structured record of your career. Your job is to make that record accurate and searchable.

1. Extraction — Raw text pulled from PDF or DOCX
2. Segmentation — Text divided into sections (Experience, Education, Skills)
3. Field mapping — Content assigned to structured fields (job title, company, start date)
4. Indexing — Keywords added to search index for recruiter queries

Failures at any stage reduce match scores and search visibility. Most candidate-facing advice focuses on stage 4 (keywords), but stages 1–3 silently eliminate qualified applicants first.

The parser opens your file and extracts character streams. DOCX stores text in XML paragraphs; PDF stores positioned glyphs that software reassembles into lines. Extraction succeeds when every word is real text—not an image, not a vector outline.

Common extraction failures: scanned paper resumes, Canva exports, infographic templates, and password-protected PDFs. Test by selecting text in your PDF—if you cannot highlight individual words, extraction will fail.

Read PDF vs DOCX for ATS for export guidance that preserves extractable text.

Algorithms look for heading patterns to split your resume into sections. Standard labels like "Experience," "Work History," "Education," and "Skills" trigger reliable splits. Creative headings ("My Path," "Expertise") may be ignored or misclassified.

Multi-column layouts scramble segmentation because the parser cannot tell whether "Python" in the sidebar belongs to Skills or is floating between Experience lines. Our two-column resume analysis shows typical failure patterns.

Within Experience, the parser attempts to identify repeating structures: employer name, job title, location, start date, end date, description bullets. Regex and ML models trained on millions of resumes guess field boundaries.

Tables that put dates in one cell and titles in another break mapping. The system may store "2020–2023" as a company name. Follow the ATS format guide for parser-friendly entry formatting.

Successfully mapped text enters a search index—like a specialized search engine for candidates. Recruiters query: "Python AND AWS AND 5 years." Your profile appears if indexed terms match.

Keyword matching against the job description produces match scores on many platforms. Terms in your summary and recent roles typically weight higher than skills mentioned once in a decade-old job. See resume keywords guide.

Workday — Common at Fortune 500. Robust parser but strict on application field auto-fill. Preview often looks plain.

Greenhouse — Popular at tech companies. Generally good PDF parsing; structured apply forms may duplicate parsed data.

Lever — Similar to Greenhouse for startups and mid-size tech. Favors clean single-column PDFs.

Platform-specific guides on our site cover quirks for Taleo, iCIMS, and SuccessFactors. Principles remain: linear text, standard headings, honest keywords.

Recruiters often open your original PDF attachment—it looks like your designed layout. But search, filtering, and initial ranking use parsed data. A beautiful resume that parses poorly will not surface in keyword searches even if a recruiter would love the design when they open it manually.

Some recruiters never open attachments until after searching the database. For those workflows, parsed data is the only version that exists.

1. Plain-text paste test—fix anything out of order
2. Switch to single-column layout
3. Rename sections to standard headings
4. Move contact info from headers into body
5. Re-export from Word/Google Docs as text-based PDF
6. Run ATS scan and resolve formatting warnings

Detailed fixes for each issue are in seven formatting mistakes.

Think of your resume as two documents: the one humans read and the one the ATS builds. They should tell the same career story. Parsing bridges upload to search—when it breaks, keywords never matter because they never reach the index.

Verify both versions with the free ATS resume checker and the format checker before every application.

You often will not receive an error message when parsing fails. Watch for these signals immediately after upload on Workday, Taleo, iCIMS, or SuccessFactors career pages:

• Empty work history fields — Experience section blank despite a full resume attached
• Wrong employer names — Dates appear as company names, or job titles land in the employer field
• Missing skills — Sidebar or table skills never populate into the profile skills section
• Garbled characters — Smart quotes, bullets, or accented letters render as ? or □ in the preview pane
• Partial education — Degree name parsed but institution missing, or graduation year shows as 0199

If you see any of these, do not submit. Fix formatting, re-export as plain DOCX, and re-upload. Submitting with corrupted auto-fill stores bad data in your candidate profile for every future application at that employer.

Beyond Workday, Greenhouse, and Lever, three enterprise platforms dominate parsing discussions because failure rates run higher:

Oracle Taleo — Oldest parser in widespread use. Strictest about DOCX, dates in MM/YYYY format, and zero tables. Two-column layouts fail roughly 41% of the time in our testing. See our Taleo guide.

iCIMS — Uses Text Kernel parsing. Moderate reliability on standard formats; struggles with retail and hospitality visual templates. Profile data persists across applications—verify auto-fill field by field. See iCIMS guide.

