Salesforce Data Cloud helps businesses break down data silos, gain a deeper understanding of their customers, and leverage AI to build smarter applications. Understanding its capabilities is key for developers creating more data-driven applications that deliver better customer experiences.

In this blog post, we’ll walk you through a clear and concise diagram of Data Cloud, highlight its core capabilities, and provide links to further reading for each capability along the way.

Access your key data sources

The foundation of any powerful data platform is the ability to access data from disparate systems in your enterprise. For Data Cloud, accessing your data is the critical first step. Data streams are capabilities that make this as efficient as possible.

Ingest structured and unstructured data

Data streams

Data streams are connections that continuously ingest data from various enterprise data sources. Connectors are available to many common applications, and data can be streamed, scheduled, or made available using zero-copy capabilities.

Each ingestion mechanism fits a broad number of use cases for web, mobile, and common enterprise applications and scales to huge volumes of data. Zero-copy capabilities allows Data Cloud to access data stored in external platforms without physically copying or moving the data itself.

Connect unstructured data

Connectors allow you to connect unstructured data files from your blob storage systems. Data Cloud doesn’t copy the files, but instead extracts their metadata to use when making their contents searchable across a vector database.

Represent data in a consistent format

Data sources have differing schemas. Structured data will have table and column names that differ but need to be represented in the same way. Data mapping is the capability that maps your source data to a common data model. For example, Contacts, Guests, Customers, and People are all individuals with the same attributes and map to the Individual object in Data Cloud. Businesses and Companies map to the Account object.

Map structured data sources to a common model

Unstructured data, such as emails, chat transcripts, PDFs, and knowledge articles, are a rich source of data that is hard to search using keywords. Indexing and embeddings help to convert your unstructured data to a format that is searchable and are key enablers for generative AI use cases.

Ingest unstructured data sources and create search indexes

Data source objects

Data source objects (DSOs) in Salesforce Data Cloud act as containers for the initial data brought into the system. They hold the information in its raw, original format before any transformations occur.

Data lake objects

Data lake objects (DLOs) are automatically created when data is ingested into Data Cloud. You can also create them manually. DLOs store structured data that are typically from sources like databases or CRM systems. Formulas and transforms can be added to augment the data when needed.

Unstructured data lake objects (UDLOs) are designed to make unstructured data, such as text files, emails, and chat transcripts, searchable. UDLOs are created manually and reference the unstructured data location. Each row within the objects corresponds to a single file on the external blob store.

Data model objects

Data model objects (DMOs) are typically mapped to data model objects (DMOs) which define a common structure for the data. There are standard data model objects provided or you can create custom objects when necessary.

Unstructured data model objects (UDMOs) are objects that form the basis for using embedding models to create vectors from your unstructured data.

Data lake objects and data model objects form the backbone of data used to drive insights and they provide data to predictive and generative AI services.

Unify data sources for the same person and business

Data about your customers and businesses is typically stored across a number of systems. Each source likely has some overlap but also contains unique details, and it is sometimes difficult to know if records across these systems relate to the same person without a common identifier. Identity resolution helps connect these dots, creating a single, comprehensive profile for each customer or company.

Unify individuals and businesses to one profile with identity resolution

Identity resolution

Built-in matching and reconciliation rules can identify data that is likely to be related to the same person or business. Matching rules allow you to add criteria based on data in the common data model with match methods, such as fuzzy matching (where entries are approximately similar but not identical), exact matching, and normalized exact matching (where entries are transformed to address issues like trailing spaces, inconsistent formatting, special characters, etc).

Data Cloud can also reconcile data based on the best value to save to a unified profile. LAST UPDATED, MOST FREQUENT, and SOURCE PRIORITY gives administrators the ability to ensure that data is consistent and of high quality.

Drive insights by enriching your data

Accessing data, mapping it to a common model and unifying data are important to enabling a consistent way to drive insights. Data Cloud capabilities help identify interests, buying habits, and online behavior, allowing you to make data-driven decisions for marketing, sales, operations, and customer service.

Enrich your data with insights, data graphs, AI and vector search

Vector search

By bringing unstructured data into Data Cloud, you can ground your generative AI, analytics, and automation search results with business-specific data that delivers deeper insights for your users and customers. Unstructured information, such as free-form text, chat transcripts, PDFs, and emails can be brought into Data Cloud, broken into meaningful chunks, and converted into machine-readable vector embeddings. These vector embeddings are then added to a search index, which can be used to perform vector search queries from apps like Prompt Builder, Agentforce, or Tableau.

Data graphs

A data graph combines and transforms normalized table data from data model objects (DMOs) into new, materialized views of your data. Because the data is pre-calculated, you can make fewer calls, and queries respond in near real-time.

