Service design optimizers benefited greatly from the development of conjoint analysis, which appeared more than forty years ago. Conjoint analysis is a powerful and popular method to estimate consumers’ preferences. As this method is based on a survey, the estimated utility functions can be subject to inaccuracies. In this paper, we propose a service design optimizer that combines robust programming and conjoint analysis. This permits one to identify the optimal service, even when the utility functions are subject to uncertainties.