AS ATS Score

Home/Reading your ATS result

How to read your ATS result

You ran a check and got a number. It is not a pass mark, it is not what the employer's system decided, and it is not comparable to the number a different tool gave you last week. This page explains what the figure is made of, what each band means, and which part of it you should actually act on.

Published 9 November 2026. Every weight quoted here is the one in the code that produces the score — if a weight changes, this page changes with it.

Read this first: the arithmetic below is ours. ATS Score is a product, and this is how the product counts. What transfers to any other checker is the shape of the problem — that a score mixes parse quality with keyword match, and that the deductions matter more than the total. What does not transfer is the exact number.

What the number is, and what it is not

Almost no applicant tracking system shows a candidate a score. Workday, Greenhouse, Lever and Taleo parse your file into fields and let a recruiter search and filter those fields; whether a given employer ranks by keyword or reads in upload order is a configuration choice nobody outside that company can see. So there is no official score to look up.

A checker — any checker, including this one — is a simulation of the first half of that pipeline: can the file be read, and does the readable text contain the things the posting asks for. That is two different measurements wearing one number, and it is worth pulling them apart before you interpret anything.

Parse quality is mechanical. Either the text survived extraction or it did not, either the date formats agree or they do not. It is objective, it is the same for every employer, and it is entirely fixable by you. Match quality is comparative. It only exists relative to one specific job description, and it changes the moment you apply for a different role.

If you ran an ATS result checker elsewhere first and the number looked nothing like this one, that is expected rather than alarming — section 5 covers the usual reasons.

How 21 rules become one number

Each rule produces a sub-score from 0 to 100. That sub-score is multiplied by the rule's weight and divided by 100, which gives the points that rule contributes:

Points = sub-score × weight ÷ 100

Score = (sum of points ÷ sum of applicable weights) × 100, rounded to a whole number.

The denominator is the part people misread. It is not always 101. When no job description is supplied, the four job-match rules are skipped rather than scored as failures, and the maximum available drops from 101 to 71. A mechanically perfect resume scores 100 either way — what changes is which failures are even possible.

Where the 101 points sit
GroupRulesPoints
Parsing1–410
Structure5–920
Bullet content10–1324
Job match14–1730
Detail and risk18–2117

Two things follow from that table. Parsing is only 10 of 101 points, so a file that reads perfectly can still score in the sixties on the strength of its bullets alone. And job match is the single largest block at 30 — which is why the same resume can move by twenty points between two applications without a single word of it changing.

A worked example, using no job description: sum of points 59.4, sum of applicable weights 71. 59.4 ÷ 71 × 100 = 83.66, which rounds to 84. Every weight, and the full arithmetic, is on the methodology page — you can reproduce that 84 by hand.

The bands, and what to do in each

These cut-offs are ours, not an industry standard, and no employer publishes a pass mark. They are calibrated to what the rule set can actually detect: below 50 something structural is broken, and above 85 the remaining points are mostly about wording.

BandWhat it usually meansDo this first
85–100 The file reads cleanly and the bullets carry evidence. Remaining losses are small and spread out. Stop optimising the file. Tailor wording to the specific posting instead.
70–84 Readable, with two or three rules losing meaningful points. Often quantification or date consistency. Open the deduction list and fix the two highest-weight losses.
50–69 Either a layout problem or a thin bullet set. Both are common and both are fixable in an evening. Check rule 1 and 2 first: if extraction is wrong, every other score is noise.
Below 50 Usually an image-only PDF, a two-column template, or contact details the parser never saw. Re-export the file single-column before touching any wording.

Read the deductions, not the total

Two resumes that both score 84 can need opposite work. The total hides which rules took the hit, so the list underneath is the actual report and the number is only its summary.

Consider a file at 86 that loses all four points of rule 1 — is there any text at all. That pattern means extraction returned almost nothing and the rest of the sub-scores were computed from scraps. The high total is arithmetic, not evidence: an image-only PDF can look fine because the few lines that did survive happened to be clean ones.