Segments

You can use segmentation to break down your data into useful segments to understand, target, and analyze your customers. You can create segments on objects from your data model, and then publish them on a chosen schedule or as needed.

Insights

Data Cloud offers multiple ways to gain insights from your data, including calculated, streaming, and real-time insights. Insights help analyze datasets for historical trends, and long and short-term customer behavior.

Model Builder

You can create and deploy generative and predictive AI models. Build models with a user-friendly interface and guided steps, or Bring Your Own Model with providers like Databricks, Amazon Sagemaker, Google Vertex AI, and OpenAI. This enables you to enrich your data with predictive insights and power generative AI capabilities with custom large language models (LLMs).

Einstein audit trail

The Einstein audit trail in Data Cloud is all about tracking how Salesforce uses generative AI. Specifically, it focuses on the Einstein Trust Layer, a set of features designed to ensure the safety and accuracy of these AI-generated responses. Data Cloud collects and stores two types of data for Einstein Generative AI: generative AI audit data and feedback data. Pre-built Data Cloud reports and dashboards can be used to analyze the usage and share insights, such as user trends, acceptance rates, and response feedback

Taking action on data changing in your enterprise

A key capability of Data Cloud is its ability to put insights into the hands of customers, employees, and partners who need it to drive tangible business outcomes.

Acting on data in Data Cloud offers a breadth of capabilities to put data into the hands of teams that need it

Data shares

Data shares provide easy access for third-party partners to use data in Data Cloud. With access to data lake objects (DLOs), data model objects (DMOs), and calculated insight objects (CIOs), you can build powerful analytical insights and machine learning models. Data Cloud objects are shared without replicating the data into a partner ecosystem. With this zero-copy data sharing, you get near real-time access to data.

APIs

You can programmatically extract your data from Data Cloud using APIs. Retrieve metadata and query key data points using well-defined REST APIs. Check out the Postman collection for details. For Python developers, there’s a connector that uses the Query API and extracts data from Data Cloud.

The Models API (Beta) provides Apex classes and REST endpoints that connect your application to large language models (LLMs) from Salesforce partners, including Anthropic, Google, and OpenAI. The capabilities of the Models API are expressed as Apex methods or REST endpoints. Generate chat, embeddings, and text, and submit feedback on any generated text.

Data actions

Data Cloud enables you to monitor data changes in data model objects and calculated insight objects, and send the change data events to a Salesforce Platform Event, Marketing Cloud, or a webhook.

A data action can enable different types of event-driven integrations and orchestrations. For example, you can:

  • Orchestrate Salesforce CRM workflows with insights and data events from Data Cloud
  • Integrate data actions in Mulesoft Anypoint by sharing aggregated event data with external partners based on criteria
  • Integrate with SaaS applications with signals from Data Cloud
  • Trigger serverless functions that work with a webhook based on insights in Data Cloud
  • Push unfiltered insights and engagements to your data lake for analysis and storage

Activations

Activation is the process that publishes a segment to activation platforms. An activation target is used to store authentication and authorization information for a given activation platform. You can publish your segments, include contact points, and add additional attributes to the activation targets. Activation targets can be file-based storage, ad services or marketing and B2C commerce clouds.

Enrichments

Data Cloud enrichments makes data from Data Cloud available in standard Salesforce components that you can add to specific record pages. For example:

  • Add related lists to key objects in Salesforce to show related information directly from Data Cloud with related list enrichments
  • And fields with data from Data Cloud to your Contact, Account and Lead record pages with copy field enrichments

Automation

You can build automation workflows based on data changing in Data Cloud, get records, and start administration jobs using Flow Builder to orchestrate changes within the context of Salesforce data. For example:

Analytics

Data Cloud allows you to use various tools and products to analyze your data with direct data connectors for Salesforce, CRM Analytics, and Tableau. Developers can also use the APIs provided for programmatic access, or use the Java Database Connectivity (JDBC) driver.

Conclusion

Salesforce Data Cloud helps developers unify and leverage their customer data across the enterprise. You can then build data-driven applications that unlock AI-powered insights and deliver new customer experiences with insights that were previously unattainable.

To reinforce your understanding of Data Cloud, watch our video series on Data Cloud Fundamentals, where we walk you through a tangible use case and explain each step in detail.

Resources

About the author

Dave Norris is a Developer Advocate at Salesforce. He’s passionate about making technical subjects broadly accessible to a diverse audience. Dave has been with Salesforce for over a decade, has over 35 Salesforce and MuleSoft certifications, and became a Salesforce Certified Technical Architect in 2013.

Get the latest Salesforce Developer blog posts and podcast episodes via Slack or RSS.

Add to Slack Subscribe to RSS