In quantum control experiments, it is desirable to build features into the field that address physical concerns such as simplicity, robustness, dynamical coherence, power expenditure, etc. With a judicious choice for the cost functional, it is possible to incorporate such secondary features into the field, often without altering the experimental procedure or apparatus. Through simulated closed-loop population transfer experiments, we demonstrate the benefit of carefully designed cost functionals. As specific examples, we address two common physical concerns: removing extraneous structure from the control pulse and finding robust fields. Removing unnecessary field components is critical if information about the mechanism is to be interpreted from the structure of the optimal pulse. Robust fields produce a stable outcome despite noise in the field and, perhaps, environmental inhomogeneities in the quantum system as is typical of condensed phase experiments.