Enhancing the Retail Customer Experience and Operational Efficiency
Background
In the rapidly evolving retail sector, innovation is paramount. Our client, a leading apparel company, recognized the necessity of embracing cutting-edge technology. Specifically, their Merchandising & Planning team identified the urgent need to leverage advanced solutions. Tasked with managing a comprehensive master data solution encompassing product, location, and vendor data across their enterprise, our client aimed to establish a scalable, efficient, and standardized framework. This framework would serve as the foundation for a robust enterprise data-as-a-service model within a federated scalable architecture.
Seeking a strategic partnership with Altimetrik, they sought assistance in enhancing the reliability, scalability, and simplicity of their master data system.
Our focus was twofold: to enhance data quality and streamline the process of ingesting and validating data, thereby reducing the time required to make it available for consumption. The overarching goal was to elevate the quality of downstream systems, ultimately enhancing the customer experience.
Implemented data filtering and notification system for issue detection.
Reduced overall data processing time from ~ 25 hours to ~6 hours, marking a significant 76% improvement.
Achieved near real-time data availability for end customers, reducing the interval from every 3 hours to immediate.
Introduced an end-to-end telemetry dashboard for monitoring, debugging, and altering data movement across the entire pipeline.
Pain Point
Despite efforts to enforce business rules on data within the existing architecture, the team faced challenges. Frequent rule changes and the constant addition of new business attributes resulted in flawed data passing through, causing significant issues for end-users.
Additionally, the process of incorporating new data was time-consuming, taking approximately 25 hours, while updating existing data required 3 hours. This delay was largely due to a monolithic pipeline generating large files encompassing entire datasets, sometimes reaching gigabytes in size.
Debugging and resolving data movement issues were time-consuming and laborious, leading to customer dissatisfaction. The sequential processing of data also strained server resources, with memory and CPU usage peaking at 90%, further delaying data publication.
Key Objectives
- Implement an additional validation layer.
- Decrease the overall time for data availability.
- Facilitate near real-time data availability.
- Develop a dashboard for visualising data movement and identifying bottlenecks.
- Enhance server efficiency
Solution
Validation Layer
Integrated an additional data validation layer just before the enterprise data exchange to effectively filter out flawed data and promptly notify the support team for swift resolution.
Parallel Processing
Fragmented the existing monolithic file into multiple smaller files, enabling parallel processing and significantly decreasing outbound file processing time from approximately 10 hours to just around 1 hour.
Real-time Data Availability
Transitioned from generating physical JSON files to directly querying data from the MDM system and transmitting it into Kafka every hour, eliminating the time required for file creation, publication, and retrieval.
Enhanced Data Tracking
Deployed a new correlation ID framework to meticulously track the flow of data from inbound to outbound, encompassing third-party systems, thereby furnishing comprehensive visibility and context for each data batch.
Optimized Queries
Enhanced efficiency by refining query logic to execute only a single query encompassing all IDs, resulting in a significant reduction in CPU usage to a mere 20%.
The Outcomes
Accelerated Data Availability
Transitioned from a 24-hour schedule to near real-time availability of master data for end consumers, ensuring they have access to up-to-date information whenever needed. Successfully reduced data errors, guaranteeing the prompt availability of accurate data to end consumers, thereby enhancing decision-making processes.
Enhanced Technical Infrastructure
Implemented architectural and design changes that effectively reduced TechDepth in the MDM system, resulting in improved data quality and optimized infrastructure utilization. These enhancements have led to improved efficiency and reliability in data processing and transmission.
Improved Data Quality
Implemented measures to enhance the quality of data transmitted to downstream systems, thereby facilitating accurate business decisions, and streamlining workflow processes. These measures have significantly contributed to the overall improvement of data integrity and reliability across the organization.
Alignment with Long-term Vision
Continued efforts are aimed at establishing a scalable, efficient, and reliable enterprise data-as-a-service system, with plans for further enhancements and adaptations to meet evolving business needs. These ongoing initiatives demonstrate our commitment to long-term success and innovation in data management and utilization.