Data Cloud Data Model Objects

Data Cloud Data Model Objects

This skill activates when a practitioner needs to design or manage Data Model Objects (DMOs) in Salesforce Data Cloud. DMOs are the semantic layer of the Customer 360 Data Model — they are logical groupings of data stored in Data Lake Objects (DLOs) and organized into subject areas. This skill covers DMO schema design, mandatory DMOs for identity resolution, the XMD (extended metadata) layer for field formatting, and the data transform types that can create or populate DMOs.

Before Starting

Gather this context before working on anything in this domain:

• DMOs do NOT support SOQL. Queries against DMO data require the Data Cloud Query API (ANSI SQL), not standard Salesforce SOQL. Attempting to query a DMO via SOQL will fail.
• Five DMOs are mandatory for identity resolution: Individual, Party Identification, Contact Point Email, Contact Point Phone, and Contact Point Address. Data streams that do not map to these DMOs cannot participate in identity resolution.
• Only one data relationship is permitted per field pair between two mapped DMOs. Multiple relationships between the same two fields are not supported.
• XMD (extended metadata) has three tiers: System XMD (platform-generated, immutable), Main XMD (org-customizable via REST API), and User XMD (per-user preferences). You can only modify Main XMD, not System XMD.
• Streaming transforms are restricted to a single DLO (no cross-DLO joins); batch transforms support joins and aggregations across multiple DLOs.

Core Concepts

DMO Architecture: DLO → DMO Layer

Data Cloud stores raw ingested data in Data Lake Objects (DLOs). DMOs are a semantic abstraction layer on top of DLOs:

• DLO — Raw storage: one DLO per data stream source. Fields are ingested as-is.
• DMO — Semantic model: curated schema with business-friendly names, relationships, and subject area classification. Multiple DLOs can map to a single DMO.
• Subject Areas — Organizational groups within the Customer 360 Data Model (e.g., Sales Order, Unified Individual, Marketing). DMOs belong to a subject area.

Field mapping from DLO to DMO is configured in the Data Stream setup. This mapping determines which DLO fields populate which DMO fields.

Five Mandatory DMOs for Identity Resolution

Identity resolution in Data Cloud requires data to be mapped to specific standard DMOs:

Data streams that omit field mappings to these DMOs cannot contribute to unified identity profiles. The identity resolution ruleset operates exclusively on these five DMOs.

XMD: Extended Metadata Layer

XMD controls how Data Cloud fields are displayed: labels, formatting, data type hints, and measure/dimension classification. It has three immutable tiers:

• System XMD — Automatically generated by the platform when a dataset or DMO is created. Cannot be modified.
• Main XMD — Org-level customization. Can be updated via PATCH /services/data/v{version}/wave/datasets/{datasetId}/xmds/main. This is the correct layer for adding field labels, aliases, and formatting properties.
• User XMD — Per-user preferences applied on top of Main XMD. Cannot be updated via API; managed by end users in the interface.

Attempting to update System XMD fields via the REST API returns an error. Identify the XMD type before attempting to modify it.

Streaming vs. Batch Transforms

Two transform types create or populate DMO records from DLO data:

Streaming Transforms:

• Execute near-real-time on new DLO records as they arrive
• Support only a single source DLO per transform (no joins)
• Cannot perform aggregations
• Use for event enrichment that does not require cross-source joins

Batch Transforms:

• Execute on a scheduled interval
• Support joins across multiple DLOs and aggregations
• Use for analytical metrics, cross-source unified views, or enrichment requiring multiple data sources

Common Patterns

Pattern 1: Custom DMO Schema Design

When to use: Onboarding a new data source that doesn't map to any existing standard DMO — e.g., a custom loyalty transaction object.

How it works:

1. Define the DMO name, API name, and subject area assignment
2. Identify the source DLO(s) and define field mappings (source field → DMO field)
3. Determine if the DMO needs a relationship to another DMO (e.g., linking loyalty transactions to the Individual DMO)
4. Create only one data relationship per field pair — multiple relationships between the same two DMO fields are not allowed
5. Map identity-relevant fields (email, phone, address) to the five mandatory DMOs in a separate data stream if they exist in the source

Why not use SOQL to query the DMO: DMOs are columnar stores, not Salesforce objects. SOQL does not work against DMO data. Use the Data Cloud Query API with ANSI SQL to query DMO records.

Pattern 2: Main XMD Update for Field Label Customization

When to use: Field labels in a Data Cloud dataset or DMO need to be renamed for business-friendly display without modifying the underlying field API name.

How it works:

1. Retrieve the current Main XMD: GET /services/data/v{version}/wave/datasets/{datasetId}/xmds/main
2. Modify the fields array in the XMD JSON: update label, alias, or formatting properties for the target field
3. Submit the updated XMD: PATCH /services/data/v{version}/wave/datasets/{datasetId}/xmds/main
4. Validate the change by querying the dataset in CRM Analytics or Data Cloud interface

Why not edit System XMD: System XMD is platform-generated and immutable. PATCH requests targeting System XMD return a 403 error. Always work with Main XMD.

Decision Guidance

Recommended Workflow

1. Inventory source data schema — Document all entities, fields, data types, and relationships in the source system.
2. Map to Customer 360 Data Model subject areas — Identify which standard DMOs the source data fits. Create custom DMOs only for entities with no standard DMO equivalent.
3. Design mandatory DMO mappings for identity resolution — Confirm that identity-relevant fields (email, phone, address, external IDs) are mapped to the five mandatory DMOs.
4. Define data relationships — Document which DMOs need relationships. Enforce the one-relationship-per-field-pair limit during design.
5. Select transform type — For each DMO population path, choose streaming (single source, near-real-time) or batch (multi-source join, scheduled).
6. Design XMD customization — Identify which fields need label changes or formatting updates. Plan Main XMD PATCH calls for each.
7. Validate with Data Cloud Query API — After DMO population, run ANSI SQL queries via the Data Cloud Query API to confirm field values and relationship traversal.

Review Checklist

• Five mandatory DMOs mapped for identity resolution (Individual, Party ID, Email, Phone, Address)
• Data relationships limited to one per field pair between DMO pairs
• XMD updates target Main XMD, not System XMD
• Streaming transforms restricted to single-DLO sources; joins use batch transforms
• DMO queries use Data Cloud Query API (ANSI SQL), not SOQL
• Subject area assignments documented for each custom DMO

Salesforce-Specific Gotchas

1. DMOs do not support SOQL — DMOs are columnar stores in the Data Cloud data lake, not Salesforce objects in the CRM object model. SOQL queries against DMO API names return errors or empty results. All DMO data access requires the Data Cloud Query API with ANSI SQL syntax.
2. System XMD is immutable — PATCH against it returns 403 — System XMD is auto-generated by the platform and cannot be modified. Main XMD is the correct target for field customization. Always retrieve the XMD type before attempting to patch it.
3. Only one data relationship per field pair between DMOs — You cannot create two separate data relationships between the same pair of fields on two DMOs. If a more complex mapping is needed, it must be modeled as a different field pairing or through a transform.

Output Artifacts

Related Skills

• data/data-cloud-data-streams — For configuring DLO ingestion and field mapping from source to DLO
• admin/data-cloud-identity-resolution — For identity resolution ruleset design and matching rule configuration
• admin/data-cloud-segmentation — For building segments that query unified DMO profiles
• architect/data-cloud-architecture — For overall Data Cloud platform architecture and deployment planning