Decoding the Conversion Gap in SME Digital Transformation: A Causal AI Framework

Abstract

Despite the proliferation of digital integration initiatives, many Small and Medium-sized Enterprises (SMEs) remain trapped in a persistent “Conversion Gap,” where digital adoption fails to manifest as tangible financial performance. Grounded in Resource Conversion Theory, this study anatomizes the structural bottlenecks of this process through a multi-stage Causal AI architecture. Utilizing time-lagged data from 649 SMEs to control for endogeneity, I integrate Gaussian Mixture Modeling (GMM), Tiered Grand-DAG algorithms, and Necessary Condition Analysis (NCA) to decode the non-linear trajectories of value realization. The findings identify a “Low Integration” cohort (34.2%) that fails to translate digital usage into realized outcomes due to a severe deficit in Absorptive Capacity (ACAP). Crucially, NCA diagnostics reveal that ‘perceived usefulness’ serves merely as a necessary baseline condition, whereas ‘user satisfaction’ functions as the primary catalyst for value conversion. Furthermore, multi-group analysis (MGA) confirms that for the most vulnerable SMEs, the causal pathway to revenue is structurally severed (β = 0.000), rendering traditional, linear training interventions ineffective. I propose a fundamental shift toward data-driven, targeted interventions to address these specific structural barriers and facilitate sustainable digital value creation in the SME ecosystem.