With the advent of the refractive era of cataract surgery, postoperative residual refractive error has become a key factor affecting visual quality. The accuracy of intraocular lens(IOL)power calculation is affected by multiple factors, including ocular biological parameters, calculation formulas, and lens constants. Currently, the lens constants widely used in clinical practice are mostly based on Western populations, which have a mismatch with the ocular characteristics of the Chinese population, such as shorter axial length and steeper cornea. Therefore, the personalized optimization of the constant has become a research hotspot. This review primarily summarizes the research progress on lens constant optimization in improving the accuracy of IOL power calculation. Currently, there is no consensus on lens constant optimization methods. Single-constant formulas can be optimized through iterative methods, while multi-constant formulas require the combination of linear or nonlinear strategies. Simplified gradient descent and data-driven techniques offer new optimization pathways, but cross-validation is needed to evaluate their performance. Segmented optimization based on axial length, corneal curvature, and anterior chamber depth has shown significant effectiveness in patients with special anatomical structures, effectively reducing postoperative refractive errors in different patient groups, but limitations remain in some extreme cases. The interaction of multiple parameters significantly impacts the predictive accuracy of complex cases. While new-generation formulas integrating multiple parameters or incorporating AI algorithms have improved accuracy, constant optimization still holds value. Currently, the complex relationships between multiple parameters and intraoperative dynamic changes require further in-depth research. Future research, including targeted population optimization studies, the development of real-time monitoring technologies, and innovative IOL designs, may make achieving zero postoperative refractive error a possibility.