Does software really reject 75% of resumes?

Search for resume advice and you will meet the same sentence within about thirty seconds. Seventy five percent of resumes are rejected by software before a human ever sees them. Sometimes it is 70%. Sometimes 98%. The number moves around, which is the first clue.

It is worth knowing where it came from, because once you know, a lot of popular advice stops making sense.

Where the number came from

The trail leads back to around 2012 and a company called Preptel, which sold a service for getting resumes past screening software. The figure appeared in their marketing. No study was published with it. No method, no sample, no definition of what counted as a rejection.

Preptel shut down in 2013. The number outlived them. An article quoted the pitch, a second article quoted the first article, a third quoted the second, and after a decade of that it reads like something everybody has always known.

A statistic that has been repeated for fourteen years and sourced for none of them is not a statistic. It is a rumour with a decimal point.

What recruiters say their software does

When people have gone and asked recruiters directly, the picture is different. In one survey of recruiters in the United States, the large majority said their system does not automatically reject anybody on formatting or content. A small number said theirs was configured to, which is worth remembering before you relax completely.

That survey was small, and it was run by a company that sells resume tools, so it has its own reasons to want that answer. Treat it as a correction to the 75% claim rather than as the final word. The honest summary is that nobody has a trustworthy number, in either direction.

What the software is built to do is less dramatic. It reads your file into fields, stores it, lets a recruiter search and filter, and sorts the results. The deciding is still done by a person. The sorting is what decides whether that person ever scrolls far enough to find you.

So why do applications vanish?

Because three other things are true, and all of them feel identical to being rejected by a machine.

There are hundreds of other applicants

A desirable role can collect several hundred applications in a few days, and sometimes a few thousand. A recruiter is not reading all of them. They search, they filter, they read the first page or two of results. If you are on page nine, nothing rejected you. Nobody got to you.

Your file did not read properly

This is the one you control. If your contact details sit in a header, many parsers skip them. If your work history is in a table, the cells can collapse into one unreadable string. If your resume is a picture of a resume, there is no text in it at all. You are then in the database as a half empty record that matches nothing. See how to format a resume so software can read it for the specifics.

You failed a question, not a scan

Plenty of application forms ask things like whether you can work in the country without sponsorship, or whether you have a particular licence. Answer one of those the wrong way and you are filtered out before anything reads your resume at all. That really is automatic, and it has nothing to do with your formatting.

What this changes about the advice you follow

A lot of popular tips exist to defeat a robot that mostly is not there. Once you drop the robot, the tips sort themselves into two piles.

The useful version of the worry

The fear behind the 75% claim is reasonable even though the number is not. Your resume really can disappear without a human reading it. It just happens for duller reasons than a machine judging your font.

So the fix is duller too. Make the file readable. Use the words your field actually uses. Put the thing you want noticed near the top, because attention runs out long before the page does. Then apply to roles you have a real case for, because no amount of formatting beats being the wrong candidate.