Data Cloud
Data Cloud is Salesforce's customer data platform (CDP) for unifying data from across an organization's systems into a single, real-time customer profile.
Definition
Data Cloud is Salesforce's customer data platform (CDP) for unifying data from across an organization's systems into a single, real-time customer profile. It ingests data from Salesforce orgs, external databases, data warehouses, marketing platforms, web and mobile event streams, and any other source, then resolves identities, builds unified profiles, computes calculated insights, and exposes the unified data back to Sales Cloud, Service Cloud, Marketing Cloud, Agentforce, and external systems. Data Cloud is the platform's strategic play for becoming the operational data layer underneath every customer engagement.
Originally launched as Customer 360 Audiences, then Genie, and finally rebranded as Data Cloud, the product has evolved into the foundation Salesforce now uses for Agentforce grounding, Marketing Cloud Engagement personalization, and cross-cloud customer journey orchestration. Data Cloud is licensed separately from base Salesforce, with consumption-based pricing measured in credits across ingestion volume, profile counts, computed insights, and activations. The data lives in Salesforce's hyperscale columnar store (built on Snowflake/Iceberg architecture) optimized for analytical and AI workloads rather than transactional ones.
How Data Cloud unifies customer data for AI and engagement
Data sources, connectors, and ingestion patterns
Data Cloud ingests from many sources. Salesforce CRM Connectors pull data from Sales Cloud and Service Cloud orgs in near-real time. Cloud storage connectors (Amazon S3, Google Cloud Storage, Azure) batch in historical data sets. Streaming connectors (Kafka, AWS Kinesis) push event data continuously. Marketing Cloud, Mulesoft, and Snowflake connectors bring data from specific Salesforce-adjacent systems. Web and mobile SDKs capture customer behavior from web pages and apps. Each source feeds into Data Cloud''s ingestion layer where it gets normalized into the standard data model.
The Data Model: Data Streams, Data Lake Objects, Data Model Objects
Data Cloud organizes data in three layers. Data Streams are the raw ingested data from each source, kept as-is for audit and reprocessing. Data Lake Objects (DLOs) are the cleaned, normalized version after initial transformation. Data Model Objects (DMOs) are the unified business model: a single Customer DMO that merges identity across all sources, a single Order DMO that consolidates transactions, a single Interaction DMO that combines touchpoints. The DMOs are what downstream consumers (Sales Cloud, Marketing Cloud, Agentforce) query against.
Identity resolution and the unified profile
Identity resolution is Data Cloud''s killer feature. Customers exist across many systems with different identifiers: a Contact in Salesforce, a subscriber in Marketing Cloud, an account in a billing system, a visitor in web analytics. Identity Resolution Rules use deterministic matching (exact email match) and probabilistic matching (similar name plus address plus phone) to merge these into one Unified Individual record. The unified profile becomes the single customer view that drives personalization, AI grounding, and cross-channel orchestration.
Calculated Insights and aggregated metrics
Calculated Insights compute aggregated metrics on top of unified profiles. Lifetime Customer Value, Average Order Size, Days Since Last Purchase, Predicted Churn Probability. The platform supports SQL-style definitions and Apex-compiled formulas. Insights run on the Data Cloud query engine optimized for analytical workloads at scale; computing an insight across millions of customers takes seconds, not the minutes or hours equivalent custom-object aggregation in Sales Cloud would take. Insights become first-class fields on the Unified Individual, available to every downstream consumer.
Segmentation and activation
Once unified profiles and insights are in place, Data Cloud supports segmentation: building audiences based on profile attributes and calculated insights. High-value customers who have not purchased in 60 days. New customers in a specific region. Customers who clicked the last campaign but did not convert. Each segment can be activated to a destination: Marketing Cloud for email/SMS campaigns, Sales Cloud as a Lead list, Google Ads as a custom audience, Meta as a custom audience, custom destinations via API. Activations push segment membership to the destination on a schedule.
Zero-copy and the federated query model
Modern Data Cloud uses zero-copy federation for some data sources. Instead of ingesting data physically, Data Cloud queries it where it lives (Snowflake, Databricks, Google BigQuery) through standard Iceberg or other zero-copy protocols. This reduces ingestion cost, eliminates data duplication, and keeps the source system as the single source of truth. Zero-copy is the architectural direction; not every connector supports it today but more do each release.
