Authors:
Ted Tao Yuan
1
;
Michelle Cai
2
and
Daniel Kao
2
Affiliations:
1
Vipshop US, United States
;
2
Guangzhou VIP Information Technology Co., China
Keyword(s):
Newsvendor Model, Flash Sale, eCommerce, Machine Learning, Bayesian Inference, Stochastic Model Applications in Inventory Management and Automation.
Related
Ontology
Subjects/Areas/Topics:
Applications
;
Operational Research
;
Supply Chain Management
Abstract:
Daily deal, or flash sale, websites offer limited quantity of selected brands and products for a short period of time. The idea is that short-term sales event of branded products drives consumer interest. Flash sale sites like vip.com negotiate great deals from various vendors on a limited quantity of selected products. In operation, all merchandises need to be allocated to regional warehouses before a short-term sales event starts. The variety and quantity of merchandises change significantly from one sales event to another. Unsold items are typically shipped back to vendors after the sales event ends. In this paper, we discuss the design and implementation of a regional warehouse merchandise allocation model and strategy to maximize sales conversion rate. Our work reveals the uniqueness of inventory planning of flash sale and its similarity to that of general online retailers. Our machine learning prediction models and Bayesian Updating strategy are highly valuable to the improveme
nt of regional warehouse efficiency and customer experience in dealing with highly volatile flash sale inventory.
(More)