Hepsiburada is Turkey’s largest e-commerce retailer offering a wide range of products such as electronics, apparel, stationery, home decor, and groceries with 65 million visitors each month.
Hepsiburada was missing sales opportunities despite growing visitors and assortment due to an inability to personalize customer experiences based on real-time insights. Data fragmentation and legacy infrastructure limitations made it difficult to capture and analyze clickstream data to drive real-time insights.
Core Compete recommended and architected a new Big Data platform to help Hepsiburada capture and analyze clickstream data to drive real-time insights that could shape customer journeys and enable more targeted campaigns.
Enabled higher degree of customer personalization for offer optimization and audience selection through deep learning ML models
Modernized infrastructure to deliver much higher level of computational and storage support to meet the growing business needs
Enabled delivery of apps with advanced visualization to drive better decision-making through real-time insights
Significantly scaled the use of machine learning to drive increased personalization
CoreCompete in Action
Daily snapshots and a longitudinal view of the customer was created across 5 data sources (SAP HANA, Product Systems, Clickstream unstructured data and Campaign Response data). Hadoop workloads had to be designed and deployed to create the right views needed to scale and deploy ML models.
Deployment of advanced machine learning models for customer personalization and audience selection for campaigns with over 150,000 customer-level features. Delivered dashboards and apps to serve up insights to decision-makers with intuitive visualizations.
Technologies such as Hadoop, Hortonworks, SAS DI & Analytics, SAS Visual Analytics and other custom applications were used to deliver the solution.
Core Compete took just 2 weeks to get the new analytics infrastructure up and running and 2 months to migrate all existing data. This rapid initial phase paved the way for an accelerated migration over just a few months.
Increase in revenue resulting from greater customer personalization enabled by 800+ machine learning models encompassing a large number of features. These models along with apps tailored for merchandizers enabled them to target the right audience in campaigns and deliver real-time customer insights to decision-makers.