The CV Formatting That Actually Breaks an ATS

Most advice about ATS formatting is a list of prohibitions with no explanation attached. No tables. No columns. No graphics. Use a boring font. Follow it and your CV ends up looking like it was typed in 1997, which costs you something with the human reader while protecting you from a problem you may not have had.

The useful version separates the small number of choices that genuinely break a parse from the much longer list of things people worry about for no reason. This article does that, with the failure mode spelled out in each case so you can judge for yourself.

If you want the underlying model first, it is in what an applicant tracking system actually is. The short version: your CV is converted into structured fields, those fields get searched, and the original file is opened afterwards by a person. Formatting damage happens in the conversion step.

The five that genuinely break things

Two column layouts

This is the biggest one by a wide margin, and it is also the most popular design on template sites, which is an unfortunate combination.

The failure: a parser reads a page as a sequence of text in reading order. When your page has a narrow sidebar and a wide main area, a naive parser reads straight across, interleaving the two. Your skills list gets shuffled into your job history, one word at a time.

What the employer's record ends up looking like: "Advanced Excel Senior Marketing Manager English C1 January 2023 Photoshop Northgate Ltd". Nobody reads that and thinks "parsing error", they see a mess and move on.

Modern parsers handle columns better than they used to, and some handle them well. The problem is that you cannot know which one you are submitting to, and the downside is total while the upside is a slightly prettier page.

The fix: one column. If you want visual interest, get it from spacing, weight and a single accent colour rather than from a sidebar.

Text inside images

A parser reads text. It does not read pictures of text.

This catches people in three ways. A CV exported as an image or scanned to PDF contains no text at all and parses to an entirely empty record. A header designed in Canva as a graphic loses your name and contact details. An infographic style skills section loses every skill.

What the employer sees: either nothing, or a CV with no name and no phone number. There is no recovery from this, because there is nothing for a recruiter to search and often no way to contact you.

The fix: check that you can select the text in your PDF with your cursor. If you cannot select your own name, neither can the software.

Contact details in the header or footer

Word's header and footer areas are structurally separate from the body of the document, and a large number of parsers skip them entirely.

What the employer sees: a candidate record with a name pulled from somewhere in the body, no email and no phone. You can be shortlisted and still not contacted, which is the most frustrating failure on this list because everything else worked.

The fix: put your name and contact details in the first few lines of the body text. It looks identical on the page.

Tables used for layout

A table used to display data is usually fine. A table used to position your entire CV is not, because the parser reads it cell by cell in an order that may bear no relation to how it looks.

Skills grids with four columns of single words are the worst version, because the reading order is unpredictable and the output is a scattering of disconnected terms.

The fix: use paragraph spacing and tab stops for alignment. If you want a skills section in two visual columns, write it as comma separated lines instead. It reads the same to a human.

Non standard section headings

Parsers look for recognisable headings to decide where each section starts. "Experience", "Work Experience", "Employment History", "Education", "Skills" are all understood. "My Journey", "Where I've Made an Impact" and "The Toolkit" are not.

What the employer sees: your jobs filed under nothing in particular, or your whole career landing in a general notes field rather than the employment history.

The fix: creative headings cost you a real function and buy you nothing. Use the boring words.

The things people worry about that do not matter

Two pages. The parser reads the whole document. Page count is a human convention, discussed properly in how long a CV should be.

Serif fonts. Times New Roman, Garamond and Cambria parse exactly as well as Arial. Font choice is a design decision about readability, not a technical one.

Bullet characters. Standard bullets, dashes and circles are all fine. Exotic symbols and emoji may not render everywhere, so a plain bullet is the safe choice, but no parser has ever rejected a document over a bullet.

Colour. Parsers read text, not colour values. A coloured heading or accent line is invisible to the software and useful to the human. Keep contrast high for readability, not for the machine.

A photo. It is simply ignored. Whether to include one is a regional convention rather than a technical question.

Bold and italic. No effect on parsing. Use them for emphasis as normal.

Hyperlinks. Fine, but write the visible text so it still makes sense without the link, because the plain text version is what gets stored. "linkedin.com/in/yourname" beats "click here".

File name. It has no effect on parsing. A sensible name still helps the human who downloads forty files, so use something like "firstname-lastname-cv.pdf".

The middle ground: dates and job titles

Two things sit between the clear failures and the harmless myths, and they cause more silent damage than people realise.

Date formats. Parsers try to convert your dates into start and end fields. They handle "January 2023", "Jan 2023" and "01/2023" well. They struggle with "'23", with seasons like "Spring 2023", and with mixed formats in one document.

What happens when it fails is subtle: the role is captured but the dates are blank or wrong, so your record shows less experience than you have, or shows a gap that does not exist. Since a recruiter often filters on years of experience, this quietly removes you without anyone noticing.

Pick one format and use it everywhere. "March 2021 - June 2024" and "Present" for the current role.

Job title styling. If you set your job title in a large bold font that looks like a section heading, a parser may treat it as one and misfile everything under it.

Keep the hierarchy clear: section headings are the largest text on the page, job titles a step below, employer and dates below that. Visually distinct, but unmistakably subordinate.

