How to Screen CVs with AI Fairly: The 7-Dimension Rubric
November 17, 2026 · 6 min read · NadeemHR Team
Why keyword filters fail
Most applicant filters work by matching words. If the job post says "customer success" and the CV says "client relationship management", a keyword filter sees no overlap, even though the work is the same. The same thing happens with synonyms, abbreviations and tools that were renamed.
The problem is larger for career changers, whose strongest evidence sits in transferable work rather than a matching job title. It also affects CVs in different formats and languages: a candidate who writes their CV in Arabic, or lays it out as a two-column design, can be scored lower for reasons that have nothing to do with ability. The result is a shortlist that rewards the people who copy the job post's vocabulary, not the people who can do the job.
There is a second cost that is easy to miss: candidates learn to game the filter. When people suspect that a machine is counting words, they stuff their CVs with phrases copied from the job post, which makes every CV look alike and hides the genuine differences you were trying to find. Understanding meaning, not matching strings, is the more durable approach.
Start with a scoring rubric
Before you open a single CV, write down what good looks like for this role. A rubric is simply a list of criteria with a weight for each, so that "strong candidate" means the same thing on the first CV and on the two-hundredth.
Weights matter because roles differ. A support agent and a head of finance should not be scored on identical priorities. In NadeemHR you enter a role title or paste a job description, and the platform builds structured requirements and a weighted rubric for you. The hiring team can then adjust the weights for that specific role before any candidate is scored.
Writing the rubric first also protects you from your own biases. It is easy to read a polished CV from a familiar employer and decide the person is excellent, then look for reasons to agree with yourself. A rubric written in advance forces you to ask what the role genuinely requires, and it gives your team a shared vocabulary for disagreeing about candidates productively.
The 7 dimensions explained
NadeemHR scores every candidate across seven dimensions. Here is what each one asks of a CV, and what evidence a CV can realistically offer.
- Skills: the technical and practical abilities the role needs, shown through projects, tools and results rather than a list of buzzwords.
- Experience: relevant work history, its depth and its recency, including responsibilities and scope.
- Education: degrees, certifications and training, weighed by how much the role actually depends on them.
- Culture fit: signals about working style and values, such as the kinds of teams and environments the candidate has thrived in.
- Communication: how clearly the CV itself is written, plus evidence of presenting, writing or client-facing work.
- Leadership: managing people, owning outcomes or mentoring others, whether or not the title says "manager".
- Career trajectory: the direction of growth over time, such as increasing responsibility, deliberate moves and sustained progress.
Weights change with the role
A rubric is only useful if it can bend to the job in front of you.
The same seven dimensions apply to every role, but their importance shifts. As an illustration, leadership might carry a high weight when you hire a team manager and a low one for a junior analyst, where skills and education may matter more. Because the weights are set by your hiring team for each role, the rubric reflects your priorities rather than a one-size-fits-all formula.
Keeping AI screening fair
AI does not make screening fair automatically. Fairness comes from how you set it up and supervise it. These practices help.
The goal is not a machine that never makes mistakes; it is a process where mistakes are visible and correctable. A ranked list is a recommendation about where to look first. It should never become a silent filter that no one questions, and it should never replace the judgement of the people who will work with the person you hire.
- Use the same rubric for every candidate in a role, and change weights before scoring starts, not after you have seen the results.
- Prefer evidence-based scores: every score should point to something in the CV that you can read yourself.
- Review the explanations behind scores, especially for candidates ranked lower than you expected.
- Keep protected characteristics out of your criteria. A rubric should describe the work, not the person's background.
- Have people review the shortlist before anyone is rejected or advanced.
- Audit decisions now and then: compare who was advanced with who applied, and adjust the rubric if something looks off.
From ranked shortlist to interview
A ranking is only the start of the conversation. In NadeemHR, the top candidates can go straight to a first-round voice interview with Layla, the AI interviewer, in English or Arabic. The interview adds evidence a CV cannot: how the person explains their work out loud.
Hiring managers then receive an executive scorecard for each candidate with strengths, potential gaps and suggested follow-up interview questions, so the next human conversation starts from a clear picture rather than a blank page.
Getting started
Paste a job description into NadeemHR and it builds the requirements and the weighted rubric for you. Then upload your CVs in bulk, including Arabic CVs, and review the ranked results with the reasoning behind each score. You can try the whole flow on one live job with the free one-time trial. Prices start as low as SAR 6,970 per quarter or SAR 23,690 per year (Startup plan, save 15%), and the details are on the pricing page. When you are ready, start from the sign-up page.
Frequently asked questions
Can AI read Arabic CVs?
Yes. NadeemHR parses CVs in multiple languages, including Arabic, and ranks them against the same rubric as CVs written in English.
Can we change the weights?
Yes. NadeemHR suggests weights for the seven dimensions from the job description, and your hiring team can adjust them for each role.
Is AI screening fair?
It can be more consistent than manual screening because every candidate is scored against the same rubric, but fairness still depends on good criteria and human review. NadeemHR supports your decision; your team makes it.
See NadeemHR on one of your jobs
Start with a free one-time trial on one live job.
