Notes from the infrastructure we ship.

Decision Traces: What Multi-System Data Fusion Reveals About Institutional Knowledge in Enterprise Hiring

We examine enterprise hiring systems across three disconnected platforms — applicant tracking, human-resource information systems, and behavioral assessments. Integrating these data sources at a Fortune 500 insurance carrier (10,765 agents hired, 2022–2025) produces three findings: keyword screening from candidate profiles failed to predict job success; personality assessments achieved stronger predictive accuracy when combined with other data sources; and speed-to-production economics varied significantly based on behavioral scores. The study demonstrates that institutional knowledge can be captured and operationalized through multi-system data fusion, revealing insights invisible within isolated systems.