Riese & Müller is growing internationally, offering an increasingly diverse product range and expanding digital services around the connected e-bike. This puts new demands on the premium bicycle manufacturer's established IT architecture. Before the collaboration, the system landscape is still tightly coupled: commerce systems for B2B and B2C are directly connected to the ERP, causing deadlocks and export bottlenecks while severely limiting the renewal.

Our goal is to bring new flexibility and creative freedom to Riese & Müller's IT landscape. This establishes the technological foundation for international scaling as well as the expansion of new business models like bicycle leasing or subscriptions. The tasks are clear: modularize the system landscape and establish clear responsibilities, optimize data flows, and standardize product data management. We're doing this through two central project streams:

First, we're developing a data hub as the central data exchange point: it receives ERP data, stores it temporarily, and distributes it in a controlled manner to target systems. Specifically, it processes changes in ERP records, converts them into a domain-oriented, R&M-specific product data model, and distributes product data to surrounding systems in a standardized way. These include the ERP as the source system, the commerce systems for B2C and B2B, as well as the R&M website. The hub is also connected to the in-house RX Cloud along with the RX Connect App for the connected e-bike and digital services. Additionally, there are data standards and industry platforms like DST and VeloConnect, through which manufacturers and suppliers centrally provide standardized product data that retailers can access. The landscape is rounded out by a Product Information Management system (PIM) and Business Intelligence (BI).

"With the data hub, we focused on architecture as a strategic decision, not just integration," explains Alexander Janthur, Founder & CEO of Turbine Kreuzberg. "We're bringing flexibility to the system landscape so that Riese & Müller can quickly launch new business models and services tomorrow on an optimal foundation – without constantly starting from scratch."

The implementation was deliberately risk-minimizing and architecture-driven. First, we modeled domains and events and derived clear data contracts from them. We developed the hub itself as a TypeScript-based, event-driven application with Change Data Capture (CDC). This means we continuously capture actual data changes and forward them to the data hub, rather than transferring the complete data inventory each time.

We developed a standardized product data model that consistently represents variants and attributes and reliably delivers enriched information to all channels. To keep shops and channels running as before, existing IDs and data formats are preserved. When data comes from multiple sources, the hub decides which source is responsible for which field and prevents mutual overwrites. New systems are connected gradually while ongoing operations remain stable. This minimizes modifications, increases data quality (clear model, data governance, workflows), and secures operations. In short: a pragmatic decoupling with immediate effect and room for further development. Strong evidence of the data hub concept's viability: Riese & Müller is now applying the architecture principles to an internally developed application for production.

With the integration of a new PIM system, Riese & Müller has a central, reliable source for product master data: attributes, variants, relations, as well as country, channel, and translation logic are maintained uniformly, versioned, and quality-assured through workflows. The object-oriented data model with inheritance represents common components across the entire product range, while still allowing enough flexibility for bike-specific texts/images and optional dependencies. Through data contracts, we define how the systems share, validate, and synchronize data with each other, so that the prepared product data reaches all channels consistently via the data hub – without modifications to the connected systems.

The selection of Centric PXM (formerly Contentserv) as the PIM technology was the result of an extensive discovery process. This included expert interviews, process, data, and requirements analyses, as well as the development of target data and governance models (roles, workflows, validations). Centric PXM stood out due to its flexibility, functional breadth (data quality, translation, extensibility), and stable cloud infrastructure.

Products are created exclusively in the ERP and receive their core attributes there. These include core product data such as production-relevant information, configuration data, characteristics and variants, blueprints/bills of materials, serial number products, prices, and inventory.

The data hub receives the "skeleton" of a product from the ERP, composed of the mentioned ERP data, and transforms it into the R&M data model. Configurable products and their variants are linked with the corresponding products from the bill of materials. The hub then sends the product data via push to the PIM, which enriches the data. This includes additional descriptive or metric attributes, linking with objects (e.g., manufacturer, color), adding images, documents, translations, and categorizations, as well as assignment to sales channels including publication and deactivation of products. The hub then retrieves the enriched PIM data via pull and also transforms it into the R&M data model.

Finally, the hub sends the refined product to the commerce systems, transforming it into the required target format. This way, the data is available consistently and ready to use across all channels.