Talent Acquisition Analytics & AI Operational Health

Role: TA Lead (Read-Only Viewer)

Calm, read-only operational telemetry across automated resume screening throughput, human-in-the-loop oversight quality, and candidate progression.

Scope:

All 18 Active Requisitions

Telemetry Synced: 14m ago (Workday API)

Applications Screened +18.4% vs prev

4,820 total

Automated triage 4,120

Expedited / referral 700

Running velocity: 803 / mo

Median Time to Screen SLA 98.6%

4.2 hrs median

Reduced from 3.8 days (pre-AI baseline).

check_circle 98.6% triaged within 24h SLA target

Advance Rate Norm: 8–12%

9.4% selectivity

453 advanced to first-round recruiter screen.

Benchmark aligned: Healthy selectivity

Override Rate (HITL) Healthy Band: 4%–12%

5.8% 280 / 4,820

Upgraded to Advance 4.1% (198)

Downgraded to Reject 1.7% (82)

Active guardrail: Rubber-stamp threshold (<3.0%) inactive

policy

Governance Operating Principle: Human-in-the-Loop Oversight RULE-GOV-094

“Healthy Band: 4% – 12%. Override rates below 3.0% trigger an audit alert for potential 'rubber-stamping' (evaluators accepting AI recommendations without critical inspection). Overrides above 15% indicate rubric drift or prompt misalignment.”

Audit Protocol COMPLIANT (v2.4)

Recruitment Progression Funnel

Stage-by-stage candidate triage, human gatekeeping, and technical conversion benchmarks.

trending_up AI-screened candidates pass HM interview at 47.2% vs. 28.5% industry benchmark

STAGE 01 Applied

4,820 100% of pipeline

STAGE 02 AI Scored

4,820 100% evaluated

STAGE 03 Human Oversight

4,820 280 overrides logged

STAGE 04 Recruiter Screen

453 9.4% conv.

STAGE 05 HM Tech Screen

214 47.2% conv.

STAGE 06 Final Panel

86 40.2% conv.

STAGE 07 Offers Extended

28 32.6% offer rt.

AI Triage Breakdown (Stage 02 Pre-Human Review) N = 4,820

Advance Recommendation: 9.4% (453)

Borderline Review Queue: 15.2% (733)

Archived (Not a Fit): 75.4% (3,634)

Median Time to Screen (Trailing 6 Months)

Screening latency from application ingestion to human confirmation status.

Observed Latency (Hours)

Target SLA Threshold (<12h)

Historical baseline: Trailing 6 months shows 71.6% reduction in review queue cycle time. Status: Highly Stable

Recruiter Override Distribution

Human-in-the-loop variance monitoring by evaluation owner.

Zero Flags

Evaluator	Screened	Overrides	Rate (%)	Primary Override Reason
Riya Sen Staff Recruiter	1,840	114	6.2%	Disputed Raft / Concurrency tenure
Alex Chen Sr. Recruiter	1,420	68	4.8%	Alternative degree / bootcamp equivalency
Kenji Sato Lead Technical Sourcer	980	71	7.2%	Open source portfolio verified
Marcus Vance Hiring Manager (Infra)	580	27	4.7%	Niche domain architecture depth

All 4 evaluators within calibrated 4.0% – 12.0% governance corridor. Rubber-stamping: None

Shortlist Quality & Manager Alignment

Downstream signal validity and talent partner operational return.

Audited Weekly

HM Shortlist Acceptance

92.4% +14.1% YoY

Marcus and Elena accepted 92.4% of recruiter-advanced candidates without dispute.

Screen False Positive Rate

3.8% Low Error

Candidates advanced who failed foundational screening rubrics.

Candidate Appeal Rate

0.64% 31 appeals

8 appeals upheld after independent secondary human review.

Capacity Liberated

~18.5h / recruiter / wk

Direct time redirected toward candidate cultivation and closing.

Overall AI Quality Index: 96.2 / 100 (A+ Institutional Grade) Model ID: v3-claude-audit-70b