How HireSift Works

Every candidate receives a composite score built from two layers. The first — and most important — layer accounts for 80% of the total score and measures core competencies: the actual skills the job requires. The second layer is a domain bonus of up to 20 points, awarded when a candidate has direct experience in the target industry.

This separation matters. Consider a dental receptionist role requiring high-volume phone handling and appointment scheduling. HireSift scores candidates on those specific capabilities — regardless of where they demonstrated them:

The hotel candidate ranks higher because they demonstrably did more of what the job actually requires. Industry experience alone does not make a good hire — demonstrated capability does.

You define the criteria. HireSift suggests them based on your job description, but you review, edit, add, and remove every item before screening begins.

Criteria come in two types:

• M

  Mandatory

  Must be met for a candidate to qualify. Examples: eligible to work in Ireland, minimum qualifications, specific certifications. Candidates who do not meet mandatory criteria are placed in a separate “Doesn't Qualify” tab — they are never hidden, always visible to you.

• P

  Preferred

  Desirable but not essential. Each preferred criterion has a weight (High, Medium, Low) that adjusts its contribution to the overall score. A High-weight preferred criterion matters more than a Low-weight one.

The employer always has the final say. HireSift produces a ranked list — every shortlisting and rejection decision is made by you.

Not all CVs are equal in quality. A CV that says “managed a high-volume phone line handling 80 calls/day” provides much stronger evidence than one that says “excellent communication skills.”

HireSift assigns each candidate a confidence rating — High, Medium, or Low — based on how specific and plausible their claims are:

• ●High — Concrete metrics, named employers, specific tools, verifiable timeframes. The CV tells a coherent story.
• ●Medium — Some specific evidence but gaps in detail. Claims are plausible but not fully substantiated.
• ●Low — Generic language, buzzword-heavy, vague dates, inconsistencies, or signs of AI-generated content. Treat scores with caution.

The confidence rating is advisory — it helps you prioritise which candidates to look at first, not to automatically exclude anyone. You can filter the dashboard by confidence level.

HireSift never produces a number without an explanation. Every candidate's score is fully decomposed:

• →Each mandatory criterion: met or not met, with a one-line justification from the CV.
• →Each preferred criterion: met or not met, with a score out of 10 and a justification.
• →Each core competency: a score out of 100 and a justification citing specific CV evidence.
• →The domain bonus: the exact reasoning for why points were awarded or withheld.
• →Flags: anything unusual, suspicious, or that the employer should verify manually.

Click any candidate card to see the full breakdown. Nothing is hidden from the employer. If the scoring looks wrong, you will be able to see the reasoning and override the decision.

HireSift is classified as a high-risk AI system under the EU AI Act (Annex III — employment and workers management systems). We comply with all applicable obligations:

• ✓Human oversight: Every hiring decision requires a human action. HireSift cannot reject or shortlist candidates autonomously.
• ✓Transparency: Candidates are informed that automated screening is used in the hiring process.
• ✓Bias testing: We regularly audit screening outputs across demographic proxies to identify and correct systematic bias.
• ✓Audit logging: All screening events, scores, and employer decisions are logged with timestamps and are available for regulatory inspection.
• ✓Risk management system: We maintain a documented risk management system covering accuracy, fairness, and data quality.
• ✓Technical documentation: Full technical documentation is available to competent authorities on request.

We take data privacy seriously. Here is what happens to your data:

• →EU-hosted: All data — employer accounts, job records, candidate CVs and scores — is stored on EU servers (Stockholm, Sweden) and never leaves the EU.
• →Auto-deletion: Candidate data (CVs, extracted text, scores) is automatically deleted 6 months after a job is closed.
• →No AI training: Candidate CV data is never used to train AI models. Your applicants' data is used solely for screening the role they applied for.
• →Right to deletion: Employers can delete any candidate record at any time from the dashboard. Account deletion removes all associated data within 30 days.
• →Access controls: Each employer can only access their own data. Row-level security ensures no cross-account data leakage.
• →GDPR lawful basis: Processing is carried out under legitimate interest (hiring assessment). Candidates are notified via the employer's job posting.

Questions? Contact us at privacy@hire-sift.com