Home / architect / data-cloud-vs-analytics-decision
data-cloud-vs-analytics-decision
Use when choosing or explaining how Salesforce Data Cloud (Data 360) and CRM Analytics (Tableau CRM / Einstein Analytics) fit together versus overlap — unified data platform vs analytics consumption, Direct Data on Data Model Objects (DMOs), identity and activation vs dashboards and recipes.
Open the skill folder on GitHub
Data Cloud vs CRM Analytics — Decision Framework
Use this skill when stakeholders treat Data Cloud and CRM Analytics as interchangeable “analytics products,” or when you must document how they complement each other in a Data 360 architecture. The guidance stops at platform boundaries and decision logic; it does not replace product-specific implementation skills.
Salesforce positions Data Cloud (Data 360) as the real-time customer data platform layer: ingest from many sources, harmonize to a canonical model, resolve identity, and activate segments. CRM Analytics is the Salesforce-native analytics and visualization layer for exploring metrics, building dashboards, and operationalizing insights—often on CRM data, and optionally on Data Cloud through supported connectivity patterns that work against harmonized Data Model Objects rather than ad hoc raw lake files. Confusing the two leads to duplicate pipelines, wrong sizing of licenses, and analytics projects that bypass identity resolution entirely.
Before Starting
Gather this context before recommending one platform, the other, or both:
- Outcome priority: Do you need cross-channel unified profiles and outbound activation (advertising, Marketing Cloud, CRM updates), or primarily dashboards and exploration for Salesforce users?
- Source spread: Is the analytical population of record only in Salesforce CRM objects, or scattered across warehouses, behavioral events, loyalty, and third-party SaaS?
- Semantic layer: Will analysts work on harmonized DMO semantics (Individual, Contact Point Email, Engagement), or on native Opportunity, Case, and custom CRM objects?
- Time sensitivity: Do segments or metrics need to reflect streaming or frequently refreshed harmonized data, versus scheduled CRM extracts?
- Security and consent: Will analytics run inside the Salesforce trust boundary with permission-set governed access, and does marketing activation require Data Cloud consent and policy tooling?
Questions to Ask Before Configuring
Each question traces to a gotcha in references/gotchas.md.
| Question | Why it matters | What a good answer adds | What proper configuration adds over just doing it |
|---|---|---|---|
| "Which use cases need cross-source identity, segments, or activation, and is Segmentation and Activation in the contract?" | Those are Data 360 responsibilities CRM Analytics does not replace (Gotcha 1) | A use-case register with an owning platform per row | No dashboard built on a "golden customer" that identity resolution never produced |
| "For each analytics subject area, does CRM Analytics read Data 360 in place or from converted datasets, and which DMOs?" | The two paths differ in freshness, scope, and maintenance (Gotcha 2) | Named DMOs and a named path per subject area | A proof of concept scoped to objects the path supports |
| "What freshness does each persona need, and what is the slowest hop?" | Latency adds up across ingestion, identity, insights, and schedules (Gotchas 3, 7) | A latency chain per persona | SLAs set on the pipeline, not blamed on a product |
| "Who owns Data 360 consumption, and which workloads draw it?" | Data 360 usage is metered as credits, not seats (Gotcha 4) | A consumption owner and a review cadence | A cost model that survives the first full ingestion |
| "Does a warehouse already hold modeled customer data?" | Zero copy federation and Data Shares can replace bulk ingestion (Gotcha 6) | An ingest, federate, or exclude decision per source | No duplicated storage and modeling |
| "Will any app or report query DMOs with SOQL?" | 12 MB result cap, no currency aggregates, no calculated insight objects (Gotcha 5) | A query path per consumer | Components that do not fail in production on volume |
What a proper decision adds over "buying both and turning on dashboards": one owner per layer, named objects and paths, and a cost and latency model written down before the first pipeline runs.
Core Concepts
Data Cloud: ingestion, harmonization, identity, activation
Data Cloud is architected as a lakehouse-style customer data platform. Data lands in Data Source Objects, is cleansed into Data Lake Objects, mapped into Data Model Objects (DMOs), and then identity resolution produces Unified Individual profiles for segmentation and activation. Its primary value is consistent cross-system semantics, identity graph quality, and governed outbound use—not chart authoring alone. Detailed layer design lives in architect/data-cloud-architecture.
CRM Analytics: analytics, datasets, and embedded experience
CRM Analytics provides datasets, lenses, dashboards, and recipes for metrics and storytelling, embedded in Salesforce for licensed users. It can consume CRM-sourced datasets through native sync patterns familiar to Salesforce teams. When Data Cloud is in play, two paths exist: converting Data 360 data model objects into CRM Analytics datasets (documented in the CRM Analytics REST API guide), and Direct Data, which queries Data 360 in place (UNVERIFIED (2026-10-03): Direct Data scope is documented in Salesforce Help only). Either way, analytics runs against the harmonized model rather than re-implementing every join in CRM.
Complementary pattern vs false substitution
The common enterprise pattern is complementary: Data Cloud becomes the system of insight for unified, multi-source truth; CRM Analytics becomes a consumer of that truth for interactive analysis and operational dashboards, alongside CRM-native datasets where appropriate. The anti-pattern is purchasing CRM Analytics alone and assuming it replaces ingestion, identity resolution rulesets, activation target configuration, and cross-cloud harmonization—those remain Data Cloud concerns.
