Field Notes / Resume
ATS Resume Checkers: What They Catch and What They Miss
The score looks like a verdict. It is a word-overlap calculation against a job description. Useful for one narrow job, misleading for every other.
A number no applicant tracking system produced, for a test no employer ran.
The score is not a verdict
Paste a resume and a job description into a checker and you get a percentage. It feels like a grade from the system that decides your fate. It is nothing of the sort.
No applicant tracking system generated that number. No employer ran that test. A third-party tool compared two documents and counted overlap. The output is a similarity measure wearing the costume of a verdict.
What they get right
Two things, and both matter.
- Parse quality. A checker reads your file the way a parser does and shows you the result. If your job titles come back scrambled, you found a real defect. This is the strongest reason to run one.
- Format defects. Tables, columns, text boxes, headings the parser fails to recognize. Mechanical problems with mechanical fixes.
Run a checker once for these two. Fix what it flags. That work is finished in an hour.
What they miss entirely
Everything a person weighs.
- Whether your evidence is any good. A bullet with a real result and a bullet describing activity score the same if they share vocabulary.
- Seniority and scope. Leading a team of twenty and attending a meeting about a team of twenty use similar words.
- Trajectory. A recruiter reads your last three titles as a story. No overlap score sees a story.
- Fit. Whether you suit this team, this stage, this problem. That is a judgment, and it belongs to a human.
Why the number seduces
Job searching offers almost no feedback. You send applications into silence. A checker hands you a number that moves when you act, and that is the only responsive signal in the whole process. So people work the number, because working the number feels like progress while the silence continues.
The percentage is measurable. The thing that decides your outcome is not. Those are different problems, and the measurable one steals attention from the real one.
How to use one properly
- Run it for parsing. Confirm your file reads cleanly. Fix defects.
- Read the missing keywords as a prompt. If a term is missing and you have done that work, add it with the result attached. If you have not done the work, leave it out.
- Ignore the headline percentage. Do not chase it. Do not compare it between roles.
- Spend the saved time on your top third. That is where a human decides.
Use a checker as a spellchecker for formatting. Then close it. The percentage measures resemblance between two documents. Nobody hires resemblance.