Causality-Centred Deep Learning–DEMATEL Framework for Technology Adoption in Turkish SMEs

This study advances an innovative methodological framework that unites deep learning, Decision-Making Trial and Evaluation Laboratory (DEMATEL), and agent-based modeling (ABM) to more accurately diagnose and address the diverse operational constraints confronting Turkish SMEs. Unlike conventional, static analyses, the proposed approach first employs factor analysis and a deep neural network to pinpoint the most pivotal performance drivers. Next, DEMATEL reveals how these drivers exert causally directed influences on other domains, such as production, marketing, and market research, thereby distinguishing net “influencer” factors from net “receivers.” Finally, ABM simulates the dynamic interplay among SMEs, each featuring unique resource endowments and strategic behaviors, under varying economic and policy scenarios. We combine prioritization (DL+SHAP), causal mapping, and dynamics into a single, transparent pipeline, and synthesize strategies via scenario-based SWOT. This integrated process uncovers high-impact levers for enhancing overall performance, demonstrating that targeted interventions in technology and finance can yield widespread improvements in other challenge areas. By converging advanced machine learning with systematic causal analysis and temporal simulation, the framework furnishes a more comprehensive, data-driven basis for strategic decision-making, offering policymakers and managers deeper insights into fostering SME competitiveness and resilience.