The abundant nanopores in shale make it an unconventional reservoir for oil and gas. Matrix compressibility and heterogeneity are crucial for fluid accumulation and migration, yet these two properties remain unclear in shale reservoirs. Herein, we investigated the matrix compressibility, multifractal properties, and their geological governing factors for nanoporous Yanchang shale from the Ordos Basin, China, utilizing petrophysical experiments, mercury injection capillary pressure (MICP), and low-pressure gas adsorption. Furthermore, a valid method was proposed to correct MICP data point by point. The results show that matrix compression in shale becomes apparent at onset pressures of 12.4–42.1 MPa during mercury injection, which can be assessed by compressibility coefficients varying within (3.85–18.77) × 10–5 MPa–1. Notably, the original MICP results overestimate the actual pore volumes due to the compression-induced error, within 16.19–102.95%, leading to potential misinterpretation of the pore size distributions, particularly for pores below 100 nm. Well-developed nanopores significantly contribute to the matrix compressibility. Among the shale compositions, organic matter (OM) is the primary controlling factor for matrix compressibility, with a positive effect. Ductile clay minerals promote matrix compressibility, whereas rigid brittle minerals resist matrix compression. Our findings also reveal the multifractal nature of pore structures in shale, with an upward-convex parabolic shape for multifractal singular spectra α–f(α) and an inverted “S” shape for the generalized dimension spectra q–D(q). Singularity spectrum width and Hurst exponent offer effective solutions to quantify the pore heterogeneity and connectivity. Pores over 100 nm tend to be relatively homogeneous and well-connected, while pores smaller than 100 nm increase heterogeneity and limit connectivity. OM, quartz, and easy-dissolve minerals positively impact pore heterogeneity but negatively affect connectivity. Nevertheless, clay minerals display no obvious correlation with pore heterogeneity and connectivity. This study provides valuable insights for predicting dynamic permeability and evaluation of unconventional reservoirs.