Abstract:A company’s perception of the business environment directly affects its investment decisions, innovative vitality, and long-term development. However, the same business environment does not necessarily lead to the same perceptions among businesses.Yet, heterogeneous perceptions are widespread, a phenomenon that warrants in-depth exploration. Based on policy instrument theory and the attention-based perspective.This paper examines enterprises of varying sizes, ownership structures, geographic locations, and industry types. Using a procedural grounded theory research methodology, this study thoroughly analyzes the mechanisms underlying the formation of corporate perceptions in complex business environments and constructs a theoretical model of “policy signal input—attention allocation—perception output—action feedback.” The study finds that differences in the combination of policy instruments, their coherence and enforcement capacity unleash heterogeneous policy signals, which serve as structural precursors to perceptual differences. Enterprises differentially allocate their attention through two stages—focus (policy signal focus) and interpretation (policy signal decoding)—resulting in a heterogeneous combination of perceptions regarding efficacy, cost, fairness, and predictability.Based on perceived differences, enterprises take feedback actions such as strategic restructuring, resource reallocation, and giving back to the environmental, which in turn influence policy provision and implementation, ultimately resulted in a dynamic evolutionary cycle of government-enterprise interaction. The study further identifies three representative types of enterprises, they are dividend-capturing, opportunity-responsive and market-positioning, then compares the pathways through which their perceptions are formed and the resulting differences. The conclusions offer important insights for resolving the mismatch between “policy universality and enterprises’ sense of benefit,” and provide a theoretical basis for accurately identifying enterprise needs and dynamically optimizing institutional provision.