Spark Ignited: Building Modern Marketing Team Using ML and Databricks

Rogers is Canada's largest telecommunications company, with over 11 million customers across the country. Machine learning is at the core of Rogers' marketing. We have over 100 production models covering most of our inbound and outbound marketing. Including acquisition, base management, cross-sell upsell and finally closing with churn. In this talk, I will share our experience building a modern marketing data science team using Databricks. I will discuss our migration process from an on-premises Hadoop cluster that led us to build a generic marketing data science framework that could be reused across all organizations and industries. Our framework consists of a suite of ETL jobs, a model feature score for training data, and a target store for model validation. Finally, I will also discuss a suite of ML models to address all key areas of marketing data science and discuss how various Databricks features can be leveraged to implement each element of the framework.