Two Machine Learning Approaches for Coupang’s Marketing Campaigns
The session will introduce two of the machine learning approaches used at Coupang to choose the best promotion for each customer: The Next Best Action (NBA) recommendation engine and the Purchase Propensity (PP) models. The NBA recommendation engine can automatically recommend the action that has the highest impact on customer’s long-term values. The PP models detect a customer’s interest to purchase particular products compiling hundreds of signals. The NBA engine's recommendation engine and the PP models are refreshed on a regular basis to account for new patterns in customer behaviors and customer values, which means the recommendations and predictions get more accurate each time. The machine learning models ran on top of the Coupang's in-house state-of-the-art AI platform. The machine learning approach overcomes manually administered, sub-optimal strategies built on human-driven ad-hoc rules and strategies and automatically powers Coupang's hundreds of campaigns every week.
Watch more Sessions
Wow the Customer