<![CDATA[This research investigates the factors influencing consumer purchase decisions within the Indonesian fashion Micro, Small, and Medium Enterprise (MSME) sector on major e-commerce platforms. Beyond testing isolated predictors, the study develops an integrated view of online purchase decision-making in which demographic characteristics operate as consumer predispositions, while discounts, free shipping, prices, ratings, reviews, and followers function as platform-based stimuli. A quantitative scenario-based stated preference design generated 4,500 observation points from 300 active shoppers. Predictive modeling was conducted using Logistic Regression, Random Forest, and Gradient Boosting, supported by SHAP (SHapley Additive exPlanations) to interpret feature contributions. Gradient Boosting achieved the strongest classification quality (ROC-AUC: 0.660), while SHAP analysis shows that age group, discounts, and free shipping are more influential than product ratings in shaping purchase likelihood. The limited role of ratings indicates a ceiling effect: when ratings are uniformly high, they become a basic trust requirement rather than a decisive differentiator. The study contributes conceptually by showing that fashion purchase decisions in saturated Indonesian marketplaces are shaped by the interaction between generational readiness, transactional incentives, and saturated trust signals. Managerially, the findings suggest that MSMEs should combine age-specific targeting, calibrated discounts, free-shipping bundles, and optimized storefront cues rather than relying solely on ratings or product quality claims.]]>
