Understanding Related Attributes vs. Direct Attributes in Salesforce Data Cloud: Key Differences and Implications
In the realm of Segmentation and Activation, distinguishing between related attributes and direct attributes is crucial for effective personalization.
By grasping the nuances between these two types of attributes, Marketers can better analyze data, improve customer personalization, and drive strategic initiative.
Let’s connect and discuss how these concepts can impact your business! #DataCloud #CustomerPersonalization #SegmentationStrategy
Further Information on Related Attributes versus Direct Attributes:
1. Related attributes are obtained from associated objects.
2. Segments can solely be established based on Profile Category, excluding engagement Category.
3. It is not possible to create a segment on the contact point object, as it falls under a Other Category, thus categorizing it as related attributes.
4. Data Cloud categorizes attributes in segmentation into two types:
- Direct attribute (1:1)—This attribute possesses a single value for the user in the DMO, such as postal code or first name.
- Related attribute (1:N)—This attribute can encompass multiple values for a specific user, such as purchases or email events.
Attribute Details:
For segmentation, Data Cloud offers two types of attributes: Direct (1:1) and Related (1:N). This evaluation originates from the object that was selected for Segment On.
- Direct attributes (1:1): Attributes that have only one value for a record in the Segment On object. For example, if Individual DMO is selected as Segment On, then Individual First Name or Individual Gender attributes are listed as direct attributes. Additionally, all attributes from other DMOs that are related to Individuals with a relationship cardinality of 1:1 are direct attributes.
- Related attributes (1:N): Attributes that have one or more values for a record in the Segment On object. For example, if Individual DMO is selected as Segment On, then Contact Point Email is active or Sales Order Grand Total Amount attributes are listed as related attributes.
In segmentation, the entire DMO hierarchy is displayed. In activation, the DMOs that are one hop away from the object are displayed.
Utilize segmentation filters to identify users who meet the criteria for the segment. Employ activation filters to determine attributes for personalizing messages.
[SEGMENT Filter = Event Attendees, (who have not made a donation in the past two months)]
[ACTIVATION Filter = Type of Event, (Webinar)]
A Direct Attribute is one that is explicitly stored about an entity, whereas a Related Attribute is associated with it through another connected entity.
Direct Attribute:
A customer's "email address" serves as a direct attribute since it is explicitly recorded in the customer profile.
Related Attribute:
A customer's "shipping state" is classified as a related attribute because it requires access to their "shipping address" to ascertain the state in which they reside. To retrieve a related attribute, it is often necessary to navigate through a relationship to another data source or table to obtain the pertinent information.
Direct attributes are those that maintain a one-to-one relationship with the target segment, indicating that each segmented entity possesses a singular data point for a profile attribute. For instance, customer data would include only one entry for postal code or first name.
Conversely, related attributes can encompass multiple data points. In the context of customer data, related attributes may include instances where an individual has several data points, such as purchases or email interactions.
The "Segment On" Unified Individual DMO addresses the process of Identity Resolution, which is defined and tailored within the Data Cloud. This process consolidates "similar" Individual Records into a single entry, i.e Unified Individual thereby eliminating the occurrence of duplicate individuals. For example:
Choose the Profile option if your dataset includes identifiers such as consumer IDs, email addresses, phone numbers, account IDs, or any other demographic you wish to segment or utilize as the initial population for segmentation. Opt for 'Created Date' within the Engagement category, as 'Updated Date' is subject to frequent changes. This mutable date field can lead to record duplication, where multiple entries with the same primary key are added to the data lake object (DLO).
The Segment On feature identifies the target object upon which the segment is constructed. The process of segmentation relies on the data model in use and the object designated as "Profile" during the creation of data streams and data modeling.
Individuals refer to a particular person or customer sourced from a specific data repository, such as Marketing Cloud Engagement.
A unified individual is defined as a customer profile that has been consolidated from various sources through the application of Identity Resolution rules.
Publish Types and Scheduling:
There are two categories of publishing: Standard and Rapid. The Standard publishing option operates on a schedule of either 12 or 24 hours, whereas the Rapid publishing option is available on a schedule of 1 or 4 hours. Once a segment is designated for Standard publishing, it cannot be converted to Rapid publishing.
When activating a segment that includes related attributes, certain restrictions apply. Specifically, a maximum of 20 related attributes is permitted per activation, and there is a cap of 100 activations that can include related attributes. Additionally, all related attributes must reside on the same data model object path.
Best Practices for Optimizing Segment Performance
Select the appropriate Segment on the target object according to your business objectives: Through further refinement, the unique members of this target object constitute your target audience. If your data is sourced from various origins, utilize the Unified Individual DMO, which enhances performance and yields more precise results.
Utilize the correct DMO type: Choose between Profile, Engagement, or Other based on the characteristics of your target dataset.
Opt for the most direct paths and avoid cyclic paths: Whenever feasible, select the most direct route between two DMOs. Longer paths result in extended join paths, increasing the workload for the segmentation engine. In a cyclic path, you begin at a DMO and, through the join relationships, return to the same DMO, such as a→b→c→a or a→b→c→b. Cyclic paths can lead to longer processing times and may cause query failures.
Reduce the volume of data to be processed: By minimizing the data that the segmentation engine must handle, the overall performance of the segment is enhanced. Employ data spaces and establish explicit filters within the segment.
Consolidate containers when feasible: When two containers with identical container paths are connected using “Or” logic, consider merging them. Additionally, merge your containers when dealing with related attributes.
Implement nested operators: Tackle complex segmentation needs within a single container to enhance segment performance.
Utilize nested segments: By nesting a segment in membership mode, the segmentation engine avoids re-evaluating the filter criteria of the nested segment, thereby improving performance.
Leverage calculated insights or data transformations for intricate operations: Calculated insights and data transformations are effective tools that can alleviate the computational demands of segmentation.
Steer clear of using skewed engagement data: Skewed data may result in one partition containing an excessive amount of information, which can prolong operation times.
Comments
Post a Comment