Using panel data from 30 universities in Guangdong Province from 2019 to 2023 as a sample, this study employs a three-stage DEA to measure performance and utilizes fsQCA to explore the configurational effects of six types of factors. The findings reveal that the average performance of Guangdong universities is 0.426, with insufficient pure technical efficiency being the primary constraint. Government support, talent scale, and research incentives are core conditions but not necessary. Four high-performance pathways exist (government-talent support type, incentive led type, talent-cooperation composite type, and comprehensive coordination type), with the incentive-led type having the highest coverage. Scientific and technological innovation capability is absent in these pathways. Science and engineering disciplines rely more on university-enterprise collaboration, while comprehensive disciplines depend more on government-talent support. The conclusions provide a basis for differentiated policy implementation.