What an ATS actually does, and how to get past it without tricks

What an ATS is

An ATS (Applicant Tracking System) is the software a company uses to receive and sort applications. Workday, Greenhouse, SuccessFactors, Teamtailor and twenty others. It is not a robot that scores people: it is a database with a search box on top.

That matters because it changes the question. The question is not how to please an algorithm. It is: when a recruiter searches that database, will I show up, and will what shows up make sense?

What it does do

• Extracts the text from your file and files it into fields: name, experience, education, skills.
• Lets people search by keyword. Someone types SQL or team management and sees who has it.
• Sorts on simple criteria: years of experience, location, answers to the form questions.
• In some companies, auto-rejects on knockout answers: right to work, availability, a required language.

What it does not do

There is no secret score out of 100. The big ATS platforms do not grade your CV, and where a feature like that exists almost nobody uses it, because the results are nonsense.

It does not read your photo or your design. It could not care less whether the CV is beautiful. Beauty matters later, when a person opens it.

It does not reward repeating a keyword twenty times. These search engines check whether the term is there, not how often.

Where a CV really breaks

The expensive failure is structural, not lexical. If the parser cannot pull the text out in the right order, half your experience lands in the wrong field or does not land at all.

• Tables. A CV laid out with an invisible table gets parsed column by column and the chronology comes out scrambled.
• Text boxes and floating columns. Same problem: the file's reading order is not the visual order.
• Details inside an image. A phone number drawn into an icon does not exist as far as the parser is concerned.
• Headers and footers. Several parsers ignore them, and that is exactly where many people put their email address.
• Fonts with unusual ligatures, or icons used as text. You get invented characters in the middle of a word.

What to do, in order of payoff

• One column, running text, a PDF generated from text rather than scanned. That alone fixes most real problems.
• The job title you are applying for, written the way the ad writes it. If the ad says Digital Marketing Manager and your CV says Marketing Lead, a search for the ad's term will not find you.
• Tools and skills under the exact name the industry uses. Google Ads, not paid search campaigns.
• The ad's own terms, but only the ones that are true. This is the part that matters.
• One file per role. Tedious by hand, which is exactly the work worth automating.

The line worth not crossing

Adding a keyword for something you have not done does not raise your odds, it lowers them. You clear the first filter and fall apart in the first interview, with the company burned and fifteen minutes wasted on both sides.

The white-text-on-white-background trick, which still circulates, has been detectable for years and in several companies is grounds for immediate rejection. This is not a moral point. It does not work.

The honest way to raise coverage is different and pays better: you have usually already done what the ad is asking for, and wrote it in different words. Turning handled the supplier relationship into supplier management is not lying, it is using the industry's vocabulary.

How much of this changes the outcome

Less than course sellers promise, and more than you would think. A well-structured CV does not win an interview on its own. A badly structured one never gets read.

What genuinely moves the needle is applying to fewer roles, better. Thirty identical applications perform worse than eight tailored ones, and take longer.