Are ATS scores accurate? What the number can and cannot know

Every resume checker, ours included, hands you a number. Jobscan gives a match rate. Resume Worded gives a score out of 100. We give six dimension scores. The first thing to understand about all of them is the same.

No employer ever sees your score. No applicant tracking system imports it. There is no threshold at which you are through.

Each tool invented its own number, measured its own way. Two tools scoring the same resume will disagree, and neither is wrong, because they are not approximating one true hidden value that a real system computes. There is no such value.

What a score can actually measure

Quite a lot, and these parts are worth trusting, because they are checkable facts rather than judgements.

• Whether the file parses. Did your name, email and phone survive extraction? Did your job titles merge into your dates? This is binary, verifiable, and the most useful thing any tool tells you.
• Which expected terms are absent. Your field has a vocabulary. Comparing your document against it is straightforward and the answer is not a matter of opinion.
• Structural facts. Section headings a parser recognises, dates in a readable format, contact details in the body rather than a header. All checkable.
• Measurable writing problems. Bullets with no numbers, lines starting with responsible for, a summary that could belong to anyone. Countable, and genuinely predictive of how a reader reacts.

What it cannot know

This is the honest half, and it is the half the marketing tends to skip.

1. 1.Whether you are actually a good fit. A tool reads a document. It does not know that the team needs somebody who has survived a migration like the one in your second job.
2. 2.How many other people applied. The single biggest factor in whether you hear back, and completely invisible to any scorer. Several hundred applicants beats any score.
3. 3.What this specific recruiter cares about. Two people hiring for the same title weigh things differently, and neither published their weighting.
4. 4.Whether your claims are true. A score rewards a quantified bullet. It cannot tell an honest 40% from an invented one, which is one reason chasing the number is a bad idea.
5. 5.The thing that actually gets you hired. A referral, a portfolio somebody loved, a conversation at a meetup. None of it is on the page being scored.

How to read your score without being misled

The before-and-after column is where a score earns its place. The absolute number means little; the same tool scoring your resume before and after an edit tells you whether the edit helped. That comparison is internally consistent even when the number itself is arbitrary.

The point where chasing it goes wrong

There is a threshold past which optimising the score makes the resume worse. It arrives when you start adding terms because the tool wants them rather than because they are true, padding bullets with numbers you cannot source, or restructuring a clear document to please a checker nobody at the company will ever run.

A reasonable rule: act on everything the tool tells you that you agree with on reflection, and ignore the rest. If it flags a line and you can explain why the line is right, the line is right. You know your work and it does not.

And the statistic underneath all of this

Most scoring tools open with the claim that software automatically rejects around three quarters of resumes. It is worth knowing that the number traces to a 2012 sales pitch by a company that shut down in 2013, with no study behind it, and that surveyed recruiters overwhelmingly say their systems do not auto-reject at all.

That matters here because the urgency around scores is largely built on it. Your resume is not being binned by a machine. It is competing for a few seconds of a tired person's attention, which is a real problem that a good score genuinely helps with, just not the dramatic one you were sold.