Type "free ATS scanner" into Google and you get fifty tools that promise the same thing: paste your résumé, get a score. Most of them are doing one of two things — counting keywords with a regex, or asking GPT to "rate" your résumé and reading whatever number it spits back.
Neither approach is honest about what an ATS actually does, and neither helps you decide what to fix.
We built a different version. Today it's at jobjab.ai/tools/ats-score, free, no sign-up, runs against any job description you paste in. This post explains what's inside it and why we shipped it as a free tool instead of paywalling.
What it actually checks
A résumé is graded on six dimensions, each scored 0-100, weighted into one headline number:
- Keyword Match — what fraction of the job's must-have terms appear in your résumé. The boring one. Most "ATS checkers" stop here.
- Impact & Metrics — do your bullets quantify outcomes (numbers, percentages, dollars), or are they job descriptions in disguise?
- Leadership Level — does the seniority your résumé conveys match what the role requires? "Led a team of 12" vs. "supported the team" reads very differently.
- Role Fit — how closely your past titles and responsibilities mirror the role on offer. A staff engineer applying to a tech lead position scores differently than a senior dev applying to the same.
- Communication — writing clarity, action-verb strength, bullet structure. The thing recruiters skim for.
- ATS Compatibility — parser-friendly structure: clean date formats, standard section headings, single-column layout, no tables or images that real ATS systems mangle.
You get a number on each, a one-line concrete fix per dimension, and one next best move — the single change that would lift your headline score the most.
Why six dimensions, not one
Because "your résumé scores 73" is useless and "Communication: 55, here's what to fix" is actionable.
If you've used the kind of tool that hands you a single number with a green-yellow-red verdict, you already know the limitation: you can't act on it. Did the score drop because the role-title alignment is off? Because the keywords are weak? Because your bullets don't quantify anything? You can't tell, so you either rewrite everything (expensive) or nothing (useless).
A breakdown lets you target. If your overall is 71 and the only red bar is ATS Compatibility: 38, the fix is "switch to a single-column template," not "rewrite my résumé."
Why it's free
Two reasons.
First, the rubric isn't the moat. What's inside JobJab — daily scans against the live job market, AI-tailored versions of your résumé per posting, on-demand interview answers grounded in your actual background, scheduled reports — that's the product. The score itself is just the first paragraph of a longer conversation.
Second, this is the kind of thing job seekers Google in a panic at 11pm. If we have something useful to put in front of them when they do, that's a better introduction than a homepage that explains how clever we are.
How accurate is it really
Honest answer: it agrees with hiring panels' first read about 80% of the time. The 20% where it's off is almost always one of two things —
It can't see the company. A 90/100 score against a job description doesn't tell you whether the company would actually hire you. Cultural fit, hiring momentum, internal politics — none of those leave a fingerprint in the JD.
It can't see your story. Two résumés can score the same numerically and tell completely different career stories. The model evaluates content, not narrative arc. A "good" 78 with a coherent story beats a "strong" 84 that reads like a list of unrelated jobs.
If you're using it as a "should I bother applying" filter, it's accurate. If you're using it as a "is my résumé objectively good" measure, it isn't, and no tool can be.
A note on the technology
The tool sends your résumé and the job description to Claude (Anthropic's model) with a structured prompt that defines the six dimensions and a calibration rubric. Claude returns the scores plus the keyword breakdown. We don't store your résumé, your job description, or the result — the API processes them in memory for one request and discards them.
The same engine runs inside the paid product. There's no "free version that's worse" — paying users get more runs per day (5/day on Applicant, 20/day on Executive) and the score gets attached to a specific saved job match, but the rubric and the calibration are identical.
Try it
Score your résumé against any job posting →
Bring a real job description — the more text the model has to compare against, the sharper the keyword analysis. A well-written posting around 800-1500 words is the sweet spot.