Why two checkers give the same résumé two different scores
One tool says 62, another says 88, and neither will tell you why. A score is only worth having when you can check it.
— ask for the arithmetic, then decide what the number is worth.
Run the same file through three tools and you will get three numbers, sometimes twenty points apart. That is not a malfunction. It is what happens when several products measure different things and all call the result a score.
Three places the disagreement comes from
The extraction, first. Each tool reads your file with a different library, and a two-column layout or a table can come out in a different order in each. Two tools working from two different readings cannot agree about what is in the document.
The rules, second. One tool counts keyword overlap with an advert. Another counts sections. Another grades achievement writing. These are different measurements wearing the same word.
The weights, third. Even where two tools measure the same things, a tool that treats formatting as half the score and one that treats it as a fifth will produce different numbers from identical findings.
What is missing from almost all of them
The working. A number with no arithmetic behind it cannot be argued with, cannot be reproduced, and cannot tell you which of its parts moved. The ATS checker here prints the multiplication, names the algorithm string the engine returned, and shows both figures when a ceiling applies. That is the whole design intent, described in why the checker prints its own arithmetic.
The unanswerable question
People want to know which tool is right. There is nothing to be right about: no checker has access to an employer's software, so there is no external score to be measured against. Any tool claiming to predict a named system's behaviour is describing a guess with a confident interface, which is the subject of what an applicant tracking system is.
What to do with several numbers
Ignore the scores and read the findings. If three tools all say your experience section was not detected, that is a real signal about extraction and worth an hour. If they disagree about whether you need a summary, that is a matter of taste dressed as a measurement.
The one thing they mostly agree about
Structure. A single-column document with plain headings, contact details in the body, no tables and a clean text layer extracts well everywhere, because every extractor is trying to do the same job. That is the argument in the ATS format guide, and it is why fixing the structure is the change that moves every tool at once.
What to do when two tools contradict each other
Read both sets of findings and look for the overlap. Anything two tools agree on is almost certainly about your document rather than about their rules: a section that was not detected, contact details that did not extract, a date format that changed halfway down. Anything only one of them says is a rule choice, and you can weigh it on its merits rather than on its confidence. Averaging the two scores is the one response with no value at all, because the numbers are measurements of different things and their mean measures nothing.
What this page does not claim
That our score is more accurate than anybody else's. Accuracy needs a truth to compare against and there is not one. What is claimed is narrower: ours is checkable. Every finding names its rule, the sum is printed, a ceiling is declared when it applies, and the improvements are priced so you can tell which one is worth doing. That makes it useful as a measurement of a document even where it would be worthless as a prophecy.
A last note on scores in general, since it applies to any of them. A high number is a floor cleared rather than a verdict earned. A four-minute invented résumé in this estate scores 97, which tells you that the measurement is about legibility and evidence rather than about a career. The fuller version of that argument, with the numbers, is in what a score cannot tell you and the honest limits of every résumé tool.
One consequence for anybody choosing which tool to trust. Prefer the one that tells you the most about how it reached its number, even where its number is less flattering. A lower score with named rules and priced fixes is actionable; a higher score with a list of generic tips is a mood. That preference also protects you from the tools that are most confident, which in this category tend to be the ones making claims about employer software they have never seen.
Get a score with its working attached
Three weighted dimensions, the multiplication printed, and both figures when a ceiling applies.
Open the ATS checkerWhat a score cannot tell youQuestions
- Why do the numbers differ so much
- Because each tool extracts the text with a different library, applies different rules, and weights them differently. None of those choices is standard, and most tools do not publish theirs.
- Which one is right
- The question does not have an answer, because there is no external truth to compare against. No checker has access to an employer's system.
- What should I look for in a checker
- Whether it prints the arithmetic, whether it names the rule behind each finding, and whether it tells you what a fix is worth. A bare number is a rating.
- Does a job description change your score
- No. This check never takes one, so the same document scores the same whichever advert you are answering.
- Should I run several checkers
- You can, and you will get several numbers. Reading the findings is more useful than averaging the scores.
Conxfolio is a free set of four career tools: a résumé builder with 37 rendered layouts, an ATS résumé checker that prints its own arithmetic, a cover letter builder that traces every proof paragraph back to the line it came from, and a portfolio builder with 20 authored designs. There is no account to create, nothing is held back for a paid plan, and no language model is used anywhere in the product, so the readers, the score and the letter are deterministic code you can check.