Executive Summary AI-generated CVs can inflate signals (titles, keywords, “impact” statements) and mask the real capability behind the claims. The assessment process should shift from “document-based proof” to “evidence-based capability testing”: validating outcomes, skills, and work behavior through structured interviews, work-sample assessments, reference checks, and credibility signals (how the candidate explains and operationalizes their claims).

Key Recommendations

1. Treat the CV as a “starting hypothesis,” not a screening verdict

• In screening, score for clarity and specificity (what they did, with what constraints, measured outcomes), not just seniority or buzzwords.
• Require a short “Claim-to-Evidence” mapping for key points on the CV:
• Claim: “Led X / delivered Y”
• Evidence: metric, artifact, role, timeline, team size, your contribution vs others
• Method: how they planned, executed, and improved results

1. Add structured, capability-based selection steps

A. Work-sample / case-based assessment (highest signal)

• Replace or reduce purely interview-based evaluation with realistic tasks:
• For roles in product/analytics/ops: a 1–2 hour take-home or timed case with rubric
• For leadership roles: a strategy-to-execution brief (what they would do in 30/60/90 days)
• For engineering/IT: debugging or design exercise, or “architecture review” based on a scenario
• Grade using a rubric tied to job competencies (not writing quality).

B. Competency-based behavioral interview (consistent scoring)

• Use STAR, but also probe “mechanics”:
• Decision trade-offs, failure points, stakeholder friction, metrics chosen and why
• “Walk me through the exact sequence of actions” and “what you personally owned”
• Keep interviewers calibrated with a scoring guide (e.g., 1–5 per competency with descriptors).

C. “Deep dive verification” questions for top CV claims

• For every shortlisted candidate, select 2–4 high-impact claims and validate them:
• “Show me how you measured success”
• “What data did you use / not use and why?”
• “What would you do differently knowing what you know now?”
• This discourages generic AI phrasing and reveals real experience.

1. Introduce credibility checks that AI can’t easily simulate

• Portfolio/artifact validation:
• Request concrete outputs: dashboards, PRDs, architecture diagrams, process maps, performance reviews, GTM decks, SOPs, training materials.
• For privacy: ask for redacted versions + explanation of what’s removed.
• Cross-check logic:
• If they claim a major outcome, confirm feasibility: team size, budget, tools, timeline, scope, dependency mapping.
• Role clarity audit:
• Ask “What was yours vs the team’s?” and “What did you personally do vs direct others to do?”

1. Use reference checks with a structured template (and align them to the rubric)

• Focus references on evidence:
• “What was the candidate’s measurable impact?”
• “How did they perform under constraints?”
• “How did they make decisions and resolve conflict?”
• Require references to describe work examples, not general praise.
• Where possible, reference the same competencies used in interviews to avoid “soft” feedback.

1. Standardize scoring to reduce the advantage of polished narratives

• Use a consistent evaluation rubric across interviewers and stages:
• Competency scores + evidence rating (specificity/accuracy/ownership)
• Separate “communication” from “competence” so AI-style writing doesn’t dominate outcomes.
• Implement a calibration meeting after each hiring loop to correct bias.

1. Screen differently at each stage (example operating model)

• Stage 1: CV screen → “evidence readiness” score
• Specificity of claims, role clarity, measurable outcomes, credibility indicators
• Stage 2: Structured interview → competency + deep-dive validation
• Stage 3: Work sample/case → role-relevant capability
• Stage 4: References → verify impact and behavior under real constraints
• Stage 5 (optional for senior roles): Panel interview + simulation exercise

Business Impact

• Better quality-of-hire: you’re measuring performance potential and real problem-solving, not storytelling polish.
• Reduced hiring risk: fewer “high-potential on paper” candidates who underperform once on the job.
• Faster and fairer process: rubrics and work samples create consistent comparisons and reduce interviewer variance.

Risks / Trade-offs

• Cost and time: work samples add effort. Mitigation:
• Start with one high-signal assessment step (not many)
• Use short, role-specific tasks with clear rubrics
• Candidate experience: timed tasks can feel heavy. Mitigation:
• Be transparent about the purpose (“to evaluate real job skills”)
• Keep scope proportional to seniority
• Over-filtering: if you make the process too strict early, you may reduce funnel size. Mitigation:
• Use deep-dive verification rather than heavy testing for every role

Immediate Next Steps (practical 30–45 day plan)

1. Build a competency rubric for 1–2 priority roles

• Define 6–10 competencies max.
• For each, write “what good looks like” and “evidence to listen for.”

1. Add a Claim-to-Evidence requirement in the application

• Ask candidates to select 3–5 CV claims and provide brief evidence.
• Use this to drive interview deep dives.

1. Pilot one work-sample assessment

• Choose the stage where you currently rely most on interviews.
• Create a simple case with a scoring sheet and run it for the next hiring cycle.

1. Calibrate interviewers

• Run a 60-minute training:
• How to score evidence vs eloquence
• Example questions for validation
• How to prevent AI-biased halo effects

1. Improve reference checks

• Use a structured template tied to the same rubric and insist on examples.

Three quick questions (so I can tailor the process)

1. What type of roles are you hiring most (e.g., sales, product, engineering, operations, leadership)?
2. Where does your current process rely most: CV screening, interviews, or tests?
3. Are you hiring volume (high throughput) or key roles (quality-first, fewer hires)?

If you need help implementing these recommendations or would like expert guidance tailored to your organization, the team at Zunavish would be happy to assist.