METRO.digital GmbH aims to revolutionize the wholesale sector. As METRO's technology unit, METRO.digital is responsible for the daily design, implementation, testing and rollout of hundreds of services and digital solutions at six locations that drive business in 24 countries.
The project at a glance
- Recommendations for 80% of products, with a 4% click-through rate
- Six-figure sales every month and rising thanks to recommendations
- Automated customer churn forecasts for marketing and CRM
- Reduced costs by replacing proprietary silo solutions
Initial situation
With sales of over €27 billion, METRO AG has been an international wholesale heavyweight for over 50 years. It uses advanced data analytics to better understand and provide the best service to its customers faster. A comprehensive portfolio of CRM analysis software to identify customer migration and generate product recommendations for different sales channels was already in place at the start of the project. However, the software installed on-premises required a great deal of manual effort. The aim of moving to the Google Cloud was to achieve scalability at the technical, business, and analytical levels while simultaneously reducing manual operational effort. New content requirements also necessitated the expansion of the existing portfolio.
Solution
A joint team from METRO.digital and codecentric was put together for the duration of the project. The team, consisting of experienced and young data scientists and data engineers, worked with agile strategies such as scrum and kanban and was able to improve the algorithms iteratively.
Only a small part of the existing product portfolio could be effectively migrated via "lift and shift". The majority of the existing applications, along with the new ones, were reimplemented as cloud native to take full advantage of the Google Cloud's capabilities. The cloud native solutions save resources and scale better.
A range of solutions for automating complex data and machine learning pipelines were evaluated and implemented in order to minimize administrative and operational tasks. The team can keep an eye on everything with monitoring and reporting through dashboards.
The main programming language was Python, which supports not only the excellent integration into the Google Cloud but also all relevant machine learning and data engineering frameworks. The technology stack also included Kubernetes, BigQuery, Apache Airflow, Google Cloud Storage, and the Google Cloud AI platform.
Result
The optimized and enhanced data science portfolio of METRO.digital now has a sustainable future in the cloud. Its advanced services enable METRO to better understand its customers’ needs, anticipate likely customer migrations, expand its existing customer business, and improve the overall customer experience. At the same time, IT infrastructure costs have been reduced and data analytics throughput increased as a result of automation.
Dr. Anna Hannemann
Product Owner "Data Science Recommendations", METRO.digital
We worked with codecentric to make the transition from manual, on-premises data science to automated data science processes in the cloud. The support of codecentric's team gave us the opportunity to scale and to prepare for a sustainable future, both technically and strategically. We have always received the best advice and support.
Dr. Anna Hannemann
Product Owner "Data Science Recommendations", METRO.digital
Any questions about the project?
Are you interested in a customized cloud-based solution for your company? Let's talk without obligation.
Head of Data
Further projects of codecentric AG
Find out about other successful projects that we have completed with our customers. Perhaps you will find inspiration for a use case in your company here.
Matthias Niehoff
Head of Data