In this paper, we will present Helm, which is the name of one of Comcast's CDPs that manages data ingestion, processing, and consumption of anonymized customer data for purposes of advanced issue resolution. We share our experiences to understand what it takes to effectively build and maintain a CDP.
The Helm platform ingests hundreds of event datasets from multiple domains, such as customer care, audit, billing, device activations and operations. To put the scale in context, the platform captures tens of billions of customer interactions from millions of customers every month. Internally, it is directly used by tens of thousands of employees to serve several hundred thousand customers, every month. An important component, Helm application programming interfaces (API), play a crucial role across several products in Comcast, serving several billions of API requests every month.
The paper is organized as follows: we will begin by introducing some of the challenges in building, maintaining, and governing CDPs. We then present an overview of the Helm CDP and its impact on customer experiences. This is followed by the platform architecture and design principles. We conclude the paper with a look at Helm’s journey so far, and possible future states.