An American fast-food chain was looking to improve its customer engagement and customer lifetime value with personalized email marketing that can easily scale across brands and geographies. It faced data quality issues due to CRM migration and lacked customer segmentation or means to know which offers were more profitable, or which had a better ratio of redemption/sends, etc.
Sigmoid built a scalable test-and-learn architecture to put the personalized recommender ML model in production. We developed and productionized the Multi-Armed Bandit model by automating pipelines in AWS, which updated a CRM platform to trigger personalized emails to end customers.
Using the MAB model, the client achieved 8% sales uplift by productionizing ML models to send over 100MN personalized emails on a fortnightly basis