Codifying construction supply chain responses under COVID-19 into a decision tree: structured decision-making under extreme disruptions

Purpose The construction industry is highly vulnerable to Black Swan Events (BSEs), as demonstrated by COVID-19 and other recent global disruptions that exposed weaknesses in procurement and supply chain management. This study develops and validates a decision tree framework to support procurement decision-making during disruptive events in construction supply chains. Design/methodology/approach Semi-structured interviews with industry professionals generated qualitative data on procurement practices before, during and after COVID-19. Interview data were systematically coded into decision criteria and strategies, then structured into a decision tree model. The model was trained and tested using a 70–30 data split and evaluated using classification metrics such as accuracy, precision, recall and F1-score. Findings The results show how key decision points, such as supplier reliability, procurement timing and risk-sharing mechanisms, shape procurement outcomes during disruptions. The model captures expert reasoning and helps reduce subjectivity in procurement decisions, while indicating that prediction accuracy can decline when multiple external shocks overlap. Research limitations/implications Model performance may be constrained in contexts where disruptive drivers overlap or evolve rapidly, thereby reducing predictive clarity. Future research can extend this framework with larger and more diverse datasets, additional disruption scenarios and hybrid/ensemble models to strengthen generalizability and decision support across project types and regions. Originality/value This study offers a structured, data-informed decision-support tool that operationalizes expert procurement reasoning for BSE conditions. It provides practical value for resilience planning in construction supply chains and establishes a foundation for further research on structured decision-making under extreme uncertainty.