Does Congestion Always Hurt? Managing Discount Under Congestion in a Game-Theoretic Setting

We study the problem of a cloud service provider (e.g., Amazon Web services) selling cloud services to a buyer (e.g., Uber) under congestion. Cloud congestion has received significant attention in recent years, leading to reduced service quality and a potential loss of customers (e.g., riders and drivers of Uber) for the buyer. One potential way to mitigate the negative effect of congestion is to offer discounts. Discount has been a useful mean to stimulate demand and reward customer loyalty. However, in the presence of congestion, the choice of discount presents additional challenges. A higher discount has both direct and indirect impacts on demand. On the one hand, a higher discount leads to higher demand (i.e., direct effect), but on the other hand, a higher demand leads to increased congestion, which in turn, lowers the demand and hurts profitability (i.e., indirect effect). Given that customers are both price and congestion sensitive, the choice of optimal discount under congestion is therefore not straightforward. We study this dynamics between congestion and discount, and explore how congestion moderates both buyer's response and seller's discount strategy. We address this issue under two popular types of contracts - quantity discount contract and market share contract. While quantity discount contract is widely practiced, market share contract is comparatively less popular. However, our results show that a market share contract performs better when the customer's discount and congestion sensitivity are both high. We also find that congestion does not necessarily hurt the buyer when customers are more intolerant to congestion, and a seller should not necessarily undertake aggressive price cuts when customers are more discount sensitive. These findings have important managerial implications as customers widely vary in their sensitivities and a suboptimal discount can pose serious business risks in a highly competitive market. In addition, we also study the buyer's problem of sourcing from multiple cloud vendors where vendors vary in their pricing and congestion susceptibilities. Our results also provide useful insights into buyer's optimal sourcing strategy when customers are both congestion and discount sensitive