Free ATS Resume Checker

What an applicant tracking system actually reads

An applicant tracking system is a database with a parser bolted to the front of it. When you attach a file to an application, the parser has one job: turn your document back into fields — name, email, phone, employers, job titles, dates, education, skills — and store them as rows a recruiter can search and filter. Nothing about that process reads your resume the way a person does. It has no opinion about your career. It is looking for text it can label.

That is why formatting decisions that look harmless on screen cause damage. Text inside a graphic is not text: a parser sees an image and stores nothing. A two-column layout can be read in the wrong order, so a job title ends up attached to the wrong employer. A header or footer is often skipped entirely, which is the most common way a phone number and an email address go missing from an application that clearly had both. Tables, text boxes and unusual section names ("Where I have worked" instead of "Experience") all reduce the number of fields that come out the other side.

The second half of the system is the search. Recruiters do not read every stored record; they filter it. They search the database for the terms in the job description — the tool, the certification, the years, the job title — and they read what comes back. A resume that parsed perfectly but does not contain the words the recruiter searches for is, in practice, invisible. This is the part a checker can measure, and it is what the score on this page is about.

One thing worth being blunt about: there is no universal "ATS score". Workday, Greenhouse, Taleo, SuccessFactors and Lever all parse and rank differently, and none of them publishes a number. Anyone who tells you your resume scored 74 out of 100 "in the ATS" is describing their own tool, not a system a recruiter uses. What a checker can honestly tell you is how much of the posting's own vocabulary your resume actually contains, and whether the sections a parser expects are present and findable.

The exact checks this checker runs

This checker is rules-based and runs entirely in your browser. No model is involved, nothing is uploaded, and no text is sent to a server, which is why it is free with no account and no card. Here is precisely what it does, in order.

First it reads the job description and extracts up to 25 keywords. Single words must be at least three characters and survive a stop-word list — the usual English function words plus the boilerplate that appears in every posting and means nothing as a resume keyword: "role", "position", "candidate", "responsibilities", "requirements", "preferred", "seeking", "years", "strong". A word that appears twice or more in the posting is weighted at double its frequency, because repetition is the clearest signal a posting gives about what actually matters.

It also extracts two-word phrases, but only from words that sit next to each other inside the same clause. The text is split on sentence and clause boundaries first — full stops, commas, semicolons, colons, brackets, bullets, line breaks — so a phrase can never be assembled across a list separator. Without that rule, "…coordination, SQL reporting…" produces the phantom phrase "coordination sql", which no recruiter would ever search for. A phrase only earns a slot in the 25 if the posting used it more than once; a one-off pair is ranked below every real single keyword.

Then it matches those keywords against your resume. Single words are matched on word boundaries, so "man" does not match "management" — with a deliberate exception for tokens that end or begin in punctuation, because a plain word-boundary test can never match "C++" or "C#" and would report them as missing on a resume that clearly lists them. Interior dots survive tokenisation so "node.js" stays one keyword. Two-word phrases are matched by proximity instead of exact order: both words must appear within five words of each other, so "reduced churn" scores against "churn reduction" and against "reduced monthly churn by 12%".

The matcher is Unicode-aware, which matters more than it sounds. An ASCII-only character class silently shreds every non-English resume: it splits "Führungskraft" into "f" and "hrungskraft", "gestão" into "gest" and "o", and reduces Arabic text to nothing at all. Accented Latin, Arabic and everything else are kept whole, so the score means the same thing in the seven languages this site serves.

Separately, it looks for the five sections a parser expects, using the words your headings and content actually use: contact details (an email address or a phone-shaped number), a summary, objective, profile or about section, experience, work history or employment, education, a degree, a university or a college, and skills, technologies, proficiencies or competencies.

The score is then arithmetic, with no hidden weighting:

• Keyword match: the share of the extracted keywords found in your resume, worth up to 70 points. Match 20 of 25 and that is 56 points.
• Sections: 6 points for each of the five sections that is present, worth up to 30 points.
• The two are added and rounded, and capped at 100. There is nothing else in the number.
• You also get the matched list, the missing list, and up to five specific suggestions built from what was missing.

