Target enrichment is used in a wide range of applications. Generation of high performance panels is an involved process that needs to take into account multiple factors, such as GC, sequence content, panel size and production variability. Here we describe the experiments and analysis undertaken to improve our design principles for high-efficiency target enrichment. Our design objective is quantitative optimization of key capture performance metrics. Towards that goal we assessed, both computationally and experimentally, multiple factors: sequence complementarity, target context, thermodynamics and our production process. We show that our design process results in high performance first-pass panels, and that for particularly difficult custom panels we are able to improve the performance through our Design-Build-Test-Learn (DBTL) cycle approach. Finally, we show how our panel design principles and data can be combined to address current and emerging target enrichment applications.