Given the inherent benchmarking capability of DEA against best practices, we focused primarily on comparing the improvement targets generated by two DEA models: the conventional additive (ADD) model and the least-distance model. This analysis is performed using a time-series dataset comprising 86 retail companies in Japan. The main contributions of this study are summarized as follows.

(1)   We conduct a comparative analysis between the improvement targets generated by the conventional ADD and the least-distance models.

(2)   We investigate the effectiveness of the improvement targets generated by the two types of models by evaluating the efficiency of these targets in the subsequent year.

Based on the results of the numerical experiments and analysis, it appears that the easy-to-achieve improvement targets generated by the least-distance model demonstrate higher effectiveness.

Further exploration of improvement targets generated by other variations of the least-distance DEA models would be beneficial through additional numerical experiments using real-world or simulated data. Additionally, conducting empirical studies by applying the least-distance DEA models to diverse fields for efficiency improvement would contribute to the existing body of research.