In today's "experience economy," service providers increasingly emphasize creating memorable, and noteworthy experiences, a crucial aspect of which entails the schedule of activities a service package comprises. In this thesis, we study the experiential-service-design problem to improve the customer's retrospective perception over the service encounter. We investigate both the service-bundle design problem, which focuses on the activity-date-bundle assignment issue and the service-package design problem, which discusses a general model to support both the empirical and normative studies.

Initially, we study the service-bundle design problem. The service bundle is a collection of services sold together. Recent research indicates that the ex-post perception of a service bundle ca...[ Read more ]

In today's "experience economy," service providers increasingly emphasize creating memorable, and noteworthy experiences, a crucial aspect of which entails the schedule of activities a service package comprises. In this thesis, we study the experiential-service-design problem to improve the customer's retrospective perception over the service encounter. We investigate both the service-bundle design problem, which focuses on the activity-date-bundle assignment issue and the service-package design problem, which discusses a general model to support both the empirical and normative studies.

Initially, we study the service-bundle design problem. The service bundle is a collection of services sold together. Recent research indicates that the ex-post perception of a service bundle can be measured by the sum of four sequence effects (i.e., peak effect, end effect, spread effect and the trend effect) of the experience profile, and the ideal service bundle is often with early peak, high end utility, and positive trend.

While the ideal bundle is straightforward to be described, it can be a complex optimization problem to offer the optimal bundles with activity-date-bundle assignment decision. We investigate the service-bundle design problem by analyzing several cases of the problem and discussing the solution methods for each of these cases. We note that some of these cases are relatively easy to solve, while others present more of an issue; these cases belong to the 0-1 sum of quadratic constrained quadratic ratio problem and it is a rather complicated optimization problem because of the non-convexity. For these complex cases, we introduce two methods to solve it optimally. The first one is the linearization technique, which transforms the 0-1 sum of ratio problem into a mixed integer linear programming problem. The second one is a geometric branch and bound algorithm, which solves the problem by optimizing several subproblems. We prove our geometric branch and bound algorithm can converge in the finite steps, and the numerical study reveals the geometric branch and bound algorithm is more efficient than the linearization technique. Our geometric branch and bound algorithm can also be an adequate heuristic when the size of the problem does not allow to solve it optimally.

The second study concerns the sequence decision in service-package design. In this study there is only one service encounter with multiple activities. The empirical study on experience profile suggests that an interior peak in the optimal service package; that is, the most engaging activity (aka peak activity) is scheduled neither at the beginning nor the end of the package. Theoretic literature, however, points to a U-shaped schedule such that the peak activity should be scheduled either at the beginning or the end.

To bridge this gap, we construct a normative model that can support both the studies. Our model is based on two psychological phenomena, acclimation process, and non-homogeneous memory decay process. We find, surprisingly, that the heterogeneity in memory-decay processes is sufficient to explain the phenomenon of interior peaks, which has been observed in practice but remains to be explained by the extent behavioral models. We show an interior peak is optimal when the memory-decay rate of the peak activity is neither too high nor too low. We characterize structural properties of the optimal service schedule, building on which we develop a novel dynamic programming algorithm that optimally solves the service-package design problem in pseudo-polynomial time. We demonstrate that when an interior peak is optimal, the service provider can schedule a "low point" (i.e., a very low-utility activity) right before the peak activity, creating an unexpectedly memorable service experience. In addition to explaining the phenomenon of interior peaks and other empirically observed service design patterns, our model sheds light on service design in the presence of the heterogeneity in memory processes. For instance, as the peak activity becomes more memorable, one might expect its optimal start time to monotonically decrease. Our results, by contrast, show such a start time may either increase or decrease.