Engineering KPI Trust​ - Enercity AG

All KPI rules are implemented exclusively in the Business Vault layer, which is the earliest permis- sible stage in the data warehouse to introduce business logic. This enforces a strict, one-way data flow where finalized results are pushed out to information marts, ensuring that core calculations can be reliably reused across countless data products, or shared with external subsidiary teams via dbt Mesh. Within the Business Vault, we maintain a separation between base business rules and reporting period rules . Base business rules are centralized into “Base Objects,” which house shared, compa- ny-wide technical definitions for key concepts and metrics. For a KPI “active electricity supply con- tracts”, these logics determine the baseline qualifications: it must be an electricity product, and the status code must indicate it is active or pending termination. Base objects bring together data from disparate source tables and systems, then apply business rules in this single location. With this, a single source of truth for each KPI is established. On this foundation, we utilize “Timeslice Objects” to apply reporting time boundaries to the es- tablished baseline. Rather than altering the core logic of the KPI, timeslice objects simply evaluate whether a specific record should be counted within a given observation window, generating specific monthly, quarterly, or year-to-date views. To calculate the month-to-date number of active electrici- ty supply contracts, only contracts with creation date within the range of the first day of snapshot’s month to the snapshot date are counted. This allows a single KPI definition to seamlessly serve mul- tiple use-cases with completely different reporting cycles. To ensure these time-based views remain consistent, we rely on the core Data Vault 2 concept of the Snapshot Date , which establishes a unified timeline across the entire data warehouse. Agreed upon by the IT data teams, a daily synchronization occurs at 6:00 AM, fundamentally answering the question of what exactly was known in the data warehouse at that specific point in time. Be- cause every single object uses this exact same timeline, information from completely disconnected source systems can still be perfectly aligned. This baseline definition is maintained centrally in the Cross-project and distributed to the decentralized data projects via dbt Mesh. Ultimately, this snap- shot architecture enables the Controlling department to enforce strict reporting cut-off dates. For example, on the April 15th cut-off for Q1, the system simply locks in the 6:00 AM snapshot from that exact day to generate the previous quarter’s KPIs. This guarantees that reported figures remain con- sistent and reproducible even when source systems change over time. For the Controlling department, this implementation design directly resolves long-standing issues with high data volatility and inconsistent reporting. Historically, many in-house systems only stored current data views, and uncoordinated user modifications often led to retroactive changes that eroded management’s trust in the delivered metrics. By anchoring the architecture on the shift-left paradigm and immutable Data Vault snapshots, Controlling now establishes the foundation for “one definition, one calculation, one trusted KPI”. Once a KPI is generated at the specific reporting cut-off date, it becomes a standardized, locked value that all business units can confidently rely on, elimi- nating the confusion of moving targets and multiple versions of the truth.

Enercity AG Success Story - Engineering KPI Trust - Share Reliable Data in a Regulated Industry​

11/12

Powered by