Targets on the data-envelopment-analysis (DEA) efficient frontier show the direction of inefficient decision-making units’ (DMUs) improvements. Some authors have correlated the improvement effects of targets with management needs. Nevertheless, targets are unattainable for DMUs with poor performance. Hence, stepwise improvement paths with different improvement effects should be provided to solve the problem of unachievable targets and flexibly meet decision-makers’ (DMs) diverse management needs in performance improvement. For this purpose, we propose novel stepwise DEA benchmarking approaches and a search algorithm to find progress-improvement strategies with different improvement effects. The contributions of our approaches are as follows. (1) Several stepwise improvement paths with different improvement effects are found, which provides DMs with more abundant alternatives for performance optimization in various management situations. (2) The upper bound for efforts in all intermediate improvement steps can be effectively controlled, which overcomes the problem of unachievable targets and makes stepwise improvement paths more feasible. To illustrate our approaches, we conducted a case study on 18 ports in China. In this case study of Chinese ports, we further highlighted the advantages of our methods by comparing them with existing DEA models. In sum, our approaches are suitable for poorly performing units that aspire to align performance improvements with management needs.