Working with Strata and AWS tools for Customer Retention and ML process standardization.
C&W Communications is a leading provider of telecommunications and entertainment services in the Caribbean and Latin America. Like many service providers, one of their main challenges is customer retention. They faced issues with customer churn and needed to develop a customized machine learning solution to predict and prevent involuntary churn. They also aimed to standardize their ML development lifecycle and automate their data science processes. Strata partnered with the client to build a scalable MLOps platform using AWS services, resulting in a 10% increase in retention and collection campaigns, a 90% reduction in operational tasks and errors, and 70% less effort for future releases. The solution enabled the client’s analytics team to focus on performance development and improvements while reducing deployment time by 95%.
Common Challenges Require Innovative Solutions
C&W Communications faced a serious business challenge when it noticed a spike in the number of subscribers involuntarily leaving its services. This issue could be tackled using machine learning (ML) techniques. The company needed to predict daily churn likelihood so they could design focused collection campaigns and take prompt actions to prevent churn. By doing this, they aimed not only to reduce the number of subscribers leaving but also to minimize outstanding debt.
Similarly, they faced common challenges often encountered when deploying machine learning models to production. These include difficulties with monitoring, the need for manual interventions, and scalability problems. Like other organizations, more than half of their machine learning projects could not be successfully deployed to production because of inconsistent, labor-intensive workflows. Model improvements were often made manually in response to declining performance, and scalability in the development process was limited. Additionally, the success of these projects depended heavily on key personnel, making staff turnover a significant concern.
The main obstacles that C&W Communications faced were two. The first was predicting each customer’s daily likelihood of involuntary churn to optimize Retention & Collection campaigns. The second challenge involved managing the entire lifecycle of the machine learning (ML) solution, including standardizing operational procedures to reduce dependence on staff and ensure scalability. Additionally, there was a need to enable or disable features remotely without deploying code, develop an automatic retraining process to address ML model drift, and prevent resource misuse, as staff members spent significant time on manual tasks rather than impactful business tasks.
AWS – Reliable Partner
C&W Communications has been utilizing various AWS services for its business needs, including deploying its data lakes. This prior experience with AWS has enabled seamless integration of model serving with the campaign management platform, leveraging the existing data lakes. This integration has significantly facilitated the collection of customer churn propensity scores for the Retention & Collection campaigns, streamlining the process and increasing efficiency. Additionally, AWS’s reputation for reliability, scalability, and continuous improvement has strengthened C&W’s confidence in its services and reliance on them for its business operations.
Strata and C&W Communications Joint Work
Strata has worked with C&W Communications for many years, providing Advanced Analytics and data science solutions, marketing automation for its main campaigns, and data-driven business consulting. Strata has extensive experience in descriptive and predictive modeling techniques, offering deeper insights through the use of multiple leading-edge algorithms and agile model development. They identify key data relationships and discover new patterns using statistical techniques for modeling and testing. Additionally, they deliver data visualization dashboards to help businesses understand their current situation or model results, employing an MLOps framework as a best practice.
A MLOps Approach for Greater Scalability
Strata worked closely with C&W to develop a comprehensive solution that addressed both their churn-prediction and ML-management challenges. The MLOps solution developed by Strata automates common data science processes, including data preparation, model training, and testing. Strata’s solution uses high-performance ML algorithms for data processing, prediction, monitoring, and automatic retraining, and is deployed in the customer’s cloud account. The system is serverless and event-driven, developed under agile methodologies to enable granular releases into production. Throughout the project, Strata collaborated with various C&W departments, including business, finance, analytics, and ML, to ensure the solution met the organization’s specific needs and requirements.
The solution features a machine learning model that predicts involuntary churn using a high-level, tree-based algorithm selected after extensive testing and experimentation. The architecture is fully integrated with AWS Step Functions, utilizing AWS Glue for the Data Pipeline and CWC’s DataLake as the data source. Amazon SageMaker Job handles the Model Pipelines, while Amazon CloudWatch manages logging and Amazon DynamoDB stores model metadata and process tracking. Overall, the MLOps framework offers high scalability and availability, shortens time-to-market, and facilitates standardized, easy management, operation, and maintenance of the ML solution.
The goal of the system is to perform daily inference, monitoring, metrics, and weekly retraining, as illustrated in the architecture diagram below. The daily process begins when new data for the assigned day becomes available. Each pipeline is designed to be atomic, meaning that incomplete or faulty pipelines will not produce corrupted data. Instead, they will only generate metadata for analysis. This setup allows each pipeline to be triggered again and to rewrite its output while creating new metadata.
High Level Architecture
Because of the large volume of data to process, a distributed data-computing solution was required. After careful evaluation, Glue Spark was selected as the most suitable option for the customer. The Data Pipelines run on Spark Glue Jobs, which are linked to AWS Glue Data Catalog, and the processed data is stored and partitioned in the project data bucket.
After the Data Pipelines are completed, the ML model workflow moves to several SageMaker Processing Jobs containers that perform various tasks, such as calculating model metrics, generating data preprocessing for the model, producing inference scores, retraining the model, and conducting the weekly best model selection process. The processing jobs run serverless, leveraging the scalability and flexibility of AWS SageMaker. This allows the team to easily manage the inference and monitoring stages of the ML model lifecycle in a streamlined and automated way.
Improved Model and Retention Campaign Performance – Reduced Manual Tasks
The implemented solution resulted in a significant positive impact, including a 10% increase in Retention & Collection campaigns, a 90% reduction in manual operational tasks and errors due to pipeline automation and standardized operations, and a 70% reduction in effort needed for launching in new markets because of the replicable and scalable architecture. After successfully developing the churn model for the first market (Jamaica), Strata was invited to create additional models for two other markets (Bahamas and Trinidad and Tobago). Additionally, the ML framework enables the analytical team to focus entirely on developing and improving model performance, making better use of their talent, and reducing deployment time by 95%.
The solution grants consistency through its own developed containers, offers flexibility and reproducibility for new market development, promotes reusability of components across projects, and supports scalability via the MLOps framework, monitoring, and explainability through metadata tracking.
The Collaborative Work Continues.
There is an ongoing contract between C&W Communications and Strata that involves delivering Machine Learning solutions and Marketing Campaigns, including daily releases of retention, stimulation, and collection campaigns for different audiences and purposes. Several model solutions are provided for different business issues and markets (churn predictions, no recharge prediction, top-up segmentation, and behavioral segmentation) and are executed on a daily, weekly, or monthly basis as inputs for the mentioned campaigns. This service also includes supporting the customer in analyzing and monitoring key KPIs, helping to identify potential issues or data-driven insights. Strata is in the AWS Advanced Tier and has validated its Services Path.
