How Does ATS Work? Keywords, Scanning, and How to Pass Any System
It does not scan your resume like a person does. It extracts the text layer of your document, performs a keyword search within, and generates results based on exact matches. Whether your resume makes the cut depends on the match between your wording and the recruiter's input rather than your intention.
What are ATS keywords? Are ATS exact matches or use synonyms?
ATS keywords are specific words the recruiter types into the platform when searching for candidates from its database. Your resume passes or fails depending on whether the text contains exact keywords. ATS search is a keyword search of a database, not a reading comprehension test.
Exact match is the default mode of almost all platforms. Assume that each platform understands only the plain text, then use find-and-replace to reflect the exact vocabulary used in the job description.
There are differences between platforms, and you cannot usually know which platform an employer uses:
| Platform | Does it recognize synonyms? |
|---|---|
| Taleo | No, e.g., "P&L ownership" ≠ "profit and loss." Used by most big enterprises and financial companies. |
| Greenhouse / Lever | Yes, basic set of common equivalents, e.g., "Python developer" = "Python engineer." Still depends on the employer configuration. |
| Workday | Not consistent, depends on the employer's setting. Same ATS, different behavior across companies. |
| Ashby | Same as Workday but better. Mostly used in tech companies. Scope of synonyms is limited, e.g., "managed team" ≈ "built team." |
Keyword synonyms and semantic inference are two distinct features. The first is a simple lookup table; very few platforms support it. Semantic inference means a deep analysis of your experience to understand whether you "built a team of 12 from scratch" (leadership), "drove ARR growth" (financial metrics), or anything else. No ATS uses this feature yet because such capability requires a large language model.
Implied competencies don't pass through either. If the job description requires "cross-functional stakeholder management," but your resume mentions "collaboration with various departments," then your resume will fail. You must use the exact phrase.
Can ATS scanners read PDFs, tables, and hidden white text?
ATS processes only the text layer of your file, not visual rendering.
Modern ATS supports PDF and DOCX. DOCX is typically easier to process because it does not require additional parsing steps. When in doubt, go with DOCX.
Text the same color as the background is invisible to human beings but fully readable by machines. Back in 2015–2019, some systems processed resumes using image-to-text conversion. This technique is ineffective now. Greenhouse, Lever, Workday, and Ashby process the text layer directly; Taleo behaves differently depending on deployment year.
When a discrepancy is detected, one of the following happens:
- Keywords are indexed but flagged for manual review
- The system flags the discrepancy and routes your resume for human verification
- Your application may be disqualified automatically
Parsed tables contain a mix of columns and headers detached from the rows' data. They tend to corrupt the structure of your resume. Do not use tables in the header, summary, and skills sections. If you need to list items horizontally, use two plain-text columns instead.
Tips to create an ATS-friendly resume: format and structure
ATS is a keyword matcher and expects a specific structure.
- Single-column design. Multi-column layouts divide the text into fragments merged incorrectly during parsing.
- Standard headings: Experience, Education, and Skills. Recruiters perform searches by section names, and creative headings like "Where I Have Been" or "What I Have Done" are not recognized.
- Contact info goes to the main text body, not headers or footers. Most parsers ignore that area.
- No text boxes or images. Text inside them is unreadable for the parser.
- No tables in sections.
Vocabulary matching is precisely what ATS needs. The trouble is sentence structure. A recruiter opens your resume and finds sentences copied from the posting. Nothing left to read: only responsibilities without accomplishments.
In this example, you use the same wording as in the description but provide more details and context. ATS recognizes keywords. Recruiters see a result.
How to beat ATS without stuffing your resume: keyword placement
A high percentage of keywords relative to visible text triggers manual review rather than automatic approval. ATS weights keywords by their placement, not frequency of usage.
| Placement | Weight |
|---|---|
| Job title | Maximum. Use the exact job title from the posting. |
| Summary | High. Up to 3–4 priority keywords in 2–3 opening sentences. |
| First bullet of each role | Medium. Each position starts with a keyword connected to the achievement. |
| Skills section | Low. Tools and technology, not soft skills. |
One keyword in your job title works better than the same keyword in 6 bullet points.
Extract 8–10 keyword phrases from the job description. Write a separate sentence for each phrase connecting a keyword with a concrete achievement. Put your top priorities into the title and summary. Others go into bullet points, each with a specific number.
Doing this manually takes 20–30 minutes per application. It's worth doing once to understand the pattern. If you're applying to multiple positions a week, JobHire.AI runs the same audit in under a minute.
