Marketing Cloud Integration Architecture

Marketing Cloud is Salesforce’s B2C marketing platform, and architecturally it is a separate platform from Core Salesforce (Sales/Service Cloud). This separation creates integration challenges that surface regularly in CTA scenarios. The data model, sync mechanisms, and integration patterns are required knowledge for any multi-cloud architecture.

Separate platform architecture

Marketing Cloud Engagement (MCE) was built on the ExactTarget acquisition and runs on its own infrastructure, database, and data model. It does NOT share the Salesforce object model. Every piece of data must be explicitly synced or integrated. CTA candidates must internalize this architectural separation.

---

Architecture Overview

Figure 1. Marketing Cloud Engagement runs on separate infrastructure from Core Salesforce, so all data must be explicitly synced or integrated. MC Connect handles near-real-time CRM data sync while the REST and SOAP APIs handle transactional messaging. Data Cloud sits above both as the unified profile and segment activation layer.

---

Data Model: Contacts, Subscribers, and Keys

The Marketing Cloud data model is fundamentally different from Core Salesforce. The key identifiers are critical to get right.

Key Identifiers

Identifier	Scope	Description
Contact Key	All of MC	Unique identifier across all channels (email, SMS, push). You assign this value.
Subscriber Key	Email Studio	Legacy term - functionally identical to Contact Key in most implementations
Contact ID	Internal	System-generated numeric ID. Not user-facing.
Salesforce Lead/Contact ID	Core SF	The 18-character Salesforce record ID

CTA best practice - key alignment

Always align the Contact Key / Subscriber Key with the Salesforce Contact ID (or an external unique identifier). Misalignment is the #1 root cause of data integrity issues in MC implementations. When MC Connect is configured, Salesforce recommends using the CRM Lead/Contact ID as the Subscriber Key.

Contacts vs Subscribers

Concept	Definition	Implication
Contact	Any record in All Contacts - synced CRM records, API imports, mobile subscribers	Counts toward MC contact billing
Subscriber	A contact you have sent email to via Email Studio	Subset of contacts
All Contacts	Superset that includes everyone	CRM syncs add contacts even if never emailed

Data Extensions (DEs)

Data Extensions are the primary storage mechanism in MC. Think of them as custom objects in Core Salesforce, but with no relationships, triggers, or built-in automation.

DE Type	Sendable?	Use Case
Standard DE	Yes	Email sends, journey entry, segmentation
Non-Sendable DE	No	Lookup data, transaction history, reference tables
Synchronized DE	No	Mirror of CRM objects via MC Connect (read-only in MC)
Shared DE	Varies	Cross-Business-Unit data sharing in Enterprise accounts

Synchronized DEs are not sendable

Synchronized Data Extensions from MC Connect cannot be used directly for sends or journey entry. You must use a SQL Query or Filter Activity in Automation Studio to copy data from the synchronized DE into a sendable DE. This is a common gotcha.

---

Marketing Cloud Connect (MC Connect)

The native connector between Core Salesforce and Marketing Cloud Engagement. MC Connect enables near-real-time data synchronization and integrated marketing workflows.

What MC Connect Provides

Capability	Description
Synchronized Data Sources	Near-real-time sync of CRM objects (Contact, Lead, Account, User, custom objects) into MC Data Extensions
Salesforce Sends	Send emails from within Sales/Service Cloud using MC templates
Triggered Sends	Trigger email sends from Apex or Flow based on CRM events
Campaign Member Sync	Sync Salesforce Campaign membership for journey entry
Tracking Data Writeback	Email open, click, bounce, and unsubscribe data written back to CRM
Journey Builder Entry	Use Salesforce data events as journey entry sources

Sync Architecture

Figure 2. Synchronized Data Extensions from MC Connect are read-only and cannot be used directly for sends or journey entry. A SQL Query Activity in Automation Studio must copy data into a standard sendable DE before Journey Builder can use it as an entry source.

MC Connect Limitations

Limitation	Detail
250-field cap	Synchronized DEs limited to 250 fields per object
No delete sync	Deletes in CRM are not reflected in MC (records remain)
One-way read	Synchronized DEs are read-only in MC
Contact billing	Every synced Lead/Contact counts toward MC contact limit
Cannot move SDEs	Synchronized DEs exist at the top level; cannot move to shared folders
No undo	Once a sync is started, it cannot be deleted - only paused
Latency	Near-real-time (exact latency varies; typically minutes, not seconds)
No custom metadata	Custom metadata types and custom settings do not sync

---

Integration Patterns

API Integration

MC exposes two primary API families covering different functional areas.

