An applicant tracking system is a database with a form in front of it. When your resume arrives, the ATS tries to extract text from the file and slot it into fields: name, email, phone, employer, job title, start date, end date, degree, skills. Everything a recruiter later searches, filters or ranks on comes from that extraction.
So the question that matters is not 'does my resume look good' but 'does the extraction succeed'. This builder is designed around that question, and every template is graded on it honestly.
What actually breaks ATS parsing
The failures are boringly consistent. They are not about fonts being too fancy or the word 'the' appearing too often — they are structural.
- Multi-column layouts, where the parser reads across the page and interleaves two unrelated jobs into one line
- Layout tables, where cell boundaries scramble the reading order
- Contact details placed in the PDF's header or footer region, which many extractors discard entirely
- Text baked into an image or logo — invisible to extraction, including scanned PDFs
- Non-standard section headings ('My Journey' instead of 'Work Experience'), which fail heading-dictionary matching
- Skill rating bars and star ratings, which carry no readable text at all
- Missing or inconsistent dates, which break the years-of-experience calculation used for filtering
- Decorative glyphs and icon fonts, which extract as mojibake
How this builder prevents each of them
| Risk | What we do |
|---|---|
| Multi-column reading order | Single column by default; two-column templates are labelled with their real parse grade |
| Layout tables | Documents are generated from structured data, never from a table grid |
| Header/footer loss | Contact details are always rendered in the document body |
| Image-only text | Exports are text-based PDFs; every word is selectable and extractable |
| Non-standard headings | Headings default to recognised labels, and the checker warns when you rename one |
| Skill bars | Skills are rendered as comma-separated text with named proficiency |
| Broken dates | Dates are structured fields, and incomplete ones are flagged as failures |
| Exotic characters | Decorative glyphs are detected and reported before you export |
Five score dimensions, all explained
Each individual check reports its status, what it found and the specific fix. Failures are sorted to the top by weight, so the first item on the list is always the highest-value thing to change.
- Parseability (28%) — can the text be extracted in the right order at all
- Structure (20%) — are sections present, recognisably named and sensibly ordered
- Content (24%) — contact details, complete dates, titles, employers, quantified bullets, action verbs
- Keywords (18%) — coverage against a target job description, or general density without one
- Formatting (10%) — font, size, margins, length, character safety
The honest caveat
Workday, Taleo, Greenhouse, iCIMS, SuccessFactors, Lever and Naukri's parser all behave differently, and each employer configures their instance further. Any product claiming a guaranteed pass rate across all of them is guessing. What we measure is the behaviour these systems share — text extraction, section mapping, field population and keyword search — which is where essentially all real-world failures happen.
Frequently asked questions
- What is a good ATS score?
- Above 85 means nothing structural is likely to break, and your remaining work is keyword tailoring per job. 65–85 means the resume parses but is leaving information on the table. Below 65, fix the failed checks before you apply — you are probably losing whole sections during extraction.
- Do ATS really reject resumes automatically?
- Outright auto-rejection is rarer than the internet suggests. What is common is ranking: recruiters search and filter within the parsed data, and a resume whose employment history failed to extract simply never appears in the results. The effect is the same, but the cause is parsing, not a robot judging your writing.
- Should I use a two-column resume?
- Not for a large-company application portal. For a startup, a referral or a role where a human opens the file directly, the risk is much lower. Our two-column templates show their parse grade openly — around 66 to 76 rather than 92-plus — so you can make the trade knowingly.
- Is PDF or Word better for an ATS?
- A text-based PDF is read reliably by every mainstream system and preserves your layout. Use DOCX only when the application form explicitly asks for it. Never submit a scanned or image-based PDF.
- Does keyword stuffing work?
- No. Hidden white text and keyword lists are detected by modern parsers and, when a human sees them, read as dishonesty. Coverage is achieved by describing real work using the same vocabulary as the posting.