AI-ENHANCED MIS PLATFORMS FOR STRATEGIC BUSINESS DECISION-MAKING IN SMES

Abstract

This study investigated how AI-enhanced management information systems (MIS) platforms influenced strategic business decision-making in small and medium-sized enterprises (SMEs) using a quantitative explanatory model grounded in a review of 40 peer-reviewed papers. Cross-sectional survey data were collected from 352 SMEs (70.4% usable response rate) representing services (33.5%), manufacturing (26.1%), retail/trade (21.0%), logistics (10.8%), and technology-enabled sectors (8.5%); 55.7% were small firms, 31.3% medium, and 13.1% micro. Descriptive results showed moderately high AI-MIS capability (M = 5.21, SD = 0.89), led by predictive analytics (M = 5.44) and automated insight/alerting (M = 5.36), while prescriptive recommendation was lower (M = 4.88). MIS integration quality was mid-range (M = 4.76, SD = 0.93), with breadth (M = 4.92) exceeding depth (M = 4.61). Managerial usage intensity was moderately high (M = 4.98), driven by usage frequency (M = 5.24) rather than scenario testing (M = 4.68). Strategic decision outcomes were positive overall (M = 5.06), strongest for decision speed (M = 5.28) and cross-functional alignment (M = 5.26), with weaker risk calibration (M = 4.72). Correlations supported theoretical associations: AI-MIS capability correlated with usage intensity (r = 0.71) and decision outcomes (r = 0.68), and integration quality correlated with decision outcomes (r = 0.60). Reliability and validity were strong (Cronbach’s α = 0.85–0.93; CR = 0.88–0.94; AVE = 0.52–0.68; HTMT ≤ 0.79), and collinearity was acceptable (VIF = 1.58–2.13). Regression results indicated significant direct effects of AI-MIS capability on decision outcomes (β = 0.41, t = 8.72, p < .001) and usage intensity (β = 0.58, t = 12.10, p < .001). Usage intensity predicted decision outcomes (β = 0.47, t = 10.24, p < .001) and partially mediated the capability–outcomes relationship (indirect β = 0.27, 95% CI [0.20, 0.35]; residual direct β = 0.14, p = .003). Integration quality independently improved decision outcomes (β = 0.26, p < .001) and moderated capability effects on usage (β = 0.12, p = .007) and outcomes (β = 0.11, p = .014). The final models explained 54% of variance in usage intensity and 62% in strategic decision outcomes.