Skip to main content
CUSTOMER STORY

Ushering in a new era of personalized retail shopping

96%

Reduction in time to insights,
from 27 hours to 1 hour

93%

Lower operational costs

emart24
PLATFORM USE CASE: Data Intelligence Platform
CLOUD: Azure

“In the retail industry, delivering an instantly gratifying experience is key to winning share of wallet. With Databricks Platform powering our data and AI initiatives, we’ve reduced analysis times by 96% while lowering operational costs by 93%.”

— Jae Kyung Lee, CIO, IT System & Digital, Emart24

As the retail industry evolves, there is a discernible trend toward leveraging new innovations — from personalization to automation of critical services — that create a self-service shopping experience that delights customers. Emart24 is leading the charge in retail-centric innovations, but they struggled with a legacy data platform that was complex to manage and very expensive to scale. After migrating to the Databricks Data Intelligence Platform, they have been able to streamline infrastructure management to more efficiently handle vast data streams from stores, diverse products, and over 1 million daily customer interactions to accelerate the shift toward cashierless shopping experiences that enhance store efficiency and customer ease for convenience stores throughout Korea.

Legacy infrastructure is complex and costly to scale

Emart24 is focusing on evolving into a technology company that is ahead of the curve rather than being a traditional retailer. By designating 2023 as the year of digital innovation, Emart24 strove to introduce and operate various digital technologies to provide new shopping experiences for customers and efficient store operations for managers. As a digital transformation leader in the Korean retail industry, they’ve introduced a number of innovations such as the Amazon-style smart COEX AI store and a smart store system that can be operated as a cashierless store.

With their eyes set on delivering additional use cases to market — including product recommendation services for business owners, automated ordering services for stores and demand forecasting for logistics centers — Emart24 needed to overcome several challenges. Their data was scattered across multiple systems, including sales systems, point-of-sale (POS) systems and customer applications (APP), making efficient management of daily transactions and operational data from more than 6,500 stores and 13 large distribution centers nationwide difficult. Their legacy data platform was not suitable for handling such a large amount of data, as it was overly complex to maintain. “Our prior data management system required a lot of time and high resource costs to analyze large amounts of transaction data, to the tune of more than 6 billion transactions and data for over 80,000 products,” said Jae Kyung Lee, CIO, IT System and Digital, at Emart24. “At this scale, our systems would take 27 hours to run a single analysis and required a resource cost of 930,000 won, making it virtually impossible to provide it to store management on a daily basis.”

They also needed to empower an increasingly diverse set of data users across the organization with varying degrees of technical and programming expertise. They needed not only a modern platform that could scale performance cost-effectively but also a tool that would empower all their data users to be productive with data.

Databricks injects efficiency and the power of AI into retail operations

Migrating to the Databricks Data Intelligence Platform on the Azure cloud, the Emart24 team was able to unlock new opportunities while fostering stronger collaboration across their data users and experts. From a data engineering standpoint, the platform has given them breakthrough capabilities and algorithm optimization resources. “With Databricks, our data team is now able to efficiently perform data compression, skipping, partition pruning and caching,” said Lee. “And we are able to more effectively manage our infrastructure with powerful batch scheduling management, automated cluster management, task monitoring and more.”

In addition, the company has made efforts to build a data culture within the company to support a naturally collaborative process. The company also revealed that it has been sharing data and conducting joint analysis through Delta Sharing with Shinsegae I&C, a partner of Emart24, for data sharing and joint analysis to improve retail customer shopping experiences.

With data pipelines performing at unprecedented speed and scale and Emart24’s data team operationalizing data at newfound levels, they’ve embarked on new AI-powered solutions that provide product recommendations that are optimized for specific stores and locations. They’re also able to deliver on new capabilities for retail stores, including more personalized experiences, cashierless operations and improved demand forecasting.

Building the future of retail on Databricks

With a unified platform on the lakehouse architecture, Emart24 has seen significant performance improvements as well as cost savings. “With Databricks, Emart24 has been able to dramatically reduce the analysis time from 27 hours to one hour, a dramatic 96% reduction in analysis time that has helped us launch new products and features to market such as our cashierless shopping experience,” explained Lee.

In addition to faster time to insight, the company has also seen a reduction in compute costs due to the automated cluster management capabilities and decoupling of compute and cloud storage. “Features such as autoscaling and optimized clusters have reduced our cloud spend from $930,000 to $70,000. This accounts for a 93% reduction in cost.”

In the future, Emart24 will use analytics through the Databricks Data Intelligence Platform more actively to strengthen their position as an industry leader and increase their corporate competitiveness. As AI services using big data play an important role in the company’s overall value chain, the company plans to promote various digital innovations in product planning, logistics management, marketing and store development. “The most important factor for success in future tasks will be Databricks, a data platform that handles big data,” concluded Lee.