||The turbomachinery aerodynamic design process is characterized both by its complexity and the reliance on designer experience for success. Complexity has led to the design being decomposed into modules; the specification of their interfaces is a key outcome of preliminary design and locks-in much of the final performance of the machine. Yet preliminary design is often heavily influenced by previous experience. While modularity makes the design tractable, it complicates the appropriate specification of the module interfaces to maximize whole-system performance: coupling of modularity and designer experience may reduce performance. This paper sets out to examine how such a deficit might occur and to quantify its cost in terms of efficiency. Two disincentives for challenging decomposition decisions are discussed. The first is where tried-and-tested engineering “rules of thumb” accord between modules: the rational engineer will find alluring a situation where each module can be specified in a way that maximizes its efficiency in isolation. The second is where there is discontinuity in modeling fidelity, and hence difficulty in accurately assessing performance exchange rates between modules. In order to both quantify and reduce the potential cost of this coupling, we have recast the design problem in such a way that what were previously module interface constraints become key system design variables. An example application of our method to the design of a generic turbofan core compression system is introduced. It is shown that nearly one percentage point of the equivalent compressor adiabatic efficiency can be saved.