How to Write a Resume That Gets Past Applicant Tracking Systems

Almost every piece of advice you have read about applicant tracking systems is either wrong or unfalsifiable. The most repeated claim — that ATS software automatically rejects 75% of resumes before a human sees them — has no verifiable source behind it. It gets cited constantly, usually by companies selling resume scanning tools.

Here is what an applicant tracking system actually is: a database. It stores applications, lets recruiters search and filter them, and tracks candidates through stages. It is closer to a CRM than to a robot gatekeeper. Very few configurations auto-reject anyone, and where auto-rejection exists it is almost always tied to explicit knockout questions the employer set up — “Are you legally authorised to work in this country?” — not to a secret formatting score.

That does not mean formatting is irrelevant. It means the risk is different from what you were told. The real risk is that the parser mangles your resume into an unreadable database record, and the recruiter searching that database never surfaces you. You are not rejected. You are invisible, which has the same effect and is much harder to notice.

What happens to your resume after you click submit

When you upload a file, the ATS runs it through a parser. The parser’s job is to turn an unstructured document into structured fields: name, email, phone, work history with employers and dates, education, skills. Those fields populate your candidate record.

Parsing is genuinely hard. A resume is a visual document; the parser has to infer meaning from layout. It looks for patterns — a line that looks like a date range, a heading that looks like a section title, text positioned as a job title above a company name. When your layout matches the patterns it expects, it works. When it does not, fields come back empty or scrambled.

Then a recruiter opens the system and searches. In practice that search is usually a keyword query against the parsed record and the raw text: “registered nurse” AND “ICU”, or a filter on years of experience, or a search for a specific certification. Candidates who match appear. Candidates whose data did not parse do not.

So the actual failure mode is boring and mechanical. Your five years at a company got parsed as zero because you wrote the dates in a format the parser did not recognise. Your job titles landed inside a text box the parser skipped. Your skills section was a graphic. Nothing rejected you. You simply are not in the result set.

The formatting decisions that actually cause damage

Text inside images, graphics, or charts

This is the one unambiguous killer. If your skills are shown as a bar chart, your name sits inside a designed header image, or your job titles are part of a graphic, that text does not exist as far as the parser is concerned. Everything on your resume must be selectable text. Open your PDF, try to select a line with your cursor, and if you cannot highlight it, the system cannot read it.

Headers and footers

Many parsers ignore the header and footer regions of a document entirely, because in most documents those contain page numbers and boilerplate. If your contact details live in the header, they may vanish. Put your name, email, and phone in the body of the document, at the top of page one.

Tables and text boxes

Tables are read inconsistently. Some parsers read them cell by cell in the wrong order, turning a clean two-column layout into interleaved nonsense: “Software Engineer Marketing Assistant Acme Corp Beta Ltd 2021–2024 2019–2021.” Text boxes are frequently skipped altogether. Neither is worth the risk. Use ordinary paragraphs and bullet lists.

Multi-column layouts

The two-column resume — a narrow sidebar for skills and contact details, a wide column for experience — is the most common design that causes trouble. Parsers read in one pass. Some handle columns correctly; some read straight across, merging your sidebar into your job descriptions. A single-column layout is not a stylistic preference here, it is insurance.

Unconventional section headings

The parser identifies sections by matching headings against a known list. “Work Experience,” “Professional Experience,” and “Employment History” are all recognised. “Where I’ve Made An Impact” is not. Your creative heading may cause the entire section beneath it to be filed as uncategorised text, which means your job history never populates the structured fields recruiters filter on.

Date formats

Write dates as Month Year – Month Year (“March 2021 – June 2024”) or MM/YYYY. Avoid seasons (“Fall 2021”), avoid years alone if you can, and be consistent across every entry. Inconsistent formats within one document are what confuse parsers most, because the pattern it locked onto in your first job stops matching in your third.

File type

Submit a .docx or a text-based .pdf. Modern parsers handle both well. The one file to never submit is a scanned or image-based PDF — a photo of a document rather than a document. If the job posting names a format, use that format; the employer knows their own system.

Being findable: how to think about keywords

Since a recruiter finds you by searching, the practical question is: what will they type? Almost always, they type words from their own job description, because that is the document in front of them.

This is where the standard advice — “include keywords from the job description” — is correct but incomplete, and where people go wrong by dumping a keyword list at the bottom of the page in white text. Keyword stuffing does not work and looks dishonest to the human who eventually reads it. What works is making sure the vocabulary the employer uses actually appears in your genuine experience.

