Our algorithms, explained
Z Jobs uses no opaque artificial intelligence to rank you. Our calculations are simple rules, written by people, published here, and none of them decides in a human’s place.
What we do not use
No machine learning, no language model, no analysis of video, voice or photos, no “success prediction”. Age, sex, origin, nationality, health, disability and family situation never enter a calculation.
How offers are ranked for you
When you search, the offers that best match your stated criteria come first. Six factors, each explained on the offer:
- Skills35 %
- Contract type15 %
- Work pattern (on site, hybrid, remote)15 %
- Pay15 %
- Career stage you chose (first job, senior role, temporary work), never inferred from your age10 %
- Location10 %
The guidance score, for the employer
When you apply, the employer can see a score from 0 to 100 that helps them decide in which order to read applications. Four factors, all explained:
- Skills: the offer’s keywords found in your skills40 %
- Experience: your roles, and those with a measurable result30 %
- Languages the offer asks for20 %
- Availability10 %
Missing information counts as half: it does not penalise you. Profile completeness has not been part of the calculation since 13 November 2026, because it says nothing about the job.
The score rejects, filters and advances no one: every decision is made by a person. You see your own score and its factors for every offer you applied to.
The questionnaires
The work profile and the career test are never used in a ranking, a score or a filter. An employer only sees your work profile if you share it.
How offers are checked
Before publication, every offer is checked automatically: discriminatory wording, age limits, pay below the legal minimum. The employer sees the issue, the article of law and a rewording. The offer is checked, never a person.
Your rights
You can ask for an explanation of a ranking or a score, have your data corrected, and object: write to hello@zjobs.online. No decision about you is made by a calculation alone. If you believe you face discrimination, you can also refer the matter to the Défenseur des droits free of charge.
The EU AI Act
Tools that evaluate candidates fall under the “high-risk” uses of the EU AI Act. Their obligations apply from 2 December 2027. Our score is made of fixed rules, without learning, and may not be an “AI system” within the meaning of the Act. We nevertheless apply its requirements today: risk management, documentation, a log of every consultation, human oversight, immediate shutdown.
Current version of the score rules: cs-2026-11-13. Every consultation of scores is logged with this version.
Sources
- Règlement (UE) 2024/1689 sur l’intelligence artificielle (EUR-Lex)
- RGPD, art. 15 et 22 (EUR-Lex)
- Code du travail, art. L1221-6 à L1221-9 (Légifrance)
- Défenseur des droits
Page updated on 13 November 2026.