Data Cloud and AI: grounding Agentforce and prompts
Data Cloud is the primary grounding source for Agentforce agents and Prompt Builder templates that need broader context than a single Salesforce record. An agent answering a customer question grounds on the customer''s unified profile, recent interactions, lifetime value, and segment membership. The combination of Data Cloud''s unified data and Einstein''s generative AI is the architectural pattern Salesforce is steering customers toward; each makes the other more powerful.
How to set up Data Cloud
Setting up Data Cloud is a multi-month exercise that touches data sources, identity resolution, data modeling, segmentation, and activation. It is significantly more complex than configuring traditional Sales Cloud or Service Cloud. Plan with an experienced data architect and a phased rollout; trying to ingest everything at once produces overwhelming complexity.
- Define the unified data model and identity strategy
Decide which customer types matter (B2C individuals, B2B contacts, both), which data sources will feed in, and what identity resolution strategy applies. The data model is the foundation; trying to change it after data is loaded is painful.
- Provision Data Cloud and configure the connected org
Setup > Data Cloud > Setup. Connect to the Sales Cloud or Service Cloud orgs that source customer data. Configure data spaces (logical containers, often one per business unit or region).
- Set up data streams from each source
Add data streams for each source: Salesforce CRM, marketing platforms, data warehouses, web analytics, mobile event streams. Configure the schema and refresh frequency for each stream.
- Map data streams to Data Model Objects
Data Cloud ships standard DMOs (Individual, Account, Order, Engagement, Email Engagement). Map fields from your data streams to the appropriate DMOs. Customize DMOs where the standard model does not fit.
- Build identity resolution rules
Setup > Identity Resolution. Define match rules with deterministic and probabilistic logic. Run the resolution job and review the unified profiles. Tune rules iteratively until the merge quality matches expectation.
- Define calculated insights
Setup > Calculated Insights > New. Build the insights that downstream consumers need: LTV, churn probability, segment membership scores. Each insight becomes a queryable attribute on the unified profile.
- Build segments and configure activations
Setup > Segments > New. Define audiences using profile attributes and insights. Configure activations to target destinations: Marketing Cloud, Sales Cloud, ad platforms, custom APIs.
- Connect to Agentforce and Prompt Builder for AI grounding
Expose Data Cloud DMOs as grounding sources in Agentforce and Prompt Builder. AI features can then query unified profiles and insights directly during agent conversations and prompt invocations.
Salesforce CRM, cloud storage, streaming, marketing platforms, custom APIs. Drives what data flows into Data Cloud.
Deterministic and probabilistic match rules that merge customer records across sources into unified profiles.
Where Data Cloud sends segment membership: Marketing Cloud, Sales Cloud, ad platforms, custom APIs.
- Data Cloud is consumption-priced. High data volumes and activations produce significant credit consumption. Monitor usage and plan budget alongside the implementation.
- Identity resolution requires iteration. The first version of match rules rarely produces perfect merges. Plan multiple tuning cycles with real data before declaring identity resolution complete.
- The data model is foundational. Changing the Data Model Object structure after data is loaded triggers reprocessing across millions of records. Get the model right early.
- Data Cloud is operationally separate from base Salesforce. The familiar admin patterns (page layouts, profiles, sharing rules) differ in Data Cloud; learning the platform-specific UX takes time.
- Activations have latency. Real-time activations exist but most ingestion and segmentation runs on minute-to-hour cycles. Plan downstream use cases around the latency profile.
Trust & references
Cross-checked against the following references.
- Data Cloud Product PageSalesforce
- Data Cloud OverviewSalesforce Help
Straight from the source - Salesforce's reference material on Data Cloud.
- Data Cloud SetupSalesforce Help
- Identity ResolutionSalesforce Help
- Segmentation and ActivationSalesforce Help
About the Author
Dipojjal Chakrabarti is a B2C Solution Architect with 29 Salesforce certifications and over 13 years in the Salesforce ecosystem. He runs salesforcedictionary.com to help admins, developers, architects, and cert/interview candidates sharpen their fundamentals. More about Dipojjal.
Test your knowledge
Q1. What does Data Cloud represent in the Salesforce Platform?
Q2. Who can benefit from understanding Data Cloud?
Q3. How does Salesforce's multi-tenant model affect Data Cloud?
Discussion
Loading discussion…