Silent Attrition: A Methodological Framework and Simulation-Based Feasibility Study for Detecting Hidden Worker Disengagement

Silent attrition refers to the gradual and largely invisible disengagement of employees who remain in an organization while exhibiting declining participation, collaboration, and behavioral commitment prior to resignation. This study proposes a methodological framework for analyzing silent attrition using digital-behavior indicators, anomaly detection techniques, HR-context signals, and supervised machine-learning models. Due to the lack of access to real enterprise workforce behavioral data, a synthetic dataset is constructed to simulate hypothesized pre-resignation behavioral patterns informed by prior research. Within this controlled simulation environment, the framework operationalizes a composite Silent Attrition Index (SAI) and a forward-looking attrition Risk Score (RS) intended to flag elevated exit risk up to 90 days in advance. The results demonstrate the internal feasibility of combining multi-signal behavioral decline, anomaly scores, and HR events into interpretable risk indicators under assumed conditions. However, the findings do not constitute empirical validation of real-world predictive performance. The primary contribution of this work lies in formalizing a reusable analytical framework and simulation-based testbed to support future empirical validation using real organizational data.