Common Patterns
Pattern: Data Cloud first, CRM Analytics as visualization layer
When to use: Multiple non-CRM sources must inform revenue, service, and marketing metrics; identity resolution is non-negotiable; activation is on the roadmap.
How it works: Land and harmonize in Data Cloud, stabilize DMO quality, then connect CRM Analytics to the relevant DMO-backed sources for exploration and embedded dashboards.
Why not CRM Analytics alone: Recipes and dataflows on CRM objects do not automatically deliver cross-cloud identity or activation semantics; you would rebuild fragile point-to-point integration.
Pattern: CRM Analytics on CRM only (Data Cloud later)
When to use: All authoritative dimensions and facts already live in Salesforce, no activation to ad platforms, and no external behavioral feeds.
How it works: Stand up CRM Analytics datasets from standard and custom objects; defer Data Cloud until cross-cloud use cases materialize.
Why not Data Cloud first: Introduces operational surface area without a clear multi-source or activation driver.
Pattern: Dual footprint with clear ownership
When to use: Large org with both embedded Salesforce KPIs and a separate enterprise BI tool; you need a written boundary so teams do not duplicate harmonization.
How it works: Document that Data Cloud owns cross-source harmonization and segments while CRM Analytics owns Salesforce-embedded analytics contracts; reference integration patterns for sync vs query boundaries.
Why not implicit assumptions: Prevents two teams from building incompatible “golden customer” definitions.
Decision Guidance
| Situation | Recommended emphasis | Reason |
|---|---|---|
| Cross-system customer profile, identity graph, segments, activations | Data Cloud as primary platform | Harmonization, identity, and activation are native responsibilities |
| Sales and service dashboards on CRM objects only | CRM Analytics on CRM data | Faster time-to-value when sources are already in the org |
| Analysts need metrics on harmonized Individual / engagement DMOs | Data Cloud + CRM Analytics (Direct Data on DMOs) | Analytics consumes the harmonized layer instead of re-joining sources |
| Real-time or high-velocity behavioral signals from web or mobile | Data Cloud ingestion and modeling | CRM Analytics is not a substitute for streaming ingestion design |
| Marketing compliance, consent, and policy for outbound audiences | Data Cloud governance patterns | Activation use cases pair with platform-specific controls |
Recommended Workflow
- Capture the primary outcome (activation, embedded BI, data science handoff, or compliance) and list every data source with its system of record.
- Decide whether harmonization and identity resolution are in scope; if yes, treat Data Cloud as the owning layer for cross-source truth.
- If analytics is required on harmonized entities, confirm the connectivity path per subject area: Direct Data in place, or DMOs converted to CRM Analytics datasets. Reject informal copies into custom CRM objects. Record the use-case register from
references/examples.md. - Document licensing and operational ownership separately for Data Cloud administration versus CRM Analytics authoring so cost models stay honest.
- Peer-review the decision with
architect/data-cloud-architecturefor pipeline depth andadmin/einstein-analytics-basicsfor CRM Analytics delivery once the boundary is set. - Archive the outcome using
templates/data-cloud-vs-analytics-decision-template.mdfor auditability.
Review Checklist
- [ ] Stakeholders can articulate what Data Cloud owns versus what CRM Analytics owns in one sentence each.
- [ ] Any “CRM Analytics instead of Data Cloud” narrative has been tested against activation and multi-source requirements.
- [ ] Each analytics subject area names its path (Direct Data or DMO-to-dataset conversion) and its DMOs, not raw stream reporting.
- [ ] Identity resolution and segment latency assumptions are documented with realistic ranges.
- [ ] Related implementation work is routed to the correct admin or integration skills.
Salesforce-Specific Gotchas
Full detail and sources in references/gotchas.md. The short list:
- CRM Analytics does not replace identity resolution, segmentation, or activation.
- CRM Analytics reaches Data 360 either by converting DMOs to datasets or through Direct Data; name the path.
- Latency stacks across ingestion, identity, calculated insights, and CRM Analytics schedules.
- Data 360 is metered by credit consumption, not seats.
- SOQL on Data 360 objects caps results at 12 MB, rejects currency aggregates, and cannot read calculated insight objects.
- A CRM Analytics sync produces a connected object, not a dataset.
Output Artifacts
| Artifact | Description |
|---|---|
| Decision summary | Short narrative of platform roles, licensing boundaries, and connectivity |
| Architecture review appendix | Table mapping use case → owning platform → follow-on skill |
Related Skills
- architect/data-cloud-architecture — DSO, DLO, DMO, identity resolution, activation design after you decide Data Cloud is in scope.
- architect/crm-analytics-vs-tableau-decision — When the debate is third-party BI versus Salesforce-native analytics, not Data Cloud.
- admin/einstein-analytics-basics — Recipes, dataflows, and dashboard implementation for CRM Analytics.
- architect/analytics-data-architecture — End-to-end CRM Analytics pipeline patterns on large CRM datasets.