65% Faster Data Processing, Zero Manual Effort
How Media-Saturn eliminated manual processes and built a future-proof system architecture
At a Glance
Manual processes, technical debt, and missing documentation in an organically grown shop system.
New microservices architecture, full process automation, and system documentation following the 4+1 view model.
65% faster processing time, zero manual interventions in data processing, first-ever complete architecture documentation.
Initial Situation
Media-Saturn-Holding GmbH operates one of the largest shop systems in European electronics retail, with over 1,000 stores and approximately 40,000 employees. The system had grown organically over the years: new features were layered on top of existing structures without adapting the overall architecture.
Customer expectations for real-time availability and fast delivery were rising across e-commerce. The technical foundation became a strategic bottleneck. Media-Saturn sought an independent partner who could objectively assess the situation and develop a viable solution, without ties to existing technology vendors.
Challenge
The core problem was not the technology itself, but the lack of transparency: no one in the organization had a complete picture of how data flowed through the system. This led to four concrete pain points:
- Staff intervened daily in processes that should have been automated. The reason: system dependencies were simply undocumented.
- Technical debt blocked progress. Years of accumulated workarounds made every change risky and time-consuming.
- The legacy system had never been systematically documented. Knowledge existed only in the heads of individual employees.
- IT and business spoke different languages. A shared model of the system landscape didn't exist.
Previous attempts to solve these problems internally had failed due to the complexity of the grown structures. Day-to-day operations simply left no capacity for it.
Methodology
Ekstend assigned an experienced Senior Architect to the project, who worked closely with Media-Saturn's internal development team and business departments over three months. A Process Analyst provided additional support, bridging the gap between technical requirements and business processes.
Rather than following a rigid framework, the approach was specifically tailored to the situation at hand: after all, full transparency over the existing system landscape had to be established before sound architecture decisions could be made. Weekly alignment sessions with IT leadership and business stakeholders ensured that every technical decision remained anchored to concrete business requirements.
Assessment & Analysis
Four weeks of systematic discovery: interviews with 12 stakeholders from IT and business, mapping of all data flows, identification of 23 technical debt items. Result: the first-ever complete picture of the actual system landscape.
Architecture Design
Evaluation of three architecture approaches (monolith refactoring, strangler fig pattern, greenfield microservices) against criteria including migration risk, time-to-value, and team competency. Decision: a hybrid approach - microservices for new functionality, gradual decomposition of the monolith.
Implementation & Automation
Delivery in two-week sprints with joint reviews. Each automated process was first run in parallel with the manual process before the manual intervention was switched off - minimizing risk without compromising the timeline.
Documentation & Knowledge Transfer
Documentation of the entire legacy system following the 4+1 view model (Logical View, Development View, Process View, Physical View + Scenarios). Three knowledge transfer workshops with the internal team so Media-Saturn can evolve the architecture independently.
Solution
The new architecture is based on microservices, with every technology decision directly tied to a business outcome:
- Java with Spring Boot as the backend framework, because the internal team already had Java expertise. This made knowledge transfer possible without retraining.
- For data storage, PostgreSQL replaces the previous proprietary solution and saves tens of thousands of euros in license costs annually.
- React for internal administration interfaces. Business departments can now make configuration changes themselves instead of waiting for IT.
- The API layer runs on GraphQL and reduces database queries per page load by a factor of 4, loading only the data actually needed.
- Apache Kafka decouples services from each other. If one subsystem fails, it no longer blocks the entire data processing pipeline.
Results
Within three months, the operational reality at Media-Saturn changed fundamentally. Standard data processing tasks that previously consumed a significant part of the workday now run fully automated. No more manual intervention. The daily ad-hoc fixes that had tied up the team have been eliminated entirely.
Equally important: the entire legacy system was systematically documented for the first time using the 4+1 view model. Knowledge that previously existed only in individual developers' heads is now accessible to the whole team. Architectural decisions can now be data-driven for the first time, rather than based on gut feeling.
The new microservices architecture allows additional services to be developed and deployed independently, without jeopardizing the overall system. Media-Saturn has solved an acute problem and built a foundation for the system's long-term growth.
Faster processing time
Manual interventions per day
Architecture views documented
Weeks to go-live
For the first time, we have a complete picture of our system landscape - and for the first time, IT and business speak the same language when it comes to architecture decisions.
Lessons Learned
The greatest efficiency gains came from transparency, not technology. Only when IT and business had the same picture of the system landscape in front of them for the first time could they jointly prioritize which processes to automate first.
It was also surprising how many manual interventions weren't technically necessary at all. Nobody had simply ever made the dependencies between systems visible. Once the data flow was documented, some problems practically resolved themselves.
For comparable projects, our recommendation: don't shorten the assessment phase. It feels slow, sure, but it's the foundation for every subsequent decision. Cut corners here, and you'll pay double later.
Technologies & Services Used
Your Contact
You might also be interested in
Facing a similar challenge?
Let's talk about how we can move your project forward too.