API	Protocol	Covers	Authentication
REST API	REST/JSON	Journey Builder, Contacts, Content Builder, Transactional Messaging, MobilePush	OAuth 2.0 (Server-to-Server)
SOAP API	SOAP/XML	Email sends, Subscribers, Lists, Tracking, Data Extensions, Triggered Sends	OAuth 2.0 (legacy: username/password)

API coverage split

REST API covers newer MC features (journeys, mobile, content). SOAP API covers legacy features (email sends, tracking, subscriber management). Many implementations require both. For new integrations, prefer REST API where available.

Triggered Sends

External systems or CRM automation can initiate real-time email sends through triggered sends.

Method	Trigger Source	Best For
MC Connect Triggered Send	Salesforce Flow, Apex	CRM-initiated transactional emails (welcome, confirmation)
REST API Transactional Messaging	Any external system	Multi-channel transactional messages (email + SMS)
SOAP API Triggered Send	Any external system	Legacy email-only triggered sends
Journey Builder API Entry	REST API call	Complex multi-step journeys triggered by external events

Programmatic Content: AMPscript and SSJS

Language	Runs In	Use Case
AMPscript	Email, SMS, Landing Pages, CloudPages	Inline personalization, conditional content, data lookups
SSJS (Server-Side JavaScript)	CloudPages, Script Activities, Landing Pages	Complex logic, API calls within MC, WSProxy for SOAP calls
GTL (Guide Template Language)	Content Builder templates	Template-level logic and loops

---

Journey Builder Architecture

MC’s visual orchestration engine for multi-channel customer journeys.

Entry Sources

Entry Source	Trigger	Typical Use
Data Extension	Schedule or API	Batch campaigns, segmented audiences
Salesforce Data Event	CRM record create/update	Welcome journey on new Contact creation
Salesforce Campaign	Campaign member added	Campaign-triggered drip sequence
API Event	REST API call	External system triggers (e-commerce, web app)
CloudPages Form	Form submission	Lead capture, preference center
Audience	Data Cloud segment	Segment-driven activation

Activity Types

Figure 3. Journey Builder orchestrates multi-channel sequences by chaining entry sources, message activities, and flow control splits. Engagement splits branch based on prior message interaction, enabling behaviour-driven personalization without external logic.

Decision Logic Types

Split Type	Logic Basis	Example
Decision Split	Contact attribute or DE field value	VIP status, location, product interest
Engagement Split	Prior message interaction	Opened email? Clicked link?
Random Split	Percentage-based random	A/B test: 50/50 split
Einstein Split	AI-predicted engagement	Einstein Engagement Scoring likelihood

---

MC Engagement vs Account Engagement (Pardot)

A critical CTA decision. Which marketing platform fits the scenario?

Dimension	MC Engagement	Account Engagement (Pardot)
Primary use	B2C, high-volume consumer marketing	B2B, lead nurturing and scoring
Platform	Separate (ExactTarget infrastructure)	Native on Salesforce Platform
Data model	Data Extensions (separate from CRM)	Prospects linked to Leads/Contacts
Volume	Millions of contacts, billions of messages	Thousands to low hundreds of thousands
Channels	Email, SMS, Push, Social, Advertising, Web	Email, Landing Pages, Forms
Journey orchestration	Journey Builder (visual, multi-channel)	Engagement Studio (email-focused)
Personalization	AMPscript, Dynamic Content, Einstein	Merge fields, Dynamic Content, HML
Lead scoring	Not native (use Data Cloud or custom)	Native, tightly integrated with Sales Cloud
Reporting	Analytics Builder, Datorama/Intelligence	B2B Marketing Analytics (CRM Analytics)
CRM integration	MC Connect (sync required)	Native (same database)
Pricing model	Contact-based + super messages	Edition-based with prospect tiers
Best for CTA	Multi-channel B2C journeys, retail, media	B2B demand gen, ABM, considered purchases

CTA decision framework

Ask three questions: (1) B2B or B2C? (2) How many contacts - thousands or millions? (3) How many channels beyond email? If the answer is B2C + millions + multi-channel, the answer is MC Engagement. If B2B + thousands + email-centric, Account Engagement. Some enterprises use both - MC Engagement for customer marketing, Account Engagement for prospect nurturing.

