Accurately assessing the multifaceted performance of urban commercial land is essential for steering consumption upgrading and refining urban spatial planning. However, existing scholarship continues to treat commercial land through a predominantly economic lens; non-economic indicators have rarely been compared with price-based metrics within a unified analytical framework. To bridge this gap, this study constructs a multidimensional benefit evaluation framework encompassing four indicators, namely, land values, shop rent, online reviews, and customer ratings, across 10,766 commercial land parcels in Chengdu, China, and applies XGBoost-SHAP methods to examine their predictive associations and spatial patterns. This study reveals the differences and commonalities among the four benefit indicators in terms of statistical characteristics, spatial distribution, and matching patterns, and further examines the ranking of their key predictors, as well as the nonlinear effects, threshold effects, and interaction effects with key factors. These findings corroborate the distinctiveness of the four benefit variables and the significance of comprehensively evaluating commercial land-use benefits. This study provides novel perspectives and empirical evidence for assessing commercial land-use benefits and their predictive mechanisms, and offers actionable guidance for formulating targeted consumption development, spatial layout, and business format optimization strategies, thereby contributing to the sustainable vitality of urban economic spaces.