You send your CV, the confirmation email arrives, and then nothing happens. Before a recruiter ever opens your file, an ATS (Applicant Tracking System) has already turned it into plain text and sorted that text into database fields. If that conversion goes wrong, your experience never enters the system at all.
A CV no software can read is not rejected: it is simply never seen. This guide explains how an ATS reads your CV, which design decisions break that reading, how to check your file in two minutes, and what changes in the Spanish and Latin American market. At the end you get a plain-text structure you can copy straight into Word.
What an ATS Does With Your CV Before a Human Sees It
An ATS is a candidate database with an application form attached. Its first job is parsing: opening your file, extracting the text, and mapping that text onto fields — name here, each job title there, dates in their own column.
- Your file arrivesPDF, DOCX or a pasted web form
- Text extractionThe document becomes a single stream of characters, in one reading order
- Structured recordJob titles, employers, dates and skills, each in its own field
- Recruiter searchFilters and keyword queries run against those fields
Nobody searches your layout. Recruiters search the extracted fields.
The step that fails is the second one. A PDF stores where each character is painted on the page, not the order in which the sentences should be read. That order lives in a separate layer: as the US federal accessibility programme puts it, a document's tag structure establishes the logical structure of the document, as well as the logical reading order that software follows.
So a CV can look flawless on screen and still arrive scrambled. The visual layer and the reading layer are two different things, and only the second one reaches the recruiter's search box.
That extracted record also outlives the application. Companies query their own database when a new role opens, sometimes months later, so a badly parsed CV keeps failing every future search without anyone realising there was a file to begin with.
The Format Mistakes That Break ATS Parsing
Nearly every parsing failure comes from the same short list. None of them is about the quality of your experience: they are container problems, and each one takes minutes to fix.
Breaks the parser
- Two or three columns: the extractor interleaves both sides and merges one job into another
- Tables used for layout, even with invisible borders
- Contact details placed inside the header or footer of the document
- Text boxes and shapes, whose content often falls outside the reading order
- Icons or bars standing in for data, such as five stars next to a language
- A PDF exported as an image, with no selectable text
- Invented section names such as My Journey instead of Work Experience
Parses every time
- One single column, top to bottom
- Standard headings: Work Experience, Education, Skills
- Contact details in the body, on the first lines of the document
- Dates written identically in every entry, month and year
- One standard font and plain bullet points
- Text you can select, copy and paste out of the file
Two of these deserve a longer explanation, because they are the ones designers recommend most often. Two-column layouts fail because the extractor reads the page as one continuous flow: your left-hand skills sidebar ends up spliced line by line into the job description on the right, and neither survives.
Headers and footers are the second trap. Word and most design tools place them in a separate object outside the main text flow, so a phone number in the footer often does not exist for the parser. The recruiter sees a promising profile with no way to contact you.
The ATS-Friendly CV Format (Copy This Structure)
An ATS-friendly CV is not an ugly CV. It is a single-column document with predictable headings, where the design lives in typography and white space instead of in boxes, grids and colour blocks.
Copy the skeleton below into a blank document, replace the placeholders with your own details, and you have a file that any parser can read end to end. Keep the section names exactly as they are: those words are what the ATS looks for to know where one block ends and the next begins.
FULL NAME
Target job title
City, Country - Phone - Email - LinkedIn profile URL
PROFESSIONAL SUMMARY
Two or three lines: your current role, years of experience, the two
skills the job posting asks for, and one measurable result.
WORK EXPERIENCE
Job title
Company name, City, Country
Month YYYY - Month YYYY
- Achievement with a number, the tool used and the outcome.
- Achievement with a number, the tool used and the outcome.
- Responsibility that matches a requirement in the job posting.
Job title
Company name, City, Country
Month YYYY - Month YYYY
- Achievement with a number, the tool used and the outcome.
EDUCATION
Degree or qualification
Institution, City, Country
Month YYYY - Month YYYY
SKILLS
Skill, Skill, Skill, Skill, Skill, Skill
LANGUAGES
Language - level (for example, English - C1)
CERTIFICATIONS
Certification name, Issuing body, YYYYTwo details inside that skeleton decide most of the parsing quality. Keep job title, company and dates on three separate lines: crammed onto one line with dashes, many parsers read the whole string as a single job title. And write dates the same way in every entry — March 2021 - June 2024, not 03/21 in one job and 2021-2024 in the next — because the field that calculates your years of experience is filled by pattern matching, and an unrecognised pattern leaves it empty.
That structure is deliberately plain, and it is also the default output of most modern editors. If you would rather not start from a blank page, an AI CV builder generates this single-column layout for you and exports a text-based PDF without you having to think about it.
The same rule applies when you create your CV with AI: the tool writes the content, but you stay in charge of the container. A brilliant summary inside a three-column template still arrives as noise.
