Retailing with Opaque Products
42 Pages Posted: 11 Sep 2015 Last revised: 14 Jul 2020
Date Written: September 11, 2015
A product is said to be opaque when one or more attributes of the product are hidden until after it has been purchased. This study considers the management and value of opaque products in the context of online retail, where opaque products are commonly offered by hiding the color or style of a product. We present the first results on dynamic inventory management problems with opaque products, which feature a new type of decision on how to allocate opaque products. We consider a model where customers arrive dynamically and choose between two substitutable products and an opaque product. Ordering inventory incurs a joint replenishment cost while storing inventory incurs holding costs. We use average cost dynamic programming to analyze policies for opaque product allocation, and we use a Markov chain analysis to study the value that opaque products provide in comparison to a setting without opaque products. We also consider conventional discrete choice models to capture the joint effects on revenue and cost from opaque products.
Under symmetric demand, we show the optimal policy for opaque product allocation myopically balances the inventory levels of the two products. We also precisely characterize the cost savings benefit as a function of the order-up-to level, which can be significant even when the opaque demand is relatively small. We also show that substantial cost savings can be achieved without sacrificing significant revenue to attract opaque product demand. Our results imply that selling opaque products can be done using simple policies and may yield sizable profit increases for online retailers, often better than a dynamic pricing strategy. Moreover, this advantage only requires a relatively small fraction of customers to purchase opaque products, and does not require changing the inherent ordering and pricing policies.
Keywords: opaque products, inventory management, substitutable products, choice models
Suggested Citation: Suggested Citation