Creating and importing events
In the Celonis Platform, events represent actions related to business entities, such as orders being created, invoices paid, or deliveries shipped. Events form the chronological backbone of process mining by capturing what happens and when. Each event belongs to an event type, which defines its identifier, attributes, and links to objects and other events. Celonis creates one event instance for each unique row in the source data.
Custom event types let you model business-specific activities when standard events don’t fully represent your processes, enabling more accurate analysis.
When creating custom event types, you can either define them from scratch or import them from existing tables in your source systems:
Create from scratch: Define event ID, attributes (e.g., timestamp, status, quantity), and relationships manually. See: Creating custom events from scratch.
Create from table : This approach streamlines event modeling by automatically generating the SQL transformations that map event types and relationships to the data. It also enables bulk event creation and configuring event-to-object relationships. See: Creating custom events from a table.
You also have the option to relabel events directly in the Knowledge Model, providing a more efficient way to handle scenarios where multiple change types exist. See: Relabeling events in the Knowledge Model.
To create a custom events from scratch using the Objects and Events dashboard:
Click Events.
Click Create - Model from scratch.
Configure the object type:
Name: Enter a unique name up to 40 characters, using only letters and numbers (no spaces or special characters).
Color: Change the display color used for the event type in the model graph view.
Description: An optional description for this event, visible when viewing the event in the table view only.
Tags: Tags are used to organize, categorize, and filter object types and event types within your object-centric data model. They don’t affect the data itself, they’re purely for management, navigation, and clarity.
Celonis processes: Choose from existing tags used as part of Celonis provided processes.
Custom processes: Enter text to create a new tag or select from existing tags.
Attributes: Configure events attributes to define the properties of an event, choose data types, and enable filtering, grouping, and analysis of your events.
You can choose from the following attribute types: Boolean, datetime, floating point, long integer, and string.
Click Save.
After completing these steps, your custom event is created and ready to use, and you can publish it to the development environment to make it available for transformations and analysis.
You can create custom events from a table in your source systems, saving you from manually recreating them. This approach streamlines event modeling by automatically generating the SQL transformations that map event types and relationships to the data. It also enables bulk event creation and configuring event-to-object relationships.
Using the import wizard, an event will follow the raw data if it fits one of the three supported structures:
A column(s) of the table represent an event(s).
A row in the table represents an event (as an event log).
A row in the table represents an event (change table with old value / new value).
For the video overview:
Before you begin
To import your objects and events, the data source must be accessible via a data connection in your data pool.
To share data between your data pools, see: Sharing data between data pools.
Importing custom event types
To import custom event types from a table:
Click Events.
Click Create - Import from table.
Select the data source you want to use and then select the tables from that data source to import.
Select how the events are stored in the table:
Events for columns:Each timestamped column represents one event type (create multiple).
Events from rows, with a column specifying activities: Each row is an activity. All unique activities are aggregated into one event (create single).
Events from rows, with columns including old and new values: Each row is a change. All changes are aggregated into one event (create single).
Click Next.
The imported events are displayed in edit mode.
Select the table to review the events generated from it.
You can optionally configure the following:
Columns: Check or uncheck columns to include or exclude them.
Add attributes: Click Add attribute to create additional attributes populated from data outside the selected table.
Rename attributes: Rename existing attributes (defaults to the column name).
Tags and descriptions: Apply existing Celonis tags or create custom tags (up to 41 characters) for filtering and reuse.
Click Next to move on to the optional Relationships step.
Relationships describe how objects and events are connected. To learn more, see: Objects, events, and relationships.
Click Next, review the events that you're creating, and then click Create.
The events are now created and can be viewed in your dashboard.
We recommend reviewing transformations now. To do this later or edit event types, click Go to events.
In some cases, a single change event is created even though multiple change types exist. While it is technically possible to generate one event per change type by adding conditions in the SQL transformation and repeating the import process, this approach is manual, difficult to maintain, and does not scale well.
A more scalable solution is to refine the event using its attributes. Instead of creating separate events in SQL, you can define a central change event and differentiate it by incorporating an additional attribute (for example, change type) into the activity name. This can be configured directly in the Knowledge Model / Studio using the Event Log Builder. By combining relevant attributes into the activity name, a single change event can be automatically expanded into multiple distinct events without duplicating the data transformation logic.
To relabel events in the Knowledge Model (KM):
Click Studio and then open the package containing the KM you want to create the event log in.
Select the KM asset and then click Event Logs.
Click Create Event Log, add a display name, and optionally a description.
An ID is automatically generated based on the display name. This is used in PQL statements but can be manually changed if needed.
Select the lead object for the event log. The dropdown lists all the objects in your perspective.
Select the events that are relevant for the process you are tracing. The dropdown lists all the events in your perspective.
To configure dynamic events by adding attributes of event types:
Click the three dots by the name of an event type, and select Relabel event.
Click Add attribute, and select an attribute from the list. The preview updates immediately to include the new attribute in the Activity Details column. You can select any number of attributes for any number of events in the event log.
Tip
Selecting numerical attributes can produce a lot of different groupings - these are handled as separate events and may break certain limits, so be careful when selecting attributes with these data types.
Drag and drop the attributes to arrange them in the order you want. We'll concatenate the attributes in the order you place them.
When you've added all the attributes you want to, select Back.
Choose your event log table settings, including:
Activity column: Select which is the activity column to visualize in the Process Explorer. By default, 'Activity' is selected.
Sorting column (optional): Events with identical timestamps will be sorted according to the values in the selected column.
While specifying a sorting column is optional, we recommend it if your data set includes activities that may have the same timestamp. If two activities belong to the same case and have the same timestamp without any configured sorting, the resulting process graph is based on implementation details (such as the alphanumerical order for CREATE_EVENTLOG). This can result in unwanted process variants.
Use the previews to check that your event log is configured correctly. We'll sample your data to create an event log table, an edge table for connections, and a graph preview. Adjust the selection of events as needed until you get the results you want.
Click Save .
For more information about object-centric event logs, see: Event logs (object-centric).