Case Study
Ziplingo Conquers Customer Churn with AWS Generative AI and Intelligent Automation
Industries
About Ziplingo
Ziplingo powers direct selling communication for companies around the world, helping them reinvent how they engage with their distributors and customers. Since 2008, Ziplingo has been at the forefront of the digital communications revolution, delivering intelligent, flexible, and highly personalized tools tailored specifically for the direct selling industry. Learn more at https://www.ziplingo.com.
Challenge
Ziplingo's Key Business Objectives
Ziplingo partnered with Cloudelligent and AWS to transform their churn prediction and customer retention strategy with the following core business objectives:
Streamlining data handling by removing 15+ hours of weekly manual work across CRMs, transactions, and other sources.
Accelerating insight delivery by replacing static dashboards that delayed churn interventions by over 3 hours.
Enhancing customer engagement through personalized outreach to improve a 12% open rate and 5% offer redemption rate.
Key Amazon Web Services Used
- Amazon SageMaker
- Amazon Bedrock
- Amazon Q in QuickSight
- AWS Glue
- AWS Glue Crawler
- AWS Glue Data Catalog
- Amazon QuickSight
- Amazon Athena
- Amazon Simple Storage Service (Amazon S3)
- AWS Lambda
- Amazon API Gateway
- Amazon Simple Email Service (Amazon SES)
Third-Party Tools
- Claude 3 (via Amazon Bedrock)
Solution
How Cloudelligent Solved Ziplingo's Challenges
1
AI-Driven Model Training and Deployment
Cloudelligent implemented a Random Forest churn prediction model trained using Amazon SageMaker. Customer data from multiple sources (including CSV and Parquet files containing orders, transactions, and user behavior) is automatically ingested into Amazon S3 on a scheduled basis. An AWS Glue job validates, cleans, and processes this data through feature extraction and engineering, with the cleaned datasets stored back in Amazon S3. The model achieves up to 92% AUC-ROC accuracy and is deployed to a serverless SageMaker endpoint, enabling real-time inference without provisioning infrastructure.
2
Real-Time Predictions with API Integration
A comprehensive real-time churn scoring system was built through an inference pipeline powered by Amazon API Gateway and AWS Lambda. When a user sends a POST request with a file path and bucket name, the API Gateway triggers a Lambda function that invokes the Amazon SageMaker Serverless Inference Endpoint with the input payload. This returns the churn rate along with the prediction file location, with results stored in Amazon S3 and immediately available for visualization and analysis.
3
Data Visualization and Generative BI
To make predictions accessible and actionable for business users, Cloudelligent integrated Amazon QuickSight with Amazon Athena and AWS Glue Data Catalog. An hourly AWS Glue Crawler updates the Data Catalog with new prediction files, enabling QuickSight dashboards to reflect the most recent churn data through incremental updates. With Amazon Q in QuickSight, users can ask natural language questions like "Why did churn increase in October?" and receive instant visualized results. The system also auto-generates narrative insights about top churn drivers by region or segment to support faster decision-making.
4
Hyper-Personalized Retention Campaigns
Cloudelligent deployed Claude 3 on Amazon Bedrock to dynamically generate tailored email content based on each customer's churn probability and behavioral patterns. These personalized messages are automatically sent using Amazon SES (e.g., 25% off a specific product) to retain high-risk users. This Generative AI layer significantly enhances campaign engagement by making each customer interaction more relevant and timelier, driving up to 40% improvement in campaign ROI and reducing manual content creation time.
Results & Benefits
Transforming data into actionable insights—Ziplingo's journey toward smarter, faster customer retention.
Cloudelligent’s AWS-powered solution enabled Ziplingo to implement proactive retention strategies, enhance decision-making, and improve operational efficiency. Key outcomes include:
Accelerated Decision-Making with Generative BI
Amazon Q in QuickSight reduced report generation from 8 hours to 5 minutes. Natural language querying delivers answers in seconds versus 30 minutes with manual methods. Amazon Q in QuickSight adoption increased 65% across non-technical teams.
Proactive Retention Campaigns at Scale
Real-time churn predictions through Amazon SageMaker and hyper-personalized emails via Amazon Bedrock helped Ziplingo achieve up to 25% churn reduction. The 92% accurate model targeted high-risk customers during critical decision windows, significantly improving engagement.
Enhanced Cost Optimization
Serverless AWS architecture reduced inference costs by up to 25% while maintaining scalability. AWS Glue and Amazon S3 minimized manual workloads and infrastructure overhead, allowing Ziplingo’s teams to focus on innovation rather than maintenance.
Why Amazon Web Services?
Why Cloudelligent?
Ready to Embark on an Epic Cloud-Native Journey?
ascend as an industry pioneer with Cloudelligent
right by your side.
