How do applicant tracking systems read a CV? Most begin by extracting text and turning familiar parts of the document—such as contact details, employers, job titles, dates, education and skills—into structured fields that recruiters can search and review.
That process is often described as a robot deciding who gets a job. Reality is less uniform. An applicant tracking system, or ATS, can store applications, parse documents, collect screening answers and support recruiter workflows. Whether it ranks, filters or automatically rejects anyone depends on the product, the employer’s configuration and the particular vacancy.
Quick answer: An ATS usually extracts machine-readable text, looks for section patterns and associates phrases with fields in a candidate profile. Recruiters may then search, filter or review that profile. A clear single-column CV, ordinary headings, selectable text, accurate dates and job-specific evidence reduce parsing problems. Keywords help only when they truthfully describe your experience.
How do applicant tracking systems read a CV? The short answer
| Stage | What may happen |
|---|---|
| Upload | The system checks file type, size and whether the document can be processed |
| Text extraction | Text is read from a document layer or, in some systems, from optical character recognition |
| Parsing | Content is associated with fields such as name, email, employer, role, date, education and skills |
| Application questions | Answers may be stored or used for employer-configured eligibility rules |
| Search and filtering | Recruiters may search terms, locations, experience or other structured fields |
| Ranking or matching | Some tools offer scores or recommendations; others do not |
| Human review | A recruiter or hiring manager evaluates the application in context |
There is no universal ATS test that all CVs must pass. A format accepted by one system can parse differently in another, and employer rules can matter more than the software brand.
An ATS is a workflow system, not one secret score
Employers use applicant tracking systems to collect applications and move candidates through stages such as new application, recruiter review, interview, reference check and offer. The system can also keep correspondence, interview notes, consent records and reporting in one place.
Resume or CV parsing is one feature inside that larger workflow. It saves a recruiter from typing every field manually and makes the database searchable. Greenhouse, for example, describes parsing as a way to locate resume details and build candidate information. Lever documents support for several common file types and explains that resumes are parsed into candidate profiles.
Some platforms add matching, artificial intelligence or automated recommendations. Others leave selection almost entirely to people. An employer can also configure knockout questions—such as legal work authorisation, a required licence or willingness to work at a stated location—without analysing the CV at all.
So “the ATS rejected me” can describe several events: a required answer was not met, a recruiter declined the application, a parsing error hid information, a role was filled, or an automated rule applied. The rejection message rarely reveals which one.
How do applicant tracking systems read a CV as text?
The first technical task is text extraction. A DOCX file stores paragraphs and formatting in a structured document. A normal PDF can also contain selectable text with coordinates showing where each word appears. The parser reads that underlying content rather than seeing the page exactly as a person does.
A scanned PDF may contain only a photograph of a page. Some systems use optical character recognition, or OCR, but quality varies with resolution, contrast, language and layout. A visually sharp scan can still produce wrong characters or reading order.
You can perform a simple check: open the final file, select all text, copy it and paste it into a plain-text editor. If the result is missing, scrambled or arranged in a confusing order, a parser may also struggle. This is not a complete ATS simulation, but it exposes many avoidable document problems.
Do not upload an image merely because it looks identical on your screen. Unless the employer specifically requests a scan, use a text-based document created from a word processor or layout tool that preserves selectable text.
Parsing turns a page into fields
After extraction, the system tries to identify what each piece of text means. A line containing an email pattern is likely contact information. A month-year range beside a company and role may be work experience. A recognised degree near an institution may be education.
Parsers combine rules, statistical models and, increasingly, language models. They use headings, punctuation, location on the page and relationships between lines. Greenhouse explains that standard resume structure guides its parsing of details such as name, email, phone, address, current title and company.
The result is not perfect understanding. A parser may mistake a client for an employer, split a two-line job title, attach a date to the wrong role or treat a sidebar as the start of the main text. Recruiters can often open the original document, but incorrect structured fields can affect search and sorting before they do.
That is why clarity helps without requiring a dull document. A restrained visual hierarchy—name, contact details, standard headings, role, employer, location, dates and evidence—works for both machines and hurried people.
How do applicant tracking systems read a CV layout?
