How to check whether your CV PDF is actually readable

Running the test

1. Open your exported PDF in whatever you normally read PDFs with — a browser is fine and is the most representative.
2. Select all the text. Ctrl+A on Windows and Linux, Cmd+A on a Mac. If nothing highlights, stop here: you already have your answer, and it is the worst one.
3. Copy, then paste into something with no formatting at all. Notepad, TextEdit in plain-text mode, or the body of a new email you are not going to send.
4. Read it top to bottom, slowly, as though you had never seen the CV.
5. Check four things specifically: your name, your phone number and email, every job title with its employer, and the dates. Those are the fields most likely to be extracted into a form.

Reading the result: five failure modes

A pass looks boring — your CV as plain text, in order, with the formatting gone. Anything else falls into one of these, and each has a different cause and a different fix.

• Nothing selects at all. The page is an image. This happens when a CV has been scanned, screenshotted, or exported from a design tool with text converted to outlines. There is no text to recover; the document has to be rebuilt from something that produces real text.
• Text selects but pastes as gibberish — random letters, boxes, or symbols. The font has been embedded without the mapping that says which shape means which character. It looks perfect and is unreadable to software.
• Text is readable but interleaved. Your job title is followed by an unrelated skill, then the next line of the job. This is the two-column layout problem: the reading order runs across the page rather than down the column.
• Your contact details are missing entirely while everything else is fine. They were placed in the page header, and headers are frequently discarded.
• Whole sections vanish — usually skills, sometimes education. They were inside a text box, a graphic, or a table that flattened badly.

Why a good-looking PDF can still be unreadable

This is the part that surprises people, and it is worth understanding because it explains why the test is necessary at all.

When a PDF embeds a subset of a font — which almost every modern exporter does, to keep the file small — the characters in the file are not letters. They are glyph numbers: an instruction to draw shape 1, then shape 2. Whether anything can turn those numbers back into the letters "Arjun Mehta" depends on the exporter also writing a small translation table into the file. That table is invisible, it is optional, and when it is missing the document looks flawless and copies out as nonsense.

So "it looks fine" tells you nothing about this, and neither does the file being a PDF rather than an image. The only way to know is to take the text out and look at it.

What we found testing our own export

It would be poor form to publish this test without running it on what this site produces, so it is now part of our automated test suite rather than something we assert.

The suite renders a CV in the Minimal ATS layout using the same code path as the download button, pulls the text back out of the resulting PDF, and checks that the name, both employers, both job titles and the section headings all appear, in reading order, with the current role before the previous one. If any of that stops being true, the build fails.

Writing it turned up something worth passing on. The candidate's name was recoverable from the file almost trivially, because it is also written into the document properties — but nothing from the body of the CV was, until the translation table described above was decoded properly. In other words, a crude check on a PDF can find the name and conclude the file is fine while the entire work history is still unreadable. If you are testing your own CV, check a job title and an employer, not just your name at the top.

Fixing what the test finds

In order of how often it is the answer:

1. Move to a single-column layout. This resolves interleaving outright, and interleaving is the most common serious failure.
2. Move contact details out of the page header into the first lines of the body.
3. Replace anything drawn as a graphic — skills charts, rating bars, a name in a logo — with plain text.
4. Remove tables used for layout, and set the same information as ordinary lines.
5. Re-export rather than converting. Printing to PDF from a preview, or exporting an image and wrapping it, both discard the text layer.
6. Run the test again on the new file. A fix you have not re-tested is a hypothesis.

What the test does not tell you

It is worth being precise about the limits, because ATS advice online is mostly overclaiming.

Passing means your file contains real text in a sensible order. It does not mean any given system will extract every field into the right box, that you will rank highly against a search, or that a human will read it. Those depend on the system, the recruiter, the role and the rest of your CV.

What passing does is remove an entire category of silent failure — the one where you are not contacted and never find out that your employment history arrived as a scramble. That is worth thirty seconds.