SAP SuccessFactors — Comprehensive profile model with international date format quirks. EU CVs with photos and graphics break extraction more often than US single-page resumes. See SuccessFactors guide.

If you apply across multiple enterprise systems, maintain one ultra-plain master DOCX that passes all four parsing stages on every platform—extraction through indexing.

Parsing builds your candidate profile. Ranking compares that profile to a specific job requisition. You can parse perfectly and still rank low if keywords do not overlap—or parse poorly and rank zero because your skills never entered the index at all.

• Parsing problems — Fix with formatting: single column, standard headings, plain-text test
• Ranking problems — Fix with tailoring: summary keywords, skills order, bullet reordering
• Knockout rules — Hard filters on years, clearance, or work authorization; tailoring cannot override

Always diagnose parsing first. A tailored resume that scores 85% in a checker but parses at 50% in the employer's ATS will underperform in production. Run both format and keyword scans before every submission.

Consider a marketing manager applying through Greenhouse. Stage 1 extracts text from a PDF exported from Google Docs. Stage 2 identifies "Experience," "Education," and "Skills" headings. Stage 3 maps "HubSpot" and "Acme Corp" into structured fields with dates 03/2021–Present. Stage 4 indexes "demand generation," "MQL," "Google Analytics 4," and "B2B SaaS" for recruiter search.

If the same resume used a two-column Canva template, Stage 1 might succeed partially but Stage 2 interleaves sidebar skills with experience lines. Stage 3 stores "HubSpot" without linking it to Acme Corp. Stage 4 indexes orphaned terms—recruiter search for "HubSpot AND marketing manager AND B2B" returns incomplete or no match despite the human-readable PDF looking polished.

This is why identical career content produces different outcomes based on layout alone. Read our what is an ATS guide for the full hiring pipeline context beyond parsing.

Three tests catch most parsing failures before an employer sees your file. Run all three when changing templates or exporting from a new tool:

1. Select-and-copy test: Open PDF in viewer, select all text, paste into Notepad. Reading order should match visual layout.
2. Field mapping review: On Taleo, iCIMS, or Workday, upload to a test profile (your own account) and inspect auto-filled employer, title, and date fields.
3. Keyword visibility scan: Run your file through a resume checker to confirm skills in the sidebar or skills section appear in extracted text—not omitted or orphaned.

If test 1 fails, fix layout before applying anywhere. If test 1 passes but test 2 fails, simplify structure further—usually tables, columns, or headers. If tests 1–2 pass but match scores are low, you have a tailoring problem, not a parsing problem.

Some legacy HR workflows still accept faxed or scanned paper resumes converted to image PDFs. OCR (optical character recognition) may extract text with errors—"rn" misread as "m," "2019" as "2O19." Modern ATS parsers expect digital-native files. If you must submit a scan, recreate the content in Word rather than uploading the image PDF.

Veterans and federal applicants sometimes have scanned service records attached as supplements— keep the primary resume as digital DOCX while attaching scans only where explicitly required. Primary application resume should always be machine-readable text.

When employers offer both upload and copy-paste fields, prefer upload with a verified DOCX if the preview looks correct. Use copy-paste only as fallback when upload parsing fails twice— some Taleo and iCIMS configs treat pasted text more reliably than problematic PDF uploads.

When you apply via LinkedIn Easy Apply or Indeed, the job board may send structured profile data, your uploaded file, or both to the employer ATS. The employer parser then processes whatever arrived—sometimes the board's summary, sometimes your attachment. Inconsistent data between profile and file creates recruiter confusion.

Best practice: keep LinkedIn profile titles, dates, and skills aligned with your ATS-safe resume, and upload the same PDF or DOCX you would submit on the company careers page when the board allows attachment override. Direct company applications remain the most reliable path for parsing control.

Understanding parsing explains why identical qualifications produce different outcomes across application channels—the file and path matter as much as the keywords inside the document.

Re-test parsing whenever you change resume templates—even minor Word style updates can alter extraction order.

Save a plain-text export of your parsed resume in your application folder for quick diff checks after edits.