Reading order matters more than decoration. A person can glance across two columns and understand that the left sidebar contains skills while the right column contains experience. A text extractor may read across the page, place both columns one after another or interleave them line by line.
Common sources of trouble include:
- important information placed only in headers or footers;
- text boxes and floating shapes with uncertain reading order;
- skill ratings shown only as bars, stars or circles;
- tables used to construct the entire page;
- icons without adjacent words for phone, email or links;
- very small type or low-contrast text; and
- scanned images without a reliable text layer.
A simple single-column layout is the safest default when the employer gives no template. It does not need to be ugly. Spacing, weight and modest colour can create hierarchy while the text remains in a logical top-to-bottom sequence.
Headings help the parser and the recruiter
Use familiar headings such as Work Experience, Education, Skills, Certifications and Languages. A creative label like “Where I Made an Impact” may be understandable to a person but harder to classify consistently.
Under each role, keep the pieces together. A robust pattern is:
Job title — Employer, location
Month Year–Month Year
Then add concise bullets describing work and results. Use the same date format throughout. If you worked through an agency or on a client site, label the relationship honestly instead of leaving the parser—and recruiter—to guess which organisation employed you.
Do not hide dates to outsmart a system. An unclear timeline creates more uncertainty for a person and can produce inaccurate fields. If you have a gap, career change or overlapping work, explain it briefly where relevant rather than distorting the chronology.
Keywords connect your evidence to the vacancy
Recruiters can search for skills, qualifications, tools, job titles and industry terms. Some matching tools compare the application with the job description. Using the employer’s ordinary terminology therefore helps the right evidence become findable.
The honest method is to read the vacancy and identify requirements you genuinely meet. If it asks for “inventory cycle counting” and that is part of your work, use that phrase in a bullet that explains what you did. If your CV only says “helped in warehouse,” neither a parser nor a person can infer the specific experience.
Do not paste a hidden list of keywords, repeat terms unnaturally or claim tools you have never used. White text is still text, and deception can be discovered in review or interview. More importantly, keyword stuffing replaces the evidence that makes a phrase credible.
Synonyms can help. A role may use both “customer relationship management” and “CRM,” or “occupational health and safety” and “OHS.” Include the full term and abbreviation once when they truthfully apply, then write naturally.
How do applicant tracking systems read a CV against a job?
There are three different mechanisms that people often combine under the word “matching.”
- Recruiter search: a person enters terms or filters and reviews the returned profiles.
- Rule-based screening: employer-configured questions or requirements move, flag or reject applications according to an answer.
- Automated recommendation or ranking: a model estimates relevance and orders or highlights candidates.
Not every employer uses all three. Even when a score exists, it may be one aid rather than the final decision. Greenhouse says its AI features are intended to support hiring work without removing human judgment; other products and configurations can differ.
Automated employment tools also raise fairness and accessibility concerns. The US Equal Employment Opportunity Commission warns that software and algorithms used to assess applicants can create disability discrimination if, for example, they screen out someone who could perform the job with a reasonable accommodation.
Applicants should follow the employer’s accommodation process when needed. Employers remain responsible for lawful selection practices; buying software does not transfer that responsibility to a machine.
File-type advice must follow the vacancy
PDF preserves visual layout across devices, while DOCX can be easier for some systems to parse and edit. There is no universal winner. Greenhouse currently accepts common formats including DOC, DOCX, PDF, RTF and TXT for candidate uploads; Lever publishes its own list. Limits and support can change.
Use the format requested by the employer. If both PDF and DOCX are accepted, choose the version you tested for text order and appearance. Export directly from the source document instead of printing and scanning.
Name the file clearly, for example Amina-Rahman-CV.pdf, unless privacy or portal instructions suggest another convention. Avoid version names such as final-final-new2. Check that the file opens, contains the intended pages and does not include comments or tracked changes.
Application forms can matter more than the CV
Many portals ask candidates to confirm fields after upload. Do not skip this review because the preview looks acceptable. Correct a parsed employer, date, phone number or qualification if the form permits it.
Answer screening questions carefully. If a question asks whether you hold a licence, “yes” should mean the exact required licence is valid, not that you intend to obtain it later. If wording is unclear, use an official contact rather than guessing.
