Häufig gestellte Fragen
What is process mining in ERP, explained simply?
Process mining is a data-driven analysis method that reconstructs and visualises how business processes actually run in practice, based on the event logs of an ERP system. Instead of relying on modelled target assumptions or interviews, it evaluates the digital traces that every transaction leaves behind, for example in procurement, order processing or invoice verification. Deviations from the intended flow, avoidable loops and bottlenecks thus become visible on a factual basis. As a scientific discipline, the term goes back to the computer scientist Wil van der Aalst, who first used it in 1998.
How does process mining differ from business intelligence?
Business intelligence focuses on key figures and aggregates and primarily answers the question of "how much", whereas process mining reconstructs the chronological flow of individual process instances and thus answers the question of "how". Process mining therefore shows which paths, sequences and loops actually exist, instead of merely outputting totals and averages on a dashboard. The two approaches are complementary: BI condenses results, while process mining explains how they came about. In practice, the exact demarcation varies depending on the industry, company size class and customising depth of the specific ERP setup.
What data and prerequisites must the ERP system meet for process mining?
The essentials are traceable document chains with timestamps for each status change, plus accessible data extracts from the transaction tables and the audit trail. At its core, every analysis requires an event log with three pieces of information: a case identifier (such as a purchase order number), an activity and a timestamp. Modern cloud ERPs often deliver such event logs directly via APIs, while classic on-premises systems require database access and a defined extraction model. Consistent master data maintenance and a clear case concept are also crucial, as incomplete or ambiguous data distorts the results.
Is there a standard for event logs in process mining?
Yes, the established exchange standard is XES (eXtensible Event Stream), which was developed by the IEEE Task Force on Process Mining and adopted as IEEE Standard 1849 in 2016; a revised version, IEEE 1849-2023, has been available since 2023. XES is an XML-based format that describes an event log as a set of cases (traces), which in turn consist of events with attributes such as activity name, timestamp and case identifier. Extensions can be used to define shared attributes with clear semantics, for instance for naming elements or for the lifecycle phase of an activity. In practice, many ERP connectors and specialised tools export the data either as XES or as a tabular CSV format.
Do I need a dedicated tool such as Celonis or Signavio for process mining?
For lightweight analyses, SQL and Python libraries such as PM4Py are often sufficient; however, as soon as conformance checking, large data volumes or regular evaluations with stakeholders are required, dedicated platforms with ready-made ERP connectors pay off. In the market, Celonis is regarded as the leading vendor and was again rated a "Leader" by Gartner in 2026, albeit in the renamed Magic Quadrant for process intelligence platforms, while SAP Signavio — acquired by SAP in 2021 for around 1.2 billion US dollars — scores particularly well in SAP S/4HANA landscapes with pre-integrated connectors. Other relevant vendors include UiPath, Microsoft, IBM, ARIS and Apromore. When making a selection, standard interfaces such as REST API and OData should be supported directly in order to avoid expensive custom integrations.
What data protection and co-determination issues need to be considered with process mining?
Event logs frequently contain personal user identifiers, so their processing is subject to employee data protection under the GDPR and the German Federal Data Protection Act (BDSG); pseudonymising user IDs together with strict purpose limitation and data minimisation is therefore common practice. Since process mining systems are in principle capable of monitoring the behaviour and performance of employees, the works council in Germany regularly has a co-determination right under Section 87 (1) No. 6 of the Works Constitution Act (BetrVG). Involving employee representatives early and concluding a works agreement are therefore established practice to keep the deployment legally sound. In practice, the specific arrangement depends on the industry, the co-determination culture and the sensitivity of the processes being analysed.