What a broken parse actually costs you

Worth being precise here, because the internet tends to describe every formatting issue as instant rejection, which is not true and makes the real risks harder to see.

A broken parse rarely triggers an automatic rejection. What it does is remove you from searches. If your skills ended up interleaved with your job titles, a recruiter searching for "SQL" may still find you, because the word is in there somewhere. But if your dates are missing and they filter for five or more years of experience, you are gone.

The second cost is the columns a recruiter scans before opening anything. Most systems show a list view with name, current title, current employer and years of experience. If those fields are empty or wrong, your row looks like an incomplete application next to thirty complete ones, and you do not get opened.

The third cost is entirely human. Many application forms ask you to correct the parsed version. A candidate who has to retype everything because their file parsed badly is more likely to abandon the application, which is a self inflicted rejection that never appears in anyone's statistics.

How to test your own CV in two minutes

You do not need a tool for the first test.

The copy paste test. Open your PDF, select all, copy, and paste into a plain text editor or an empty email. What you see is roughly what a parser sees. Read it top to bottom. If your skills are tangled into your job history, you have a column problem. If your name and phone number are missing, you have a header problem. If nothing at all pastes, your file is an image.

The order test. In that pasted text, check that each job title is immediately followed by its employer, dates and bullets. If content from another part of the page has been spliced in, the reading order is broken.

The dates test. Check that every date survived in the same format you wrote it. Missing dates are the most common silent failure.

That test catches the majority of real problems, and it costs nothing. For a more precise view of the structured fields an employer's system would build, a parsing check gives you the extracted record itself rather than the raw text.

A safe layout, described

If you would rather just be told what to do, here it is in full.

One column, full page width. Name and contact details as the first two lines of the body, in normal text. A short professional summary next, in prose. Then Experience, with each role as: job title, employer, location, dates on one or two lines, followed by bullets. Then Education in the same shape. Then Skills as grouped comma separated lines. Optional short sections after that: certifications, languages, volunteering.

Standard headings. One date format. A single font in two or three sizes. Bold for job titles and headings. One accent colour if you want one, used for headings and a rule line. Nothing in the header or footer except perhaps a page number. Exported to PDF from a text based document.

That layout parses cleanly in every system, and it can still look good. Design quality on a CV comes from spacing, alignment and restraint, not from sidebars. The complete content checklist for filling that shape is in how to write an ATS friendly CV.

Where the templates come from

It is worth understanding why so many CVs have the exact layout that causes problems, because it is not carelessness.

Template galleries are ranked by how they look as thumbnails. A two column design with a coloured sidebar, an avatar and a row of skill rating bars looks striking at postage stamp size, so those templates rise to the top and get downloaded most. Almost none of them were designed with parsing in mind, and the ones sold as "ATS optimised" carry no certification behind that phrase, because no such certification exists.

Word's own built in templates are a smaller version of the same problem. Several of them place contact details inside a header and several use tables for the layout grid, both invisible to you while you edit.

None of this means a designed CV is a bad idea. It means the design has to survive being converted to plain text, and the copy paste test tells you within seconds whether yours does. If you like a two column template enough to keep it, keep it for direct email applications and speculative approaches, where a human opens the file immediately, and use a single column version for anything submitted through a portal.

PDF or Word

Briefly, because it belongs here.

Text based PDF is the safer default for almost every application. It preserves your layout for the human reader and has been parsed reliably for many years.

Send Word only when the advert or the recruiter asks for it, which happens most often with recruitment agencies, since they frequently rebrand CVs before sending them to clients. In that case a simple Word document is genuinely more useful to them.

Never send a PDF that is a scan or an exported image, and never send a file exported from a design tool as flattened artwork.

Frequently asked questions

Do ATS systems really reject two column CVs?

They do not reject them, they mangle them. The record built from your file may be scrambled, which affects whether you appear in searches. Some modern parsers cope well, but you cannot tell in advance which one you are dealing with.

Are icons for phone and email a problem?

The icons themselves are ignored, which is harmless as long as the text next to them is real text. The failure case is a contact block designed as a single image, where the details disappear entirely.

Should I remove all formatting to be safe?

No. A plain text wall is harder for a human to read, and the human decides. Keep clear headings, bold job titles and generous spacing. Restraint, not removal.

Does a text box work if the text inside it is selectable?

Often not. Text boxes are frequently read out of order or skipped, even when the text is selectable. Use normal paragraphs instead.

My CV was designed for me. How do I check it?

Do the copy paste test. If the pasted result reads in the right order and contains your contact details, the design is fine regardless of how it looks. If not, ask the designer for a single column version, or rebuild it yourself for online applications and keep the designed one for direct sending.

What about CVs submitted through LinkedIn Easy Apply?

Your profile carries most of the weight there, since the filtering happens inside LinkedIn. The attached file still gets opened later, so it should still parse cleanly, and the profile itself deserves its own attention, as covered in optimising your LinkedIn profile.

See the extracted version of your CV

The copy paste test shows you the raw text. What it cannot show is which fields an employer's system built from that text: which roles, dates, sections and skills were recognised as such. CvLaunch reads your file and shows you the structured result, so you can see exactly where a layout choice cost you a field. It takes a few seconds, it is free, and there is no sign up.