ATS resume guide: what the software actually does

What an ATS actually is

An applicant tracking system is recruitment software. Its core job is unglamorous: receive applications, pull the text out of each file, store it in structured fields, and let recruiters search and track candidates through stages. Workday, Greenhouse, Lever, Taleo, iCIMS and SAP SuccessFactors are all examples.

The step that affects you is parsing. When you upload a file, the system extracts the text and tries to sort it into fields — name, email, phone, employers, job titles, dates, education, skills. It does this using structural cues: recognisable section headings, consistent date formats, and a predictable reading order.

Then a recruiter searches. They might filter for a job title, a location, a qualification, or a specific skill, and read the results. That is the actual mechanism, and understanding it changes what you should optimise for. Your goal is not to trick a filter. It is to be correctly indexed, so that when a recruiter searches for what you can genuinely do, you are in the list.

The 75% rejection myth

You have probably read that applicant tracking systems automatically reject 75% of resumes before a human sees them. The figure gets repeated everywhere and it does not describe how these systems are configured in practice.

Most systems do not auto-reject on the strength of a resume at all. Where automatic screening does exist, it is usually applied to explicit application questions — work authorisation, willingness to relocate, a required licence, minimum years of experience — which you answer in a form, not in your resume. Those are knockout questions, and they are the honest kind of filter: you either meet the requirement or you do not.

Why this matters: if you believe a robot is scoring your prose, you write for the robot, and writing for the robot produces keyword-stuffed documents that repel the human who actually reads them. The realistic risk is more mundane and more fixable — your file parses badly, your details land in the wrong fields, and you never surface in the searches that would have found you.

How parsing really fails

Parsers fail in specific, predictable ways. Nearly every real problem we see falls into one of these.

Text that is not text

If your resume is an image — a scan, a photograph, or a design exported as a flat picture inside a PDF — there is nothing to extract. This is the single worst case, and it is silent: the upload succeeds and your application contains no searchable information whatsoever. Always export from a text-based source.

Content in headers, footers and text boxes

Many parsers skip header and footer regions, and some ignore floating text boxes. This matters most because those are exactly where people put contact details. A resume that parses cleanly but has no phone number is worse than one with a plain layout.

Reading order in multi-column layouts

A parser reads the document's underlying content order, which is not always the visual order. In a two-column design, that can interleave the columns — your skills sidebar ending up spliced into the middle of a job description. Modern parsers handle many two-column layouts correctly, but the failure is invisible to you, which is why single-column is the safer default for online applications.

Tables for layout

Tables used to position content are a common source of scrambling. Cell-by-cell extraction can separate a job title from its dates or merge unrelated entries. A simple table for a skills grid is usually survivable; a resume whose entire structure is a table often is not.

Unconventional section headings

The parser is looking for recognisable labels. "Experience", "Work Experience", "Employment History", "Education", "Skills" are all understood. "My Journey", "Where I've Made an Impact" and "The Story So Far" are not — and a section the parser cannot classify may be dropped from the structured fields entirely.

Graphics carrying information

Skill rating bars, donut charts, icon-only contact details and timeline infographics contain no extractable text. If a five-star graphic is the only place you mention Python, then as far as the system is concerned you have not mentioned Python. (These also tell a reader nothing verifiable — see the skills section of our resume guide.)

Formatting rules that matter

Short version of everything above, in the order it is worth caring about.

That last one is worth expanding. Write "Search Engine Optimisation (SEO)" the first time, then use the acronym. A recruiter searching either term finds you. The same applies to "Chartered Accountant (CA)" and "Amazon Web Services (AWS)".

Keywords, done honestly

Keywords matter, but not in the way the phrase "keyword optimisation" suggests. You are not trying to hit a density target. You are trying to describe your real experience in the words the employer uses, so that their search finds it.

The method is straightforward:

1. Read the posting and list the skills, tools and responsibilities it names.
2. Mark the ones you genuinely have.
3. For each of those, make sure it appears in your resume — in the employer's vocabulary, inside a real bullet describing real work.
4. Leave the rest out.

The vocabulary point is the one people skip. If the posting says "stakeholder management" and your resume says "worked with clients", you have the experience and will not be found. Adopt their phrasing where it honestly describes what you did.

Stuffed: Skilled in project management, project planning, project coordination, project delivery, project execution, and project lifecycle management.

Natural: Managed the delivery of a six-month CRM migration across three teams, from planning through go-live, including vendor coordination and stakeholder reporting.

The second version contains more genuinely useful keywords than the first, in context, and a human can read it without wincing. That is the target.

Tricks that backfire

Every one of these circulates as clever advice. Each is a bad idea, and it is worth knowing why.

• White or 1pt hidden keyword text. The parser extracts it and shows it to the recruiter as normal text. The trick is invisible only to you. Recruiters who spot it read it as deliberate deception.
• Pasting the whole job description into the document. Same outcome, more obvious.
• Keyword lists in the file's metadata. Some parsers read metadata, most ignore it, and no recruiter is impressed by it.
• Claiming skills you do not have to clear a filter. This works exactly as far as the phone screen, then costs you the relationship with that recruiter for future roles.
• Paying for an "ATS score". There is no standard score. Vendors parse and rank differently, and no third party can tell you how a specific employer's configuration will treat your file.

The reason all of these fail is the same: they optimise for a gatekeeper that mostly does not exist, at the cost of the human who does.

Test your own resume in one minute

You do not need a paid tool. This approximates what a parser sees:

1. Open your exported resume — the actual file you would send.
2. Select all the text and copy it.
3. Paste it into a plain text editor (Notepad, TextEdit in plain mode, or any code editor).

Now read what you get, and check:

• Did any text come through at all? If not, your file is an image and no system can read it.
• Are your name, email and phone number present?
• Are the sections in the order they appear visually, or have columns interleaved?
• Did every job title keep its employer and dates alongside it?
• Did the section headings survive?
• Are there strings of odd characters where bullets or icons used to be?

If the plain text version reads sensibly and contains everything important, your resume will parse. If it does not, you have found a problem worth fixing before your next application — and you have found it in a minute, without guessing.

ATS checklist

• Exported as a text-based PDF, not an image or a scan
• Contact details in the body, not in a header or footer
• Single-column reading order
• Conventional section headings
• Consistent MMM YYYY dates throughout
• No information carried only by a graphic, chart or rating bar
• Acronyms spelled out once alongside the short form
• Terms from the posting used where they honestly apply
• No hidden text, no pasted job description, nothing you would not defend in an interview
• Copy-paste test passed

Where to go next

Formatting is only half of it — a perfectly parseable resume still has to be worth reading. The resume writing guide covers what to put in each section, resume tips covers line-level edits, and resume best practices covers the conventions around length and structure. Our shorter ATS walkthrough is a quicker read if you want the essentials.

Frequently asked questions