What your score means, and what it does not

The labels on the score are fixed bands. 80 and above is Excellent: the resume speaks the posting's own language and the parser will find every section it looks for. 60 to 79 is Good — usually a resume with all five sections and a real gap in vocabulary, which is the easiest gap on this list to close. 40 to 59 is Needs work: typically half the posting's terms are absent, and often a missing skills section is costing 6 points on its own. Below 40 means the resume and the posting have very little language in common, which is what a generic resume sent to a specific job looks like from the inside.

Two honest caveats. Because sections are worth 30 points, any complete, well-formed resume starts at 30 before a single keyword is matched — so a score in the thirties is not a compliment, it is the floor. And the score is only ever about one posting. The same resume, scored against three different jobs, will return three different numbers, and that is the point: tailoring is per job, not per career.

The number is also not a prediction of an interview. It measures whether a recruiter's search would surface you and whether a parser would file you correctly. What happens after a human opens the document is about evidence, and no keyword count can substitute for a bullet that says what changed because you were there.

The one rule that outranks everything else here: never add a keyword for something you have not done. The missing list is a prompt to describe work you did in the words the employer uses, not an invitation to claim work you did not do. Every term you add has to survive the interview where someone asks you about it.

A worked example: 38 to 92 on the same career

Both resumes below belong to the same person applying to the same posting, a Customer Support Specialist role at a SaaS company. Nothing was invented between the two versions: the employers, dates, degree and actual work are identical. The scores are the real output of the checker on this page.

The first version scored 38, matching 3 of 25 keywords. All five sections were present — that is the 30 points doing its work — so almost the entire gap was vocabulary. The bullets said "Answered emails and phone calls from users every day", "Helped people with their accounts and payments" and "Worked with the developers when something was broken". Every one of those describes real work in words the posting never uses. The missing list came back with zendesk, saas, csat, sla, ticket, queue, billing, onboarding, escalate, inbound and eleven more.

The second version scored 92, matching 22 of 25. The same three jobs, rewritten in the posting's vocabulary and with the numbers filled in: "Own the inbound ticket queue in Zendesk, holding first response time under 2 hours against a 4-hour SLA across 320 tickets a month". "Resolve billing questions and refunds end to end, cutting billing escalations 38%". "Maintain the help centre knowledge base, 60 articles, which deflected 22% of inbound tickets". The summary changed from "Friendly and hard-working professional" — a sentence that contains no searchable term at all — to "Customer Support Specialist with 5 years in SaaS customer support, running a Zendesk ticket queue against a 4-hour SLA and lifting CSAT from 82% to 94%". The skills list changed from "Communication, teamwork, problem solving, Microsoft Office" to the tools and processes the posting names.

Three keywords were still missing at 92: "respond", "within" and "keep". Those are verbs from the posting's own sentence structure, not skills, and chasing them would make the resume worse. A perfect 100 is not the target and the last few points are usually noise.

The seven things that cost the most points

In the order they are worth fixing, based on what the scorer weights and what the missing lists keep showing:

• No skills section. It is 6 points on its own, and it is the single easiest place to carry the posting's tool names in a form a recruiter search will hit.
• A summary written in adjectives. "Detail-oriented team player" contains nothing searchable. Lead with the job title from the posting and one measured result.
• The employer's job title instead of the industry's. If you were a "Client Happiness Ninja" and the posting says "Customer Support Specialist", put the standard title in and keep the real one alongside it.
• Tools named only in the skills list. A tool that appears once in a list and never inside a bullet reads as familiarity, not experience — and phrase matching rewards it appearing near the work you did with it.
• Text inside images, headers or footers. Parsers routinely skip all three, which is how contact details vanish from applications that plainly had them.
• Two-column layouts and tables for anything but decoration. They are the most common cause of a title being filed against the wrong employer.
• One resume sent to every posting. The score is per job; the missing list changes every time, and so should about ten lines of the document.