How does Data Cloud Work (Part 2) - Harmonization & Transformation

Welcome back to Part 2 of my “How Does Data Cloud Work?” series.

In Part 1, we explored the two primary approaches to implementing Data Cloud—whether within an existing org or through a new, standalone environment. Now, let’s go deeper into what makes Data Cloud truly intelligent: its ability to harmonize and transform data at scale.

These next steps—Connectivity, Transformation, Harmonization, and AI Mapping—are the foundation of how Data Cloud translates massive, scattered data into actionable insights. Together, they enable seamless unification of enterprise information, which in turn powers AI engines, digital labor models, and Salesforce’s Agentforce capabilities.

Let’s unpack how this journey works, step by step.

Understanding Connectivity, Transformation & Harmonization

In the previous segment, we learned about the flow of information in the Data Cloud. Now, it is time to move to the next part of the discussion—connectivity. You have learned how the data cloud functions and structures, but it all starts with connectivity. 

How does this incredible connection process take place? 

Let's break it step by step —

1. OOB Connectors 

OOB Connectors, or 'Plug & Play' as commonly known, play a significant role in the connection process. It's as simple as choosing the OOB connectors, connecting them to the Salesforce org, and gaining access to all the objects. This information can be linked to Salesforce Service, making the connection process straightforward and efficient. 

• All Salesforce 
• SFTP
• All Hyperscalers
• MuleSoft
• >50 New Last Quarter Beta 

Example: Receive the Data from Oracle ERP connected to SalesforceNow.

2. API, Kinesis & SDKs

Additionally, there are other forms of connection. The Data Cloud also consists of APIs and SDKs. Kinesis is an AWS way of integration. 

So, in this case, you can extract streaming information for the customer's Kinesis pipeline. 

• Streaming & Batch API
• Kinesis
• Web or Mobile SDKs

For example- You can receive streaming data from the customer's kinesis pipeline. 

So, the first two parts (i.e., the OOB connectors, API, and Kinesis—they involve the data connection part).

3. Data Transformation

The critical question is, how does data transformation happen? Now that we have looked into various integration methods, the connection will happen through pipelines and integration. 

• Streaming Transforms
• Batch Transforms
• Visual SQL Experience

Example– Light Data Cleansing or Bucketing 

There are two ways of Transformation:

I. Streaming Transformation

In this method, you receive real-time data immediately. It is a real-time transformation. In this case, you get the data immediately and process it on a request basis. Let's move to the next one. 

II. Batch Transformation

The batch transformation is more like a synchronous transformation. The data visualization can also happen when you opt for Batch Transformation. 

III. Visualize/SQL Experience

If needed, you can visualize the data or use the SQL experience option.

4. Canonical Data Model - The Harmonization Stage

The canonical data model involves harmonization. Herein, this process is industry-specific. 

• OOTB Data Model
• Industry-centered Data Models
• Customizable & Extensible Models

Example– Loyalty, Objects, FINS/HLS Specific  Data Models.

5. AI Mapper

For instance, you might have observed in a data loader that once you have mapped all the columns of a sheet, you can match it to the respective configuration. 

Example: First Name/ Last Name automatically maps to the correct attributes of the Individual. 

Automatically map source field to the OOTB Data Model.

Example– First Name/ Last Name automatically maps to the correct attributes of the Individual. 

Mapping Fields In The Data Cloud 

Ingestion & Modeling

Ingestion— As I shared earlier, the data is first ingested before proceeding with any other process. It is stored in the form of objects. This information is retrieved from the source via connectors to establish communication between servers. 

This way, data is constantly accessed. Through data streams, the nature and timings of connections are firmly established. 

While I have already shared a few connectors, here is more. This includes Salesforce CRM Connectors, B2C Commerce Connectors, Google Cloud Storage Connectors, Market Cloud Personalization Connectors, and more.

Data Modeling— Once all the data streams are ingested into the system, there is a source-to-target mapping experience. For instance, you can utilize the Customer360 Data model’s notion of an individual ID tag, with the source field corresponding to an individual data stream, customer journey, service issue, or other data. 

In the following flowchart, I will simplify the methods of ingestion and modeling in detail. Check below:

Data Cloud Harmonization

Now, let's get into the details of the Data Cloud Harmonization. 

As we have looked at above, harmonization is simply bringing data of various kinds from a diverse source together to function in an integrated manner. 

To simplify it further, information from multiple sources is visualized when it is 100% aligned with the respective sources. Also, note that every source has a unique data point aligned with your campaigns (let's say, a marketing campaign). 

Now, let us take an example:—

Consider,  there is a marketing campaign where you are tracking likes, reshares, and open rates for email. 

When it is harmonized, it reflects under one single view. Check below:

Why is the Harmonization Of Data So Important?

Harmonization is a crucial step in unifying data, which is the soul of the data cloud. This step allows the data to merge with all the procedures:

In the context of a marketing campaign, it's important to note that no single data point contains all the relevant information. This underscores the need for data harmonization. 

Without harmonizing all data onto a single platform, analyzing each source individually looms large. This is a significant challenge for companies, consuming vast amounts of time and energy.

 So, when the information is not harmonized individually, you won't be able to gain insights from a larger perspective, eventually hindering the unification part. 

Now, let's delve into the next concept- the Harmonization Center. This is a centralized platform where all data is harmonized, ensuring consistency and facilitating comprehensive analysis. 

What is the Harmonization Center?

Let us consider Marketing Cloud Intelligence. One crucial part is the Harmonization Center, designed with a no-code approach. This approach enables non-tech enthusiasts to access the technology as much as it helps tech-savvy experts. 

There are four different tools:

• Data Classification
• Patterns & harmonized dimensions
• Media name builder app
• Fix & Maintain

These are the different tools dedicated to data harmonization. 

What’s Next?

Now that we are thorough with the harmonization process of the data cloud, it is time to move to Chapter 3. I shall take you through a detailed ride for zero-copy data and related insights here. Stay tuned, as I see you next week.