Salesforce Data Cloud Consultant cheat sheet

End-to-end pipeline: Data Stream → Data Lake Object (DLO) → Data Model Object (DMO) → Identity Resolution → Unified Profile → Insight/Segment → Activation.

A DLO holds raw ingested records in lakehouse storage; a DMO is the harmonized, modeled object. Data only becomes usable for identity resolution, segments, and activations after DLO-to-DMO mapping.

Standard profile DMO is Individual; Contact Point Email, Contact Point Phone, Contact Point Address, Party Identification, Account, and Engagement are also standard DMOs.

Data spaces partition streams, DLOs, DMOs, segments, and activations; objects are not shared across data spaces and a single segment cannot span data spaces.

Standard DMO categories are Profile, Engagement, Related, and Other; custom DMOs are created in the data model with a category, primary key, and fields.

Core APIs: Ingestion API (write records), Query API (SQL reads), Profile API (single-individual profile reads), Connect API (segment/activation operations), Metadata API (deployment).

Data Cloud is a lakehouse: Parquet files in object storage queried by a distributed SQL engine, so data is not copied into the core CRM relational database.

Consumption is metered in credits across profiles, segmentation and activation volume, data storage, and API/query usage.

The Unified Profile is the consolidated individual record read by CRM components, Flow data actions, segments, and activation targets.

Ingestion options include the Salesforce CRM connector, Ingestion API (bulk and streaming), SFTP, Amazon S3, Google BigQuery, Snowflake, SharePoint/Google Drive, Marketing Cloud, MuleSoft, and the Web/Mobile SDK.

Every data stream requires a source object, a primary key (optionally composite), and a timestamp field; the timestamp drives incremental and scheduled refresh behavior.

Streaming data streams land records in near real time and can feed streaming insights and real-time data actions; batch streams refresh on a configured schedule.

DLO-to-DMO field mapping supports direct mapping and formula-based transformation; unmapped DLO fields are not available downstream for insights, segments, or activations.

Data transforms apply SQL-based filtering and reshaping during ingestion, reducing stored volume and improving fit against the standard data model.

The Ingestion API accepts JSON payloads posted to an ingest resource path under /services/data/vXX.X/ingest/sources/{source}/{object}, authorized by an OAuth connected app.

Custom DMOs are defined with a category, a primary key, and fields; relationships to Individual or Account are typically modeled as Related DMOs.

Deleting a data stream also removes its DLO; if ingestion resumes, the stream, DLO, and mappings must be recreated.

A ruleset has two components: match rules (which source records are the same entity) and reconciliation rules (which field value wins on the unified record).

Match rules may be exact or fuzzy; fuzzy rules tolerate typos and formatting variants and are tuned per field with similarity thresholds.

Reconciliation options include source priority, most recent, and most frequent value; a default reconciliation rule applies to any field without a field-specific rule.

Output is a Unified Individual record per resolved entity; Party Identification stores source record identifiers and their link to the unified profile.

Rulesets are defined per object type (Individual or Account); only one ruleset is active for a given object type at a time.

Identity resolution runs on a schedule or on demand; results change as new or updated source records arrive.

Match rule sequence matters: rules are evaluated in order, so broad rules placed early can capture matches that narrower rules were intended to handle.

Identity resolution links records; it does not merge or delete source records in the data lake.

Calculated Insights are SQL definitions over one or more DMOs that produce metric objects (for example lifetime value or order count) on a refresh schedule.

Writeback enabled calculated insights persist results as DMOs, making the metrics usable in segmentation and activation instead of display-only.

Streaming Insights run continuously against streaming data for near real-time metrics used in personalization and real-time segments.

Segments are built on a DMO (usually Individual) using include and exclude criteria, and can filter on fields of related DMOs through relationship paths.

Segment membership is persisted in Segment Membership DMOs, with each segment represented by its own segment record and on/off state.

Segments can be published to activation targets, or consumed inside Salesforce through Flow and Data Cloud segment elements.

Refresh cadence depends on segment type and edition: batch segments refresh on a schedule, while streaming segments update as qualifying events arrive.

Data Graphs in Profile Explorer present a unified view of related records for exploration and personalization; they are not the object that defines segment criteria.

An activation is the combination of a segment or insight, an activation target, field mappings, and a refresh schedule; each segment-target pair is a separate activation.

Common activation targets include Marketing Cloud Engagement, Marketing Cloud Personalization, Salesforce CRM, Google Ads, Meta Ads, Amazon Ads, Snowflake, S3, and webhooks.

Ad platform targets typically require hashed email or phone identifiers and target-specific required fields before publishing succeeds.

Data Actions expose Data Cloud inside Flow: fetch a unified profile, update profile attributes, or request a personalization decision at run time.

The Profile API returns a single individual's unified profile in real time; the Connect API handles segment, activation, and metadata operations for integrations.

Deactivation sends removal or opt-out signals to a target for profiles that exit a segment, where the target supports it.

CRM integration surfaces include Unified Profile components on CRM records, Data Cloud segments in Flow, and Data Cloud data actions in Flow.

Activation refresh frequency governs how quickly membership changes reach the target; streaming-capable targets update as members qualify.

Architecture layers: ingestion, object storage holding Parquet files, a distributed SQL query engine over the lake, and a metadata/catalog layer serving profiles, insights, and segments.

Data Cloud runs on Hyperforce, so the selected region and cloud provider determine where data is stored and processed.

Access is controlled with permission sets plus data space membership; users see only the objects inside the data spaces they belong to.

Data masking policies hide sensitive field values from users lacking the required permission while keeping the underlying values usable in matching.

Consent is modeled with consent-related DMOs (including contact point consent) and can be applied as segment criteria before activation.

Privacy Center connects to Data Cloud to locate and act on data subject requests, such as access and deletion, across ingested sources.

Data kits/bundles and the Metadata API move data model, stream, and segmentation metadata between orgs; the underlying data records are not deployed.

Retention settings on data streams and DLOs delete aged records; retention should be configured before high-volume ingestion to control storage and compliance exposure.

Monitoring relies on the Setup Audit Trail for configuration changes and on ingestion, identity resolution, and activation job status for pipeline failures.