How the score is built. What it measures.
A short, plain account of what we weigh, how we validate it, and what is out of scope.
What goes into a score.
Four signals. Documented weights. We do not change these silently.
How precisely the resume's surface terms align with the job description's required and preferred skills.
Contextual meaning. We compare embeddings so equivalent phrasing (e.g., 'led' vs 'managed') counts when it should.
How cleanly an ATS can parse the document. Section headers, date formats, columns, tables, headers and footers.
Quantified achievements, action verbs, the presence of measurable outcomes inside bullets.
What the score is anchored to.
The score is calibrated against documented ATS parsing behaviour, not a single self-reported survey.
- We maintain parser fixtures for the major ATS families (Workday, Greenhouse, Lever, iCIMS, Taleo, SuccessFactors). The format compatibility check runs against these fixtures directly.
- Keyword and semantic weights are calibrated against published guidance from each ATS vendor and recurring patterns we observe in real job descriptions.
- When a vendor changes parsing behaviour, we update the relevant fixture and the format check. The weights themselves change only if the underlying signal is shown to be unreliable.
- We log internal regression scores against a held-out set of synthetic and consented user resumes to catch regressions before they reach production.
Honest limits.
If we cannot measure it cleanly, we do not put a number on it.
- We do not guarantee interviews or offers. We optimize ATS compatibility and content alignment.
- Our score is a prediction trained against documented ATS behaviour, not a direct read from any specific vendor.
- Outcomes vary by industry, role seniority, geography, and the surrounding market.
- We do not publish self-reported survey numbers as marketing claims. If we cannot measure it cleanly, we do not quote it.
Run the score on your resume.
New accounts get 1 free optimization. Upload, paste a job description, and read each weighted breakdown for yourself.