A decision framework for procurement fraud detection: Wisdom from academia and industry

Abstract

Procurement fraud poses significant financial and reputational risks to organizations, yet existing efforts to address it are fragmented between academia and industry. While academic research has proposed sophisticated fraud detection models using machine learning and data analytics, these solutions often lack practical applicability due to limited guidance for practitioners. In this study, we address this gap by proposing a decision framework for procurement fraud detection, synthesizing insights from existing literature and a real-world data analytics project with a global brewing company. We identify three critical decision problems in developing fraud detection models: (1) constructing fraud indicators, (2) determining the aggregation level, and (3) selecting the model validation method. By evaluating alternatives for each decision, we offer practical solutions that organizations can tailor to their unique procurement processes and risk profiles. The proposed framework combines the knowledge from literature and practical insights, offering actionable guidance for practitioners while bridging gaps between academic research and industry practice. This study contributes to the field by formalizing decision-making challenges in procurement fraud detection and fostering collaboration between academia and industry.