This document discusses discrete choice experiments for designing marketing surveys. It describes how discrete choice experiments work, including using choice sets with profiles that combine different attribute levels. An example choice set for marketing a new laptop is provided. Statistical models like the multinomial logit model are used to analyze results. The document also discusses optimal design criteria like D-optimality and Bayesian optimal designs that account for unknown parameters. Computational methods like Mysovskikh quadrature are presented for efficiently calculating Bayesian D-optimal designs. The document concludes with discussing how discrete choice experiments require nonlinear model design methods and how Bayesian robust designs and software can help implement such studies.