Now consider a file at 74 where the entire loss sits on rule 11, quantified results. Extraction was perfect, headings parsed, dates consistent, keywords present — the resume simply claims responsibilities without numbers. Nothing about the file needs rebuilding. Three edited bullets would close most of that gap.

Same distance from 100 in neither case, and yet one file needs a re-export and the other needs an afternoon of writing. If you only remember one instruction from this page: fix by weight, not by total. The rule worth 12 points moves the number four times as far as the rule worth 3.

Why another checker disagrees

Because they are usually measuring different things. A job-match tool compares your resume to one job description and reports the overlap — without a description it has nothing to divide by, so it either refuses to score or scores something else. A parse-focused checker measures whether the file can be read at all, and will happily score a resume that matches no particular job.

This is why an 84 here and a 52 on a subscription tool are not a contradiction, and why neither number can be compared across tools month to month. It is also why a checker that reports one number without telling you its denominator is not giving you much to act on — the denominator is the whole interpretation.

The longer comparison, including which parts of a subscription product you actually get for the money, is on the Jobscan alternative page.

What changes when you add a job description

The denominator goes from 71 to 101, and rules 14 to 17 switch on: hard-skill keyword coverage (weight 12), job-title alignment (8), industry terminology (6), soft-skill match (4). Thirty points that were previously excluded now have to be earned.

So the score usually falls. A resume that read 88 without a description might read 71 with one, and nothing about the resume got worse — it is being measured against thirty points of criteria that did not exist a minute ago. The drop is information, not damage: it tells you how much of the posting's language is missing.

Which is also the reason a checker that scores without asking for the posting can only ever tell you about the mechanical half. If your goal is a specific application, supply the description.

A high score is not an interview

A score of 100 says the file is readable and the bullets are evidenced. It says nothing about whether you can do the job, whether the recruiter has already filled the role, whether an internal candidate exists, or whether a human reads your summary and finds it generic.

What this measures is the layer that silently removes people before any of that: files that cannot be parsed, sections that cannot be found, dates that cannot be compared. That layer is worth clearing, and clearing it does not guarantee anything downstream. We would rather say that plainly than imply a number can promise a callback.

For the rule-by-rule mechanics behind each deduction — what a parser does with a two-column layout, why a header contact block disappears, which headings it recognises — see what an ATS actually checks.

What a re-run tells you

Change one class of thing, then run it again. If you rewrite three bullets and re-export the file in the same pass, a movement of eleven points cannot be attributed to either change, and you have learned nothing you can repeat on the next application.

A score that does not move after an edit is also a result: it usually means the edit was outside these 21 rules. Adding a personal statement full of soft skills, for example, moves nothing unless a job description is supplied — rule 17 is the only place soft skills are weighed, and it is one of the four that gets skipped without a posting.

Questions people ask after seeing their score

What is a good ATS score?

In our bands, anything from 70 up is workable and 85 up is strong. A number like 82 or 84 sits in the 70–84 group: the file is readable, and two or three rules are costing real points. What matters more than the number is where those points went — if they are on quantification, you have writing to do; if they are on parsing, fix the file first.

Why did my score drop when I added a job description?

Because thirty points of job-match criteria switched on and the denominator grew from 71 to 101. Nothing about your resume changed. The lower number is the one that refers to that posting.

Does this match what Workday sees?

No, and nobody outside Workday can tell you what it sees. This is a transparent heuristic over rules that are published in full; it is not a copy of any vendor's parser. Where a rule depends on local hiring convention rather than mechanics, the methodology page says so.

My score is 100. Am I done?

With the file, yes. With the application, no — a perfect mechanical read still has to survive a human reading it, and the tailoring that matters from here is per-posting wording, which moves rules 14 to 17 only when you supply the description.

See which of the 21 you fail

Upload or paste, get the score in about ten seconds. No account, no card, and the file stays on your machine.