The performance of a chemical reaction is critically dependent on the electronic and/or geometric structures of a material in heterogeneous catalysis. Over the past century, the Sabatier principle has already provided a conceptual framework for optimal catalyst design by adjusting the electronic structure of the catalytic material via a change in composition. Beyond composition, it is essential to recognize that the geometric atomic structures of a catalyst, encompassing terraces, edges, steps, kinks, and corners, have a substantial impact on the activity and selectivity of a chemical reaction. Crystal-phase engineering has the capacity to bring about substantial alterations in the electronic and geometric configurations of a catalyst, enabling control over coordination numbers, morphological features, and the arrangement of surface atoms. Modulating the crystallographic phase is therefore an important strategy for improving the stability, activity, and selectivity of catalytic materials. Nonetheless, a complete understanding of how the performance depends on the crystal phase of a catalyst remains elusive, primarily due to the absence of a molecular-level view of active sites across various crystal phases. In this review, we primarily focus on assessing the dependence of catalytic performance on crystal phases to elucidate the challenges and complexities inherent in heterogeneous catalysis, ultimately aiming for improved catalyst design.