What an ATS-Friendly CV Actually Means

I spent years in tech at Amazon, Google and Spotify, on both sides of the hiring table. Then I built ReqHunt, which reads the careers pages of 8,317 companies every hour. That means I look at the raw output of applicant tracking systems all day — the actual API responses, the actual application forms, the actual field names.

So I want to be blunt about something: most of what you have read about ATS-friendly CVs is wrong, and it is wrong in a specific, traceable way. It was written by companies selling resume templates, and the fear is the product.

Here is what actually happens to your CV.

New to this? Start with the background piece — what an ATS resume actually is — and come back here for the execution detail.

The myth that will not die

You have seen the claim. Something like "75% of resumes are rejected by the ATS before a human ever sees them." It appears in thousands of articles. It has no credible source. It traces back to marketing copy from resume-optimisation vendors, gets cited by a blog, and that blog gets cited by the next one.

The claim requires applicant tracking systems to do something they mostly do not do: automatically reject applications based on a keyword-matching score.

It has a sibling you have also seen: the claim that 80% of jobs are never advertised. We traced that one and checked it against 100 live roles. It is fabricated in the same way, and it survives for the same reason — it makes job searching feel like a rigged game that a product can unrig for you.

I can tell you what the systems behind the roles I track actually are, because I poll them directly.

Which ATS you are actually applying through

Here is the real distribution across the 8,317 companies in the ReqHunt catalog:

Two systems account for 96.9% of the companies we track. Neither Greenhouse nor Ashby scores your CV and rejects you automatically. There is no keyword threshold. There is no percentage match that gates your application.

Notice something else about that table: Workday is not on it. Neither is Taleo. Those are the systems that generated most of the ATS horror stories, and they are enterprise products used by large legacy corporations. If you are applying to European startups and scale-ups, you are overwhelmingly applying through Greenhouse or Ashby, which are modern, recruiter-facing tools built in the last decade.

The advice written for 2010-era Taleo is still circulating in 2026. It is aimed at a system most of you will never touch.

What the ATS actually does with your CV

Three things happen, and only one of them is automated in a way that can hurt you. (For the full definition of an ATS and the myths around it, read what an ATS resume is.)

It parses your CV into structured fields. The system reads your uploaded file and tries to extract your name, email, phone number, work history, education and skills into a database. This is the step where a badly formatted CV genuinely causes damage — not by triggering a rejection, but by producing a garbled profile.

It stores your application for a human to review. A recruiter opens a list of candidates. They read, filter, and search. Greenhouse and Ashby both give recruiters powerful search across applications, which is where keywords genuinely matter — not to pass a gate, but to appear in a search.

It applies knockout questions. This is the real automated filter, and almost nobody writes about it. When an application form asks "Are you legally authorised to work in the EU?" or "Do you have 5+ years of experience with Kubernetes?", the answer can automatically disqualify you. That is a hard rule set by the recruiter, applied to a form field you filled in yourself.

Knockout questions reject far more applications than any keyword algorithm. And the advice for handling them has nothing to do with CV formatting.

The real risk: parsing failure

If you take one thing from this article, take this. Your CV does not get rejected for lacking keywords. It gets damaged by formatting that parsers cannot read, and then a human sees a profile that looks empty or scrambled.

Here is what actually breaks parsers, in rough order of how often I see it cause problems:

Multi-column layouts. The parser reads left to right across the full page width, so a two-column CV interleaves your job titles with your skills list. The output is nonsense. This is the single most common cause of a mangled profile, and it is the format most "designer" CV templates use.

Text inside images or graphics. A skills section rendered as a chart, a header rendered as a logo image, a name in a decorative graphic — none of it is text as far as the parser is concerned. It is invisible.

Headers and footers. Many parsers ignore them entirely. If your contact details live only in a header, they may not exist in the parsed profile.

Tables used for layout. Tables holding actual tabular data usually survive. Tables used to position blocks of unrelated content across a page usually do not.

Unusual fonts and heavy styling. Text set in a font that does not embed cleanly can extract as garbage characters.

The fix for all of these is the same, and it is boring: a single-column CV, real text, standard section headings, saved as a PDF from a word processor rather than exported from a design tool.

What "ATS-friendly" should actually mean

Strip away the template-vendor marketing and the practical definition is short.

Use one column. Use real text, not images of text. Put your contact details in the body of the document, not the header. Use conventional section headings — Experience, Education, Skills — because parsers are trained on them. Use a standard font. Give each role a clear job title, company name and date range on their own line, because that is exactly the structure the parser is looking for.

Save as PDF unless the form specifically asks for .docx. Both Greenhouse and Ashby handle PDF well. The old advice to always submit Word documents is another artefact of legacy systems.

That is genuinely most of it. There is no secret formatting trick, because there is no algorithm to trick.

A structure that parses cleanly

You do not need to buy a template. Parsers want predictable structure, and predictable structure is easy to type. Here is the skeleton, which you can copy into any word processor:

FIRST LAST
City, Country | email@example.com | +00 000 000 000 | linkedin.com/in/yourname

EXPERIENCE

Job Title
Company Name | City, Country | Jan 2023 – Present
- What you did and what changed as a result, with a number where you have one.
- Second bullet, same shape.

