How we score, and what we never do
Claims about resumes are easy to make and hard to check. This page explains exactly what the match score means, how we keep results truthful, and where the limits are.
The match score
- Read the job. An AI model extracts 12 to 25 requirements from the posting: tools, skills, certifications, responsibilities. Each is marked required or nice to have, with common synonyms.
- Count what the resume shows. Required items weigh 2 points, nice-to-have items 1. A requirement counts only if the resume text clearly contains it (or a listed synonym).
- Compare. The score is the share of points covered, calculated by plain code on your original and again on the tailored resume. The model does not decide the number.
What the score is not: a prediction of an interview or job offer. It measures how clearly your resume demonstrates what the posting asks for, which is what recruiters and applicant tracking systems look for first.
How we prevent invented content
- Strict rules. The rewriting step may only reorder, shorten and rephrase facts present in your resume.
- Second opinion. A separate pass compares the result with your original and flags any unsupported claim.
- Automatic removal. Code deletes any line containing a number that does not appear in your original, and any employer that is not in it.
- Gaps stay gaps. Requirements you do not meet are shown to you as gaps, never added to the resume.
Limits you should know
- AI can still make mistakes. Read every line before you send the resume. You are responsible for what you submit.
- If your resume is thin, the score cannot reach 90% honestly. We show the real number.
- Links to job posts on some sites cannot be read automatically; paste the text instead.
- Scanned PDFs without selectable text cannot be read.
Your data
Unpaid previews are deleted after 7 days and paid results after 30. Your resume is not sold, not used for marketing and not used to train models. See the privacy policy.