TruConnect Delivers Affordable Connectivity for Underserved Customers with Quantiphi and AWS

TruConnect, a nationwide wireless service provider, is bridging the digital divide. It offers subsidized wireless services to eligible low-income consumers through benefits provided by the Federal Communications Commission’s (FCC) Lifeline Program and the Affordable Connectivity Program (ACP). TruConnect sought to improve subscriber retention and streamline customer acquisition, vetting, and onboarding processes. By partnering with Quantiphi, an AWS Premier Partner, TruConnect deployed an AI-powered solution that automates data extraction from customer documents and creates a central data repository that enables rapid analytics and reporting. This approach lays the foundation for an innovative Customer 360 solution and predictive churn modeling.

Given the surge in subscribers resulting from the pandemic and introduction of the ACP benefit, TruConnect required a cost-efficient customer evaluation solution that was compliant with federal program regulations. With its infrastructure built on Amazon Web Services (AWS), TruConnect  collaborated with AWS Premier Partner Quantiphi. As an applied AI and data science software and services company, Quantiphi was onboarded to automate the process of extracting customer information from various sources, such as drivers’ licenses, and securely store this data in a centralized data repository. 

To actively evangelize the new ACP program benefit and to target the right customers, Quantiphi helped TruConnect leverage demographic data that efficiently identified new markets and reached relevant underserved customers. Powered by Quantiphi’s document AI processing solution, TruConnect’s data extraction capabilities achieved a 97% accuracy rate and reduced the overall document pipeline to less than a minute. This dynamic solution operates on Amazon Simple Storage Service (Amazon S3) for storing all levels of processing data. AWS Lambda is used for triggering pipeline modules, Amazon Textract for extracting text components from documents, and Amazon SageMaker for the end-to-end deep machine learning models development.

Subscriber usage is dynamic and variable, therefore data and network usage are top-of-mind costs for TruConnect. Since these expenses directly impact the bottom line, the company sought a cost-efficient solution to manage and optimize subscriber data usage. Historically, TruConnect followed an arduous and expensive process requiring manual segmentation for hundreds of customers daily. After engaging Quantiphi and leveraging AWS technology, TruConnect was able to harness the power of advanced analytics and machine learning (ML) to segment customers into buckets. This enabled the prediction of customer usage and the implementation of a simulation pipeline comparable to the actual production pipeline. The simulation pipeline ingests data from Amazon Redshift, a cloud data warehouse, using AWS Lambda, and Amazon Glue, a serverless data integration service, as well as several other AWS services.

Travlin McCormack, chief technical officer at TruConnect, elaborates on the operational transformation resulting from Quantiphi and AWS, “The Quantiphi solution running on AWS allows us to move hundreds of thousands of users between these buckets to ensure that their usage matches the bucket they're in. It is helping us drive our costs down.”

TruConnect achieved an estimated monthly savings of $100,000 per month from digitally transforming its bucket allocation and replacing its manual processes.

Customers are at the heart of TruConnect’s subscription-based business, requiring them to have a consistent pulse on their customer base. Leveraging the Amazon SageMaker ML platform, Quantiphi designed a churn prediction model for TruConnect to drive loyalty and key insights. The ML model was developed in Amazon SageMaker Studio and uses the Amazon SageMaker pipelines system for inference and deployment. The inference pipeline also includes AWS Lambda data fetch in batch and Amazon API Gateway handles real-time inference requests. 

The model provides predictions on customer retention based on existing customer data. Information can include how long a customer has been with TruConnect, their geographic location, what kind of device they are using to access the service, or whether they were previously a subscriber. McCormack explains how the churn prediction model is laying the groundwork for a comprehensive Customer 360 solution for TruConnect: “This is the foundation of our Customer 360 solution because it tells us what essentially the customer satisfaction rate is based on the usage and the patterns that we're seeing.” The churn prediction model also has a real-time predictor score to estimate the longevity of a given applicant at the time of sign-up. 

Prior to implementing a churn prediction model, TruConnect had to estimate customer network usage manually, and as a result, inaccuracies would prove to be very costly. However, now with the AI-powered churn prediction solution they can save an estimated $1 million per month on excess network expenditures. TruConnect aspires to continue to bring connectivity to underserved populations across the United States and is excited to continue enhancing its Customer 360 Solution to better serve its customers.