Job Title
Company Name | City, Country | Mar 2020 – Dec 2022
- Bullet.

EDUCATION

Degree, Field
University Name | City, Country | 2016 – 2020

SKILLS

Named tools and technologies, comma separated, spelled the industry way.

Every element there exists for a parsing reason. Contact details sit in the body rather than a header. Job title, company and dates are on their own lines in a consistent order, which is exactly the pattern parsers look for. Section headings use the conventional words. Bullets are hyphens rather than decorative symbols. There is one column.

It is plain. Plain is the point. Once it parses correctly, visual polish is for the human reader, and the human reader is looking at your work history, not your margins.

Where keywords actually matter

Keywords do matter, but not for the reason you have been told. They matter because a recruiter at a company with 400 applicants will search that pool. If the role is for a data engineer and they search "Airflow", your CV needs to contain the word Airflow to appear.

That is a search problem, not a filtering problem, and it changes the advice substantially.

It means you should use the specific tools, languages and frameworks by name, spelled the way the industry spells them. It means writing both the acronym and the expansion the first time — "Search Engine Optimisation (SEO)" — because you do not know which one they will search. It means the words should appear in the context of real work you did, because a human reads the result of that search and a keyword-stuffed skills block is transparently obvious.

What it does not mean is mirroring the job description word for word. Recruiters read these for a living. A CV that has clearly been find-and-replaced against the posting reads worse than an honest one.

The timing problem nobody mentions

Here is the thing that actually moves your odds, and it has nothing to do with your CV.

Applying early matters enormously. A recruiter reviewing applications in the first days of a posting is reading a shortlist. The same recruiter three weeks later is triaging hundreds and is likely already interviewing candidates. Identical CVs get different outcomes depending on when they land.

This is why ReqHunt exists. We poll company career pages every hour, so roles surface as they open rather than after they have syndicated to the large aggregators. Most of the roles in our catalog never appear on LinkedIn at all, which is a longer argument I have made elsewhere.

You can spend a week perfecting your CV formatting, or you can apply to the right role on day one. The second one is worth more.

What to actually do this week

Open your CV. If it has two columns, rebuild it as one. If your name or contact details are in the header, move them into the body. If anything important is an image, make it text.

Then check the parse. Both Greenhouse and Ashby show you the parsed result during the application flow — the fields they extracted, pre-filled for you to correct. That preview is the most honest ATS test available, and it is free. If it comes back mangled, your formatting is the problem. If it comes back clean, stop optimising and start applying.

Then the work moves downstream. The right ten columns in an application tracker will tell you whether you are losing at the CV stage or somewhere else entirely — which is worth knowing before you rewrite anything again. And if you want to see what this catalog data looks like applied to a single sector, our breakdown of what 2,400 company boards show about AI hiring is built the same way this article is.

Frequently asked questions

Do applicant tracking systems automatically reject CVs?

Not in the way the internet claims. Greenhouse and Ashby, which cover 96.9% of the companies in our catalog, do not score CVs against keywords and reject below a threshold. What does automatically disqualify applicants are knockout questions on the application form itself — work authorisation, years of experience, location requirements. Those are answers you type in, not anything the parser extracts from your CV.

Is a 72 ATS score good?

There is no ATS score. Greenhouse and Ashby do not assign your CV a number, and no recruiter sees one. The scores you have seen come from third-party CV-checking tools, which invented their own scoring model and then market it as though it reflects what employers see. A 72 on one of those tools means you scored 72 on that tool. It predicts nothing about your application.

What is the best CV format for ATS?

Single column, real text, conventional section headings, contact details in the body, saved as PDF from a word processor. That is the whole answer. Format arguments beyond those points are mostly aesthetic preferences dressed up as technical requirements.

How do I convert my CV to ATS-friendly?

Rebuild it as one column, move anything from the header or footer into the body, replace any image or graphic that contains text with actual text, and use standard headings. Then upload it to a real Greenhouse or Ashby application and look at the fields it extracts. That preview is the only conversion check worth trusting.

Should I submit a PDF or a Word document?

PDF, unless the form explicitly asks for something else. The old advice to always use .docx dates from legacy systems with weak PDF parsers. Modern systems handle PDF fine. What matters far more is that the PDF was produced by a word processor rather than exported from a design tool, because design exports frequently render text as vectors or images.

Is a two-column CV really that bad?

For parsing, yes. Most parsers read across the full width of the page, which interleaves the content of your two columns into unreadable output. A human reviewing the parsed profile sees a mess. If you love your two-column design, keep it for direct email applications and maintain a single-column version for online forms.

How many keywords should I include?

There is no target number, because there is no score to hit. Include the specific technologies, tools and methods you actually used, named the way your industry names them, in the context of the work you did. The purpose is to appear when a recruiter searches their applicant pool, not to satisfy an algorithm.

Does using a resume-scanning or ATS-checker tool help?

Those tools test against a model of ATS behaviour that mostly does not match how Greenhouse and Ashby work. The free and accurate test is to start a real application on one of those systems and look at the parsed fields it shows you before submission. That is the actual parser, on your actual CV.

Do I need a different CV for every application?

No, and the returns on doing it are lower than the returns on applying early. Keep one well-formatted CV and adjust the top few lines to match the role. Time saved there is better spent finding roles in their first days rather than their third week.