PDF or Word: Which File Passes the ATS Filter
Both work, with one condition: the file must contain real, selectable text. The format matters far less than how the file was produced.
| File | How a parser handles it | When to use it |
|---|---|---|
| PDF exported from a text editor | Read correctly; layout stays fixed on any device | Default choice for email and most application portals |
| PDF exported as an image, or a scan | Zero text extracted: the parser sees one picture | Never, unless the employer asks for a signed scan |
| Word (.docx) | Read correctly and the safest bet with older systems | When the portal explicitly asks for Word, or you are unsure |
| Old formats (.doc, .odt, .pages) | Inconsistent; some parsers skip the file entirely | Convert to .docx or PDF before sending |
The image PDF is the silent killer, and it is easy to produce by accident: exporting from a design tool, scanning a printed copy, or taking a screenshot of a page. The principle is the same one that governs document accessibility — when a file has no renderable text, software cannot read or interact with its content, and optical character recognition has to be run first.
How to Check Your CV Is Readable Before You Send It
You do not need special software to test your own file. Three checks, in this order, catch almost every parsing problem:
- The select-all test. Open your PDF, press Ctrl+A and then Ctrl+C, and paste into a plain-text editor such as Notepad or TextEdit. If nothing pastes, your PDF is an image. If the text pastes but jumps between columns, your layout is interleaving.
- The reading-order test. Read that pasted text from top to bottom, as a stranger would. Every job title should still be followed by its own company, dates and bullets. If a skills list has landed in the middle of a role, the columns are the problem.
- The search test. In the PDF, press Ctrl+F and search for your own phone number and for one skill from the middle of the document. If the viewer finds neither, that content is invisible to the parser too.
Each failure points at one fix. No text pasted at all means re-exporting from the original editable file instead of from the design tool. Interleaved text means moving to a single column. A phone number the viewer cannot find means pulling it out of the header and into the first lines of the body.
Anything that fails these three checks fails inside the ATS as well, and no amount of keyword tuning compensates for text the system never extracted.
What Changes in the Spanish and Latin American Market
The parsing rules are universal, but the expected content is not. A CV written for the Spanish or Latin American market carries fields that a US template does not, and some of them interact badly with automated filters:
- Photo, ID number and date of birth. Still common in Spain and much of LatAm, and harmless as text, but never place them inside an image, a table or a sidebar: that is where they take the rest of your contact block down with them.
- The data-protection clause. A short line at the end is fine. Keep it as normal body text, not as a footer or a text box.
- Accents and the letter ñ. Text-based files handle them correctly; files exported as images turn them into recognition errors after OCR, which is one more reason to avoid design exports.
- Portal forms. On LinkedIn, InfoJobs or Computrabajo, your profile fields often outrank the attached file. Fill them in with the same job titles and dates as your CV, so both agree.
- The Europass format. The European Commission describes it as one of the most recognised CV formats in Europe, and its export is text based, so it parses cleanly. It is a safe fallback for public-sector and European applications, though its fixed layout gives you little room to prioritise.
Parsing Is Step One, the Match Is Step Two
Once your file is readable, the ATS stops being an obstacle and becomes a search engine. From that point the question is no longer whether the system can read you, but whether what it read matches the job posting.
That is a different job, and it is where the real gains are. Learning which keywords to pull from a job posting and where to place them only pays off once the parser can see the page. The same goes for the tailoring itself: matching your CV to a specific job description is what turns a readable file into a shortlisted one.
If you want to know which requirements your current CV already covers and which ones it misses, you can compare your CV against the job posting and get the gaps listed for you, one requirement at a time. Fix the container first, then close the gaps.
Frequently Asked Questions
Does an ATS reject a CV that has a photo?
No. A photo is ignored by the parser, not penalised, as long as it is a separate image and not the container for your text. The real risk is a CV exported entirely as an image, where the photo is the only thing the system finds.
How many pages should an ATS CV have?
One or two. Length is not a filter criterion in itself, and no parser stops at page one. Two pages are justified when you have more than about ten years of relevant experience; beyond that you are diluting your own keywords.
Do ATS systems read PDFs correctly?
Yes, provided the PDF was exported from a text editor and its text can be selected. The old advice to always send Word comes from a time when PDF extraction was unreliable. Today the deciding factor is whether the file has a real text layer.
Does hiding keywords in white text work?
No, and it can get you disqualified. The parser extracts white text exactly like any other text, so the recruiter sees a keyword list that does not appear in the visible document. It reads as an attempt to game the filter.
Is a CV made in Canva or a design tool ATS-friendly?
Only if you export a text-based PDF and choose a single-column template with no text boxes. Most attractive templates use columns and shapes, which is exactly what breaks extraction. Run the select-all test before sending one.
What should I do if the portal asks me to retype my CV?
Fill the form in completely, even when you also attach the file. Portal fields feed the ATS directly and are usually what the recruiter filters on, so an empty form with a perfect attachment still ranks badly.