Due to recent advances in computational power and sophisticated software, modeling many complex systems has come within reach for the first time, promising understanding these systems better than ever before. A prime example of the impact of these advances is on the use computational models to construct performance landscapes. As interpretive tools, performance landscapes are incredibly powerful for interrogating evolutionary trends, competitive dynamics in an ecosystem, and even niche partitioning. If carefully crafted, these landscapes can lead to important insights in the evolutionary systems that they model. In this synthesis, we cover best practices for constructing performance landscapes, including: selecting parameter inputs and outputs, building morphospaces, constructing physical models, parameter spaces and surrogates, and performance space analysis. We conclude with a call for additional work linking biology and mathematics around the unique challenges presented by performance-space construction and validation.