Grading system
How scoring works
Every number in KairoMatch, what it grades, and what it doesn't.
KairoMatch shows you three different scores. They are not versions of each other and they are not meant to agree — each answers a different question. Here is exactly what each one grades, what it does not, and how the maths works.
The three scores
Résumé score
· this documentConsidered read + your material- What it measures
- How well your résumé reads to a recruiter and an ATS.
- What it does NOT measure
- It does not measure whether you are qualified — a beautifully written résumé for a role you have not done still scores well here.
- Where you see it
- Aim's headline number, the résumé review page, and the source pane.
- What moves it
- Rewriting the document — clearer structure, quantified bullets, ATS-safe formatting. Uploading a new version re-scores it.
Scale: 0–100
Market standing
· your occupationConsidered read + your material- What it measures
- Where you stand against what the market asks for your occupation, as a level from None to Expert.
- What it does NOT measure
- It does not rate the résumé document, and it is not a fit score for any particular job you are applying to.
- Where you see it
- Aim's standing header and its Score & skills plate.
- What moves it
- Adding real experience, skills or scope to your timeline — then re-running the career analysis.
Scale: a level, None → Expert
Overall Fit
· this postingDeterministic- What it measures
- How well you match ONE job posting.
- What it does NOT measure
- It does not rate your résumé's quality (that is only 5 of the 100 points) and it says nothing about your standing in the wider market.
- Where you see it
- The job-fit canvas, each job card, and the Latest/Best fit tiles.
- What moves it
- Analyzing a posting you match better, or closing the gaps this one lists — must-haves move it most.
Scale: 0–100%
Why isn't my résumé score the same as my fit score?
They answer two different questions. The résumé score grades the DOCUMENT — how it reads to a recruiter and an ATS. The fit score grades YOUR HISTORY AGAINST ONE POSTING — how much of what that job asks for you can evidence. The document is only the "Résumé score (document quality)" bucket inside fit, worth 5 of 100 points; the other 95 are requirements, skills, experience and domain coverage for that specific role. That is why a well-written résumé can be a poor fit, and a plain one can be an excellent fit.
How the fit percentage is built
- Must-have requirements37
- Experience & seniority18
- Skills proficiency (O*NET)15
- Nice-to-have requirements13
- Domain / keyword alignment12
- Résumé score (document quality)5
Weights sum to 100 and are read live from the scorer — this page cannot show a version of the formula we no longer run.
- Partial evidence still counts. An adjacent or transferable match on a must-have earns 65% of that requirement's credit, and 75% on a nice-to-have — because real hires match most requirements, not all of them. Missing preferred qualifications never zero their bucket (it floors at 30%).
- A failed dealbreaker caps the result at 40%. If the posting requires something you genuinely do not have — a licence, work authorisation, a hard location — no amount of strength elsewhere pushes past that cap. A dealbreaker we could not verify from your résumé costs 5 points instead, and the card says which one.
- The displayed maximum is 99%. There is always something left to improve, so we never print a perfect score.
- Only the buckets a posting actually has are counted. A job description that lists no nice-to-haves has its 13 points redistributed across the rest — so the percentage is always out of what was really asked for.
- The band is read off the number: 75% and up is HIGH, 55% and up is MEDIUM. The badge can never disagree with the percentage.
Career stage
Where you sit inside your current occupation — early, established, advanced, or ready for the next role. It is computed from the history you already stored: how long you have worked, how long you have held the current role, the seniority words in your own job titles, and any team-size or budget language in the bullets you saved. This costs nothing, and it never reads anything you have not entered.
title ladder (intern/junior → associate → mid → senior/lead/staff/principal → manager/director/VP) + total years + years in current role + management-scope evidence → early (<3 yrs or entry titles) · established (3–8 yrs at level) · advanced (senior title AND ≥6 yrs) · next-role-ready (advanced AND ≥2.5 yrs in the current role, or management scope already evidenced)It is not a ranking against other people and it is not a promotion forecast — it is a reading of your own recorded history against the usual shape of a career in this occupation.
The market standing levels
Market standing is reported as a level, not a number. The model still returns a 0–100 and the level is banded off it — those cut points are below. The 0–100 appears in two places — the Aim standing card and the standing plate. Everywhere else your standing reads as its level. Deterministic weights · O*NET-grounded · free, no run to show it. A two-point move is noise rather than progress.
None— Level 1 of 6
- None
- no score yet
- Not analysed yet — no run, so no level. Not a verdict about you.
- Basic
- 0–39
- Early against what the occupation asks; most of the market's requirements are not yet evidenced in your history.
- Developing
- 40–54
- Building — a real share of the occupation's demands are evidenced, with clear ground still to cover.
- Proficient
- 55–69
- Solidly inside the occupation's expectations for the work you have evidenced.
- Advanced
- 70–84
- Above the bar on most of what the occupation asks, with depth the market recognises.
- Expert
- 85–100
- At the top of what the occupation asks — the evidenced ceiling of this scale.
The ranges are read live from the banding function the app runs — this page cannot publish a cut point the product does not use. None means no analysis has run; it is not a zero. A measured 0 is Basic.
Where each number comes from
- Considered read + your material
- A considered read of your material — your history, résumé, or pasted text — that forms a view. It writes down its reasoning, and the same input can produce a slightly different read.
- O*NET · BLS data
- Straight from U.S. Department of Labor data (O*NET and BLS) for your occupation. We report it; we don't adjust it.
- Deterministic
- Arithmetic on your device over data you can see. Same inputs, same number, every time — nothing is guessed.
- Database
- Computed in our database over public reference data, not over anything you typed.
- Measured usage
- Measured from your own real usage — token counts and runs we recorded.
What we will never do
- We never invent a percentage. If an analysis could not be grounded well enough to produce a number, you see an em-dash and the reason — never a plausible-looking score.
- We never show a number without saying what it is about. Every score carries its scope: this document, this posting, your occupation.
- We never quietly change what a score means. When the maths behind a number changes, the explainer behind its ⓘ changes in the same release.
- We never let two screens disagree. Every number on every surface comes from one shared derivation — if the rail and the dashboard could differ, that is a bug, not a nuance.
These are the formulas. For the plain-language version — what happens to your résumé, which public data each number is built from, and what KairoMatch will never do — see How KairoMatch works.