Discrete choice experiments have gained importance over the years while supporting the marketing industry significantly. Several author-groups have contributed to the theoretical development of discrete choice experiments and for finding optimal choice designs under the multinomial logit model. The author-groups Street–Burgess and Huber–Zwerina have adopted different approaches and used seemingly different information matrices under the multinomial logit model. The information matrix plays a crucial role for finding optimal designs in both approaches. Since the expressions for the relevant matrices look very different and it is not obvious how the two approaches are related, this has given rise to some confusion in the literature. We resolve this confusion by showing, in general, how the information matrices under the two approaches are related.There have also been some confusion regarding the inference parameters expressed as linear functions of the systematic component of utilities corresponding to the options. We theoretically establish a unified approach to discrete choice experiments and introduce the general inference problem in terms of a simple linear function of such utilities. This allows us to show that the commonly used effects coding under the A-criterion for the non-singular full-rank inference problem inherently attaches unequal importance to the elementary contrasts of attribute levels. On the contrary, we see that the orthonormal coding leads to attaching equal importance to the elementary contrasts of attribute levels. However, for a singular full-rank inference problem involving the full set of effects-coded parameters, we show that the orthonormal coding provides an equivalent approach to obtain A-optimal designs.