6. Conclusion and Outlook
Choice-based revenue management has received a considerable boost in attention over the past decade, in particular assortment optimization has recently been intensively worked on. Various choice models have been investigated with the aim of establishing efficient solution approaches to the assortment optimization problem, which is important for RM control policies since these problems need to be solved in real-time in order to dynamically generate offer sets. Usually, a particular choice model was first analyzed on its own, and subsequently in conjunction with certain constraints on capacity, allowable prices, consideration set structures, etc. We expect that more work will appear in the coming years on this, using less common choice models such as the paired combinatorial logit model (Zhang et al. (2017)). In particular, we see that there is scope for future research on multi-stage choice models that have seen a lot of attention in the marketing literature but almost none in revenue management. As discussed above, in marketing, two-stage models (also called consider-then-choose models) are common, and there is also some work on multi-stage models that address the situation where customers’ consideration sets may change over several stages.
Modeling of choice processes that incorporate several stages are also of interest in the context of incorporating ancillary revenues into the optimization. Customers can be modeled to first choose a product, and in a second stage to choose ancillary services; Bockelie and Belobaba (2017) propose a sequential consumer choice process along these lines. The increasing importance of ancillary revenues in traditional RM industries like airlines, car rentals, casinos, or cruise lines has so far not been adequately reflected in the academic literature.