Some systems treat the form as the authoritative record and the CV as an attachment. A complete CV cannot compensate for a required blank field. Conversely, auto-filled fields are not evidence that the parser understood every bullet correctly.
How do applicant tracking systems read a CV before human review?
The best preparation serves both audiences. A machine benefits from predictable structure; a recruiter benefits from relevance and evidence. Neither benefits from a page built around myths.
A strong bullet answers three questions: what did you do, in what context and what changed? “Updated inventory” is vague. “Recorded daily stock movements for three active sites and investigated differences before weekly ordering” gives a task, scale and purpose without inventing a percentage.
Use numbers when they are accurate and meaningful, not because every bullet needs a metric. Safety, care, judgment and quality do not always reduce honestly to a percentage. Specific nouns and verbs are more valuable than inflated claims.
SOAKJAM’s complete guide to preparing a professional CV covers contact details, experience, education, gaps, privacy and final checking. This article focuses on what happens after that document reaches a tracking system.
A practical ATS-friendly CV checklist
- Follow the vacancy’s instructions before any general advice.
- Use a simple one-column reading order unless a required template says otherwise.
- Place name and contact details in the main body, not only a header.
- Use familiar section headings and consistent dates.
- Write job title, employer and dates as real text.
- Describe relevant skills in evidence-based bullets.
- Use vacancy terminology only where it truthfully applies.
- Avoid ratings made only from graphics or icons.
- Export a text-based PDF or clean DOCX in an accepted format.
- Copy the final document into plain text and inspect the order.
- Open the uploaded file or preview and correct parsed fields.
- Keep a copy of the vacancy and exactly what you submitted.
Common ATS myths
- “Every ATS gives every CV a score.” Some products offer matching or ranking; many workflows depend on search, questions and human review.
- “A PDF is always rejected.” Major platforms support PDF, but the file must contain extractable text and meet the employer’s rules.
- “One missing keyword causes automatic rejection.” That can be true only if an employer creates a relevant rule or search. There is no universal keyword gate.
- “White-text keywords beat the system.” Hidden stuffing is deceptive, may be exposed and does not demonstrate competence.
- “Fancy design is the only problem.” Unclear evidence, incomplete questions and failure to meet a requirement matter even with perfect parsing.
- “The ATS makes the hiring decision.” Software can influence visibility or apply configured rules, but employers choose the process and remain responsible for it.
Frequently asked questions
Should I use “CV” or “resume” in the file?
Use the term normal in the vacancy and country. In many places they mean the same short job-application document; in US academic contexts a curriculum vitae is usually longer and different.
Are two-column CVs always unreadable?
No, but their reading order is less predictable. A tested single-column version is a safer default for an unknown system.
Can an ATS read tables?
Some can, but extraction order varies. Use tables only when the employer provides one or when you have verified the output; do not construct the entire CV from nested cells.
Should I add every keyword from the job description?
No. Include requirements you genuinely meet and show evidence. Copying terms you cannot defend creates a misleading application.
Does an ATS reject employment gaps?
There is no universal gap rule. Employer filters and recruiter judgments vary. Clear dates and a brief honest explanation are safer than hiding the timeline.
Can I know exactly why I was rejected?
Usually not from a standard notice. The decision may involve eligibility, competition, timing, recruiter review or automation. Ask politely if feedback is offered, but do not assume a parsing failure without evidence.
How do applicant tracking systems read a CV? Structure first, evidence always
How do applicant tracking systems read a CV? They extract text, infer structure and place information into fields that an employer can search, filter and review. Some systems add matching or automation, but there is no single algorithm shared by every company.
The durable strategy is not to write for a mythical robot. Write an accurate application whose order a machine can follow and whose evidence a person can trust. Use ordinary headings, accepted files, clear dates and the language of the vacancy where it honestly matches your experience.
If the document survives a plain-text check and tells a recruiter what you did, where you did it and why it matters, it is doing the real job of a CV.
Editorial review pending
Editorial information
SOAKJAM articles are designed for clarity, useful context and transparent sourcing. Important facts should be checked against the linked primary sources.
Found an error, outdated detail or a useful source we should consider?
