.png&w=384&q=75&dpl=dpl_B2k3vFW3cCgG1Wa3Z3Uh89ayHsjZ){kind=small}
The problem: too many resumes, too little time
A position posted on a popular job board receives between 150 and 300 applications in the first week. Reviewing each resume manually takes 6 to 8 minutes, meaning a recruiter needs 15 to 40 hours just for the first filter on a single opening.
That is not sustainable. And the problem is not just time: cognitive fatigue causes screening quality to drop dramatically after the first 50 reviews.
Key Takeaway
AI does not replace the recruiter's judgment in screening. What it does is eliminate the 15-40 hours of initial manual review, delivering a prioritized list of the best candidates so the human can make the final decision.
Step 1: Define your screening criteria
Before activating any AI tool, you need clear, objective criteria. Without them, AI cannot evaluate anything.
Knockout criteria (mandatory)
These are binary filters that automatically disqualify:
- Minimum years of experience in the role
- Certifications or degrees required by law or regulation
- Geographic location (if the role is on-site)
- Start date availability
Scoring criteria (weighted)
These are factors that add points to the profile:
- Relevant industry experience (weight: 25%)
- Specific technical skills (weight: 30%)
- Career growth trajectory (weight: 20%)
- Experience with specific tools (weight: 15%)
- Educational background (weight: 10%)
Step 2: Set up your screening tool
Option A: With Selenios (fully automated)
- Create the job opening in the Selenios dashboard
- Define knockout and scoring criteria in the configuration form
- Activate candidate intake
- Selenios analyzes each resume in real time, extracts structured data, and assigns a score
- Candidates are automatically classified as: Recommended (score above 80), Possible (60-80), and Not a Fit (below 60)
The entire process takes under 60 seconds per candidate, including data extraction, compatibility analysis, and classification.
Option B: With ChatGPT (semi-manual)
If you do not have access to an AI-powered ATS, you can use ChatGPT as a support tool:
- Copy the candidate's resume text
- Use this prompt: "Analyze the following resume for the [role] position with these criteria: [list criteria]. Assign a score from 0-100 and explain the main strengths and weaknesses. Resume: [paste text]"
- Repeat for each candidate
This option works but is significantly slower (2-3 minutes per CV) and does not scale for high volumes.
Step 3: Configure automated scoring
Scoring is the heart of the system. Here is how it works:
Recommended scoring model
Each criterion has a weight and a scale. The AI evaluates each dimension and calculates a composite score:
- Relevant experience (0-30 points): years in similar roles, industry, level of responsibility
- Technical skills (0-30 points): match with required technologies or competencies
- Career trajectory (0-20 points): professional progression, job stability, quantifiable achievements
- Requirements fit (0-20 points): location, availability, salary expectations
Total score: 0-100
Step 4: Review and calibrate
No AI system is perfect from day one. Follow this calibration process:
Week 1: Full review
Manually review the first 30 candidates classified by the AI. Compare your evaluation with the system's. If discrepancies exceed 15%, adjust the criteria weights.
Weeks 2-4: Sample review
Review a random sample of 20% of candidates classified as "Not a Fit" to verify good profiles are not being missed. Adjust knockout criteria if necessary.
Month 2 onward: Ongoing monitoring
Once calibrated, review 5-10% of classifications monthly. Correlate screening scores with hired candidate performance to continuously refine the model.
Expected results
When you correctly implement AI screening, these are the typical outcomes:
- 75% reduction in screening time: from 23 hours per week to under 6
- 30% improvement in shortlist quality: candidates reaching interviews are more relevant
- Reduction of unconscious bias: AI evaluates competencies, not name, gender, or university
- Improved candidate experience: faster responses and more agile processes
Mistakes to avoid
- Overly rigid criteria: if no candidate passes the filter, your knockouts are too strict
- Not calibrating the system: AI needs human feedback to improve
- Relying solely on the resume: CV screening is the first filter, not the only one. Complement with conversational interviews
- Ignoring candidate experience: even if you automate the backend, the candidate should feel a human, professional process
How does AI resume screening work?+
AI analyzes each resume by extracting structured data such as experience, skills, and education, and compares it against predefined job criteria. It assigns a compatibility score and automatically classifies candidates. Tools like Selenios do this in under 60 seconds per candidate, including extraction, analysis, and classification.
Is automated resume screening reliable?+
Yes, when configured correctly. Modern AI systems achieve 85-92% accuracy in candidate classification, comparable to or better than human screening which has a 20-30% error margin due to fatigue and unconscious bias. The key is defining clear criteria and calibrating the system in the first few weeks.
How much time does AI resume screening save?+
An average recruiter spends 23 hours per week reviewing resumes manually. With AI, initial screening drops to under 60 seconds per candidate, saving up to 18 hours per week. That time is redirected to higher-value interviews and evaluations where human judgment is irreplaceable.