Abstract
An opaque product refers to a subset of products that only differ in some secondary attribute (such as color or style). After a customer purchases an opaque product, one of the products in the subset is then allocated to the customer. Selling opaque products has become popular on e-commerce platforms, since it is a source of flexibility that reduces demand variability and hence inventory cost. In this paper, we seek to understand how the design of opaque product configurations and the fraction of customers choosing opaque products impact the value of flexibility arising from opaque selling.
The key finding from our study is the following insight: a modest opaque selling strategy where (i) the opaque options contain small subsets and (ii) only a small fraction of customers buy opaque products, achieves substantial cost savings (compared to non-opaque traditional selling). Moreover, the cost savings of this modest opaque strategy is on the same order as a fully flexible scenario where all customers buy an opaque product. Our analysis relies on exploring a novel connection of multi-product inventory management to the balls-into-bins framework along with a simple balancing policy for opaque selling. As a byproduct of our analysis, we show that the simple balancing policy is asymptotically optimal. For extensions outside of our theoretical analysis, we conduct extensive numerical experiments to show that our core insights continue to hold.
The key finding from our study is the following insight: a modest opaque selling strategy where (i) the opaque options contain small subsets and (ii) only a small fraction of customers buy opaque products, achieves substantial cost savings (compared to non-opaque traditional selling). Moreover, the cost savings of this modest opaque strategy is on the same order as a fully flexible scenario where all customers buy an opaque product. Our analysis relies on exploring a novel connection of multi-product inventory management to the balls-into-bins framework along with a simple balancing policy for opaque selling. As a byproduct of our analysis, we show that the simple balancing policy is asymptotically optimal. For extensions outside of our theoretical analysis, we conduct extensive numerical experiments to show that our core insights continue to hold.
Original language | English |
---|---|
DOIs | |
Publication status | E-pub ahead of print - 19 Nov 2019 |
Externally published | Yes |