A mid-size Canadian enterprise company developing subscription and usage billing and rating solutions for data, voice, SMS and API usage. The company’s SAAS platform allows its customers to create robust billing solutions that are device/solution agnostic, with customizable business rules to rate traffic, categorize charges and define a variety of service and account levels.
The client platform ingests a variety of usage information across a number of different sources. Traditionally, billing solutions have been designed to enable billing and rating on a per service basis and requires its customers to configure and update multiple accounts to handle those services. The platform has the ability to send notifications on a per service basis, set renewal terms for individual services etc. Increasingly, customers of the client are looking to generate insights into end-user usage across multiple services. The customers want to develop correlations across multiple data services in order to optimize billing their users, and upsell/recommend packages to their users and improve retention.
Crater Labs implemented a recommendation system, leveraging a graph convolutional network. By representing the relational data within its data warehouse as a graph, Crater Labs is able to build associations across the client’s services, seamlessly establishing usage patterns across services for each of the customer’s users through a single model. With our machine learning solution, billing and rating can factor data, voice, SMS and API services simultaneously, and generate insights into how the Client’s customers can optimize their billing and rating plans. Currently, our solution specifically makes recommendations on picking optimal data plans based on cross-service usage, and also predictively identifies the end-users who are at risk of churn.
Crater Labs created a graph-convolutional neural network that immediately established linkages across the billing and rating services provided. Without having to modify its data warehouse, Crater Labs’ was able to develop a model that provides the Client’s customers with real-time insights into recommendations it can make to its customers to optimize billings, commitments, and provide pro-active advice to minimize cost while maximizing service levels. In addition, our model is able to provide real-time signals as to when a customer may leave.
Benefits and ROI
As a part of a pilot project run earlier this year, the solution Crater Labs developed was able to successfully identify situations indicating user churn, and also make accurate recommendations on billing and rating plan changes to increase end-user satisfaction. The client expects a double-digit ROI on its investment in the project.