A QUANTITATIVE STUDY ON AI-DRIVEN EMPLOYEE PERFORMANCE ANALYTICS IN MULTINATIONAL ORGANIZATIONS

Abstract

This study addresses the practical and ethical challenge of using AI-driven employee performance analytics to inform people decisions across heterogeneous multinational contexts. The purpose is to quantify the association between analytics capability and use and individual performance, and to test mechanisms and boundary conditions that explain when such systems relate to better outcomes. Adopting a quantitative, cross-sectional, case-based design, we surveyed employees and managers in five enterprise cloud/HRIS environments (multinational firms), yielding a pooled sample of N = 3,274 nested within organizations. A focused literature base of 39 peer-reviewed articles informed constructs and hypotheses. Key variables were AI Analytics Capability and Use, Perceived Fairness of Analytics (procedural and distributive), Manager Analytics Literacy, Organizational Culture for Data Use, and Employee Performance (Likert composites with split-source and objective KPI subsets where available). The analysis plan specified descriptives, correlations, and hierarchical regressions with heteroskedasticity- and cluster-robust inference; mediation was assessed via bias-corrected bootstrapping, and moderation via centered interaction terms with simple-slope and Johnson–Neyman probes. Headline findings indicate a positive main effect of analytics capability/use on performance after controls, a significant indirect pathway through perceived fairness (partial mediation), and reliably stronger slopes where manager analytics literacy and data-use culture are higher (positive interactions); convergent patterns held for manager-rated and de-identified KPI subsets. Implications are that technical investments alone are insufficient: organizations should pair governed pipelines, versioned metrics, and explainability with manager enablement and data-culture rituals to raise perceived fairness and translate model outputs into accepted, high-quality feedback and performance decisions across jurisdictions.