The extraction exercise

Take the job description and pull out three categories:

• Hard requirements: named tools, certifications, languages, qualifications. “Salesforce,” “CPA,” “Python,” “RN licence.”
• Function words: what the role does. “Forecasting,” “onboarding,” “incident response,” “stakeholder management.”
• The employer’s dialect: the same job is called different things in different places. If they say “client success,” do not only say “account management.” If they say “paid media,” do not only say “PPC.”

Then check which of those you have honestly done, and make sure the words appear in the bullet points describing when you did them — not in a keyword appendix.

The second version contains almost the same terms. It also survives a human reading it, which the first does not.

Spell out abbreviations, once

A recruiter might search “SEO” or “search engine optimisation.” A search for one does not return the other. Write the long form with the abbreviation in brackets the first time it appears — “search engine optimisation (SEO)” — and use the short form afterwards. Do this for certifications too: “Project Management Professional (PMP).”

Match the job title, where it is honest to

Internal job titles are often meaningless outside the company. If your title was “Growth Ninja” and the work was digital marketing, write Digital Marketing Manager (internal title: Growth Ninja) or simply use the functional title. You are translating, not lying. What you cannot do is promote yourself — calling yourself a Senior Manager when you were a Coordinator is a misrepresentation that will surface in reference checks.

The structure that parses cleanly

There is no secret template. There is a conventional order that every parser has been trained on:

1. Name and contact details in the body, top of page one. Name, phone, email, city and country, LinkedIn URL. No photo in most markets — check local convention.
2. A short summary of three or four lines, if it says something specific. Skip it if it would only say “results-driven professional seeking opportunities.”
3. Work experience, most recent first. For each: job title, employer, location, dates, then bullet points.
4. Education, most recent first. Qualification, institution, year.
5. Skills, as a plain comma-separated or bulleted list of real, checkable competencies.
6. Certifications, licences, languages where relevant to the role.

Each experience entry should be laid out so the relationship between title, employer, and dates is unambiguous on a single line or two adjacent lines:

Operations Analyst
Beta Logistics, Manchester, UK
March 2021 – June 2024

Bullet points that survive both readers

Your bullets have to work for a keyword search and for the human who reads them eight seconds later. That means specific verbs, real numbers, and the vocabulary of the field.

The rewrite added searchable platform names, a time frame, and two metrics. It did not add length for its own sake — it replaced “responsible for” with what actually happened. If you do not have exact numbers, use honest approximations and say so: “roughly 40 tickets a week,” “a team of 6–8 depending on season.”

Test your own resume in five minutes

You do not need a paid scanning tool. Two checks catch nearly everything:

The plain text test

Open your resume, select all, copy, and paste it into a plain text editor — Notepad, TextEdit in plain text mode, or any online plain-text box. Strip all formatting. Then read what you get.

This is approximately what the parser sees. Look for: text that disappeared entirely (it was an image), columns that interleaved, dates separated from their jobs, bullet characters that turned into junk symbols, and your contact details missing (they were in the header). Anything broken here is broken in the database too.

The eight-second test

Give the document to someone who does not know your job. Ask them, after eight seconds of looking, to tell you what you do, roughly how senior you are, and where you last worked. If they cannot, the layout is failing the human half of the audience regardless of how well it parses.

What actually gets you filtered out

Since we started by dismantling the auto-rejection myth, it is worth naming the things that genuinely do knock applications out:

• Knockout questions. Work authorisation, willingness to relocate, minimum qualification, licence held. These are explicit and answered in the application form, not on your resume. Answer them accurately — a false answer here is grounds for withdrawal of an offer later.
• Missing mandatory fields. Half-completed applications get filtered or deprioritised. Yes, retyping your history into the form after uploading the file is infuriating. Do it anyway.
• Genuinely not matching. Sometimes the honest answer is that the recruiter searched for a certification you do not hold. No formatting fixes that.

A note on resume-scoring tools

Tools that give your resume a “match score” against a job description are running their own keyword comparison, not the employer’s actual system. They are useful for one narrow purpose: spotting vocabulary from the job description that is genuinely missing from your experience. They are not useful as a target to optimise. Chasing a 90% score usually produces a resume stuffed with borrowed language that reads as though nobody wrote it, because effectively nobody did.