Design of a Real-Time Pricing System for E-commerce

Abstract

The high volumes and stiff competition call for a smart (both economically and computationally) pricing systems in modern E-commerce. The three market factors, e.g., the Demand for the item, the Competition prices, and the Buyer’s characteristics, are instrumental in deciding the sale prices for items of the same quality. These factors are volatile in nature, and therefore, the ability to modify the prices in real-time gives a boost to a seller’s profitability. In this paper, we propose a novel design for a real-time pricing system to offer differentiated price suggestions. We build an algorithm that uses the three factors and the target sale quantity for decision making. The pricing is formulated as an optimization problem and is solved by using the technique of Linear Programming (LP). The key parameter in the LP equations is the sale probability, which is derived from historical price requests and ‘price request to sale’ ratio. The simulated results demonstrate that, by executing multiple subsequent optimization cycles, the pricing solution generates higher revenue than static pricing choices. In order to reduce the time required to serve price requests, we have decoupled the pricing calculation from the request paths, leading to an extremely fast pricing solution.