<p>The deep-seated contradiction between the green development of Chinese cities and the barriers to government information has catalyzed a transformation toward the digital governance paradigm centered on constructing a digital government. This study took 278 cities in China as the experimental objects and used the policy of big data management institution reform as a quasi-natural experiment. By constructing the DDML-DID model, this study examined the impact of digital government construction on green total factor productivity (GTFP). Various machine learning algorithms indicate that a one-unit increase in the digital government construction led to a significant enhancement of urban GTFP by 0.9% to 4.7%. Furthermore, policy individual treatment effects varied, ranging from − 0.1 to 0.2. Mechanism analysis indicated that the digital government construction encourages the improvement of GTFP through a dynamic chain transmission path with varying policy sensitivities, such as enhancing governmental digital attention, improving digital infrastructure construction, promoting digital technological innovation, and promoting digital economic development. Heterogeneity analysis indicated that the promoting effect of digital government construction on GTFP was more significant in ecological function areas, resource-based cities, non-two-control areas and central cities. The research conclusion of this study provides valuable inspiration for the government to break through the institutional roots that hinder the urban green development through digital transformation.</p>