Data extractions and transformations for object-centric process mining
Celonis uses extraction and transformation tasks to convert your business data into an object-centric data model, giving you a flexible, system-agnostic view of processes across your organization. You can use predefined extractions and transformations or create custom versions suited to your source systems. The data you extract and transform is then stored in an object-centric data pool.
The flow of data here is displayed in this diagram:
Connect: The first stage is for your source system to be connected to the Celonis Platform. The method of connection depends on the source system you're connecting to, with further details explained here: Connecting data sources.
Extract: Once you've connected your source system to the Celonis Platform, you can then use data jobs (known as extraction tasks) to extract that data. Extraction tasks pull relevant data from your source systems (like SAP ECC, Oracle EBS, or other databases) so that it can be transformed into the object-centric model. See: Extractions
Transform: With the data from your source systems available to the Celonis Platform, you can then use transformation tasks to convert that data into objects, events, changes, and relationships. See: Transformations.
Store: The objects, events, changes, and relationships that are created from your data are stored in a database in an OCPM data pool. See: Object-centric data pool.
Extractions pull relevant business data from your source systems (like SAP ECC, Oracle EBS, or other databases) and prepare it for transformation into the object-centric model. Think of extractions as data collection pipelines: they gather the right data without altering it.
Extraction tasks work by:
Identify relevant tables and columns: They locate the source tables that contain the information you need (e.g., invoices, purchase orders, shipments) and then select the attributes to extract (e.g., document ID, vendor, date).
Extracting raw data: The raw data is then pulled into the Celonis Platform so that transformations can process it. This data is kept unaltered, so it's not turned into objects or events at this stage.
Using predefined extractions or creating custom extractions
Predefined extractions: Use predefined extractions when you work with standard processes like Accounts Payable, Order Management, or Procurement. They automatically know which tables and fields to pull from SAP ECC or Oracle EBS.
Creating custom extractions: Create custom extractions when your source system isn’t covered, or you have custom tables, attributes, or business processes. Define the tables and columns to extract using SQL, and combine them with custom transformations to populate new objects or events.
Transformation tasks take the raw data from extractions and turn it into objects, events, changes, and relationships in your object-centric model. They shape your data so you can analyze processes effectively.
Transformation tasks work by:
Creating objects: Transformation task map raw source data to object types (e.g., invoices, purchase orders, shipments) and then identify changes to each object and store them in change tables.
Creating events: The tasks use objects and their changes to generate events that reflect real business actions. They then assign attributes to each event, such as timestamps, IDs, or amounts.
An example of transformations within Objects and Events:
Customizing or extending transformations
While we recommend starting with predefined transformations when possible, as they cover common processes and speed up implementation, you also have the option to customize or extend transformation tasks. This allows you to combine custom SQL with predefined transformations to fit your business requirements.
Customize transformations when you need to adjust how Celonis populates objects or events (e.g., missing data, table joins, or ID mappings).
Extend transformations to populate custom attributes or create new object/event types without replacing the original logic.
The object-centric data pool is where Celonis stores all your object-centric data: objects, events, changes, relationships, and transformations. Each object and event type gets its own table, and changes or relationships are tracked in supporting tables. By default, a single shared model lives in each data pool, giving you a central workspace and a single source of truth.
The data pool supports development (test) and production (read-only) environments, so you can safely build and validate transformations before using them in production. You control access with data pool permissions, and URLs or API calls always point to the correct model.
For most users, start with a single data pool and model. Enable multiple models only if you need strict data separation for compliance or organizational reasons. Remember: object-centric and case-centric assets can coexist in the same data pool, but each model must be entirely one type or the other.