Data Transformations in Salesforce Data Cloud: From Raw Data to Business-Ready Intelligence

Salesforce Data Cloud(Data 360) enables organizations to ingest massive volumes of customer data from CRM systems, marketing platforms, digital channels, and external data sources. However, raw ingested data is rarely ready for analytics, identity resolution, or activation.

This is where Data Transformations in Salesforce Data Cloud play a foundational role.

Transformations help convert fragmented, inconsistent, and source-specific data into clean, standardized, and business-ready datasets that can power trusted insights and personalized experiences.

This blog explains what transformations are, how they differ from formulas, the types of transformations supported (Batch and Streaming), how to set them up, and real-world use cases.

Transformation in Data Cloud

What Are Transformations in Salesforce Data Cloud?

Transformations in Salesforce Data Cloud are declarative mechanisms used to clean, enrich, reshape, and prepare data after ingestion.

When data enters Data Cloud, it is stored in Data Lake Objects (DLOs) or mapped to Data Model Objects (DMOs). At this stage, the data often mirrors the source system structure and values.

Transformations allow you to:

• Standardize inconsistent values

• Derive new attributes

• Enrich data using reference datasets

• Aggregate records for analytics

• Prepare data for identity resolution, insights, segmentation, and activation

A critical architectural principle is that transformations do not modify the original ingested data. Instead, they generate transformed outputs that can be safely reused across downstream processes.

Why Transformations Are Essential

In real-world enterprise environments, data challenges are unavoidable:

• Country values arrive as India, IND, IN

• Phone numbers follow different formats

• Transaction data is too granular for analytics

• Business logic exists outside source systems

Without transformations:

• Identity resolution accuracy decreases

• Analytics become inconsistent

• Segmentation quality suffers

Transformations ensure Data Cloud acts not just as a data store, but as a trusted intelligence layer.

Transformations vs Formula Fields: A Clear Distinction

Salesforce users often compare transformations with formula fields. While both involve calculations, their intent and execution model differ significantly.

Formula Fields

• Evaluated at runtime

• Limited to record-level context

• Suitable for simple calculations during ingestion or display

Transformations

• Executed as part of data processing pipelines

• Designed for high-volume datasets

• Reusable across identity, insights, and activation

Salesforce documentation positions transformations as data engineering tools, while formulas serve lightweight derivation needs.

Types of Transformations in Salesforce Data Cloud

Salesforce Data Cloud supports two primary transformation types, each serving different latency and complexity requirements:

1. Batch Data Transforms

2. Streaming Data Transforms

1. Batch Data Transforms

What Are Batch Data Transforms?

Batch Data Transforms run on demand or on a defined schedule. They are designed for complex data preparation scenarios involving joins, aggregations, filtering, and enrichment across large datasets.

They are created using a visual transformation editor, making them accessible to admins and architects without requiring custom code.

Capabilities of Batch Data Transforms

Batch transforms support:

• Joining multiple Data Lake Objects

• Aggregating metrics (sum, count, average)

• Applying conditional logic

• Creating derived fields

• Filtering records

• Producing reusable transformed datasets

These capabilities make batch transforms ideal for periodic, heavy-processing workloads.

Day-to-Day Example

A retail organization ingests:

• Customer profiles from CRM

• Transaction data from an order system

A batch transform:

• Joins customers with transactions

• Aggregates total spend and purchase count

• Creates a “Customer Value Tier” field

This transformed dataset is then used for:

• Identity resolution

• Calculated insights

• Marketing segmentation

When to Use Batch Data Transforms

Use batch transforms when:

• Data does not require real-time availability

• Complex joins and aggregations are needed

• Processing large historical datasets

• Preparing data for analytics and reporting

2. Streaming Data Transforms

What Are Streaming Data Transforms?

Streaming Data Transforms apply transformation logic as data is ingested, enabling near real-time data preparation.

They are defined using a SQL SELECT statement and are typically applied to streaming ingestion sources.

Capabilities of Streaming Data Transforms

Streaming transforms:

• Process records continuously

• Apply cleansing and normalization logic instantly

• Read from a source DLO and write to a target DLO (or mapped DMO)

• Are optimized for low-latency use cases

They are best suited for simple but time-sensitive transformations.

Day-to-Day Example

A digital application streams user events into Data Cloud. Each event record contains multiple phone number fields.

A streaming transform:

• Normalizes phone numbers

• Converts them into a standardized structure

• Writes clean records immediately

This ensures downstream systems always operate on standardized data without waiting for a batch run.

When to Use Streaming Data Transforms

Use streaming transforms when:

• Data freshness is critical

• Immediate normalization is required

• Transformations are relatively lightweight

• Supporting real-time personalization or alerts

  
  

Batch vs Streaming: How to Choose

Aspect	Batch Data Transforms	Streaming Data Transforms
Execution	Scheduled or manual	Continuous
Latency	Minutes to hours	Near real-time
Complexity	High (joins, aggregates)	Moderate (SQL-based)
Best For	Analytics, insights, segmentation	Real-time readiness

How to Set Up Transformations in Data Cloud

Step 1: Ingest Data

Data is ingested into Data Lake Objects via connectors or APIs.

Step 2: Choose Transformation Type

• Select Batch Transform for scheduled processing

• Select Streaming Transform for real-time processing

Step 3: Define Transformation Logic

• Batch: Use the visual transformation editor

• Streaming: Define SQL-based transformation logic

Step 4: Run or Schedule

• Batch transforms can be scheduled or run on demand

• Streaming transforms run continuously

Step 5: Use Transformed Data

Transformed outputs can be used for:

• Identity Resolution

• Calculated Insights

• Segmentation

• Activation

Real-World Use Cases

Identity Resolution Improvement

Standardized emails and phone numbers improve match accuracy.

Analytics and Reporting

Aggregated customer metrics enable reliable insights.

Marketing Personalization

Clean, enriched attributes enable accurate segmentation and activation.

Final Thoughts

Transformations are a core architectural pillar of Salesforce Data Cloud. They bridge the gap between raw ingestion and meaningful business outcomes.

By choosing the right transformation type—batch or streaming—and applying consistent logic, organizations ensure that data remains trusted, reusable, and activation-ready across the entire customer lifecycle.

Official References

• Connect & Transform Datahttps://trailhead.salesforce.com/content/learn/modules/data-cloud-connect-and-unify/connect-and-transform-data

• Batch Data Transformshttps://trailhead.salesforce.com/content/learn/modules/batch-data-transforms-in-data-cloud-quick-look/get-started-with-batch-data-transforms-in-data-cloud

• Streaming Data Transformshttps://trailhead.salesforce.com/content/learn/modules/streaming-data-transforms-quick-look/get-started-with-streaming-data-transforms-in-data-cloud