Screen inbound CVs against a structured scorecard
Last updated 11 August 2026
CV screening lets recruiters rank a high-volume applicant pool against a written scorecard, by extracting evidence for each criterion from every application and scoring it consistently, typically cutting first-pass reading time while creating an audit and bias-testing obligation that must be settled before it goes near a live vacancy.
| Dimension | Score | What that means |
|---|---|---|
| Impact | 3/5 | Meaningful savings for one team |
| Effort | 4/5 | Custom build, about a quarter |
| Data readiness | 2/5 | Needs one tidy export |
| Company size | 51–200 · 201–1000 · 1000+ | |
What problem this solves
A single vacancy draws six hundred applications. A recruiter gives each one somewhere between six and thirty seconds, and by Friday afternoon the standard applied at the top of the pile is not the standard applied at the bottom.
Good candidates are missed because their CV is laid out badly. Two recruiters shortlisting the same pool produce different lists, and neither can explain afterwards why a particular person was rejected.
How it works
- Write the scorecard first, as a hiring manager would: the specific criteria, what evidence counts for each, and how they are weighted.
- Extract from each application the evidence relevant to those criteria, rather than scoring the document as a whole.
- Score each criterion separately and record the passage the score was based on.
- Present a ranked shortlist with reasons to a recruiter, who decides — the system does not reject anyone.
- Test the scoring for adverse impact across sex and race before use, and repeat it on a schedule.
- Publish the audit summary and notify candidates where the law requires it, and keep the whole process under human review.
What you need to start
- A written scorecard agreed with the hiring manager before any automation — the scorecard is the product, the model only applies it
- An independent bias audit: New York City requires one within the prior year, plus a published summary and 10 business days’ notice to candidates
- Legal review for your jurisdiction: the EU AI Act lists systems that analyse and filter job applications as high-risk under Annex III
- A recruiter who makes the actual shortlisting decision and can explain it
Expected outcomes
| Metric | Typical range | Source |
|---|---|---|
| First-pass screening time | Reduced; no independent figure | — |
| Scoring consistency across a pool | Same rubric applied to every CV | — |
| Measured bias in LLM CV ranking | White names favoured in 85.1% | View source |
Real-world signal
An AIES 2024 audit of language-model CV retrieval across more than 500 resumes and 500 job descriptions found white-associated names favoured in 85.1% of cases and female-associated names in 11.1%, with Black male names disadvantaged in up to 100%.
A study of 391 New York City employers found only 18 had posted a Local Law 144 bias audit report and 13 a transparency notice, showing how rarely the disclosure requirements are met in practice.
The EU AI Act classifies AI systems intended to be used for recruitment or selection, in particular to analyse and filter job applications and to evaluate candidates, as high-risk under Annex III, point 4(a).
Common questions
How much data do you need to start?
None historically, which is the trap. The system works from the scorecard and the applications, so nothing stops you switching it on — and that is exactly why the bias testing and the legal review have to come first.
Does this decide who gets rejected?
It should not. Use it to rank candidates and surface evidence against criteria, with a recruiter making the shortlisting decision. Automated rejection raises legal exposure sharply and removes the human judgement an audit assumes is present.
What are the legal obligations?
They depend on where you hire. New York City requires an independent bias audit within the prior year, a published summary and 10 business days’ notice to candidates. The EU AI Act treats application filtering as high-risk. Get this reviewed before launch.
How long does it take to get running?
The technical build is short. Agreeing the scorecard, commissioning the audit and clearing legal review is what sets the timeline, and none of those steps is optional.
Related use cases
Resolve routine support requests without an agent
Automated first-line support lets customer service teams resolve repetitive requests — order status, returns, refunds — without a human agent, by answering from policy documents and order systems and handing anything unresolved to a person, typically absorbing a large share of contact volume while the team keeps the complex cases.
Extract terms from supplier contracts
Contract term extraction lets legal and procurement teams see the obligations buried across a supplier portfolio, by reading each agreement and pulling renewal dates, liability caps, indemnities and price mechanisms into a structured register, typically producing output around the standard of a junior reviewer that still needs a qualified check.