---

Data Cloud + Marketing Cloud

Data Cloud (formerly CDP, now Salesforce Data 360) is the strategic integration layer between MC and the rest of Salesforce.

Capability	Description
Unified Profile	Merges CRM, MC, Commerce, and third-party data into a single customer profile
Segment Activation	Create segments in Data Cloud, activate them as MC journey entry sources
Calculated Insights	Compute engagement scores, LTV, propensity - use in MC personalization
Real-time Events	Stream behavioral events (web, mobile) to Data Cloud, trigger MC journeys
Identity Resolution	Match and deduplicate contacts across systems before syncing to MC
Zero-Copy Partners	Connect Snowflake, BigQuery, Databricks data without replication

Strategic direction

Salesforce’s strategic direction (Marketing Cloud Next, announced 2025) is to rebuild MC natively on the Salesforce Platform with Data Cloud as the foundation. For CTA scenarios set in the future state, position Data Cloud as the integration hub between CRM and marketing, replacing MC Connect for data sync and using Agentforce agents and flow-based orchestration alongside Journey Builder. Journey Builder is not being deprecated in the near term.

---

CTA Scenario Use Cases

Scenario 1: B2C Retailer - Omnichannel Welcome Journey

Situation: A national retailer with 5M loyalty members needs a welcome journey triggered when a customer registers in-store (POS) or online (Commerce Cloud). Journey must send email, push notification, and SMS over 14 days with personalized product recommendations.

Architecture Decision:

Component	Choice	Rationale
Platform	MC Engagement	B2C, high volume, multi-channel
Data sync	Data Cloud segment activation	Unifies POS + Commerce + CRM profiles
Journey entry	API Event via Data Cloud activation	Real-time entry on registration
Personalization	AMPscript + Einstein Recommendations	Dynamic product content per customer
Channels	Email Studio + Mobile Studio (push + SMS)	Multi-channel orchestration
Fallback	MC Connect for CRM writeback	Track engagement back to Service Cloud

Scenario 2: Financial Services - Regulatory Compliance with Marketing

Situation: A bank must send mandatory regulatory disclosures alongside marketing campaigns. Marketing team uses MC for campaigns, but compliance requires audit trails, opt-out enforcement, and content approval workflows.

Architecture Decision:

Component	Choice	Rationale
Transactional emails	REST API Transactional Messaging	Separate from marketing sends, guaranteed delivery
Marketing emails	MC Engagement Journey Builder	Campaign orchestration with suppression lists
Consent management	Salesforce CRM as master	Single source of truth for consent, synced via MC Connect
Audit trail	SOAP API tracking extract + Data Extract Activity	Compliance reporting from MC tracking data
Content approval	Content Builder + approval workflow	Locked content blocks for legal-approved text

Scenario 3: Healthcare - Patient Engagement Across Channels

Situation: A healthcare provider wants appointment reminders (SMS), wellness content (email), and post-visit surveys (CloudPages) - all personalized based on patient data in Health Cloud.

Architecture Decision:

Component	Choice	Rationale
Platform	MC Engagement	Multi-channel (email + SMS + web)
PHI handling	Data Cloud with Shield encryption	HIPAA-compliant data handling
Data sync	MC Connect with field-level filtering	Sync only non-PHI fields; use tokenized patient IDs
Appointment reminders	Triggered Send via CRM Flow	Real-time SMS on appointment creation
Surveys	CloudPages with SSJS	Custom forms writing back to Health Cloud via API
Consent	CRM consent objects	Health Cloud consent model as single source of truth

---

Common Gotchas

Data model mismatch

MC’s flat Data Extension model does not map cleanly to Salesforce’s relational object model. A CRM Contact with related Opportunities, Cases, and Activities becomes multiple flat DEs in MC. Plan data model mapping carefully. Denormalization is often required.

Contact billing surprises

Every record synced via MC Connect counts as a billable MC contact, even if you never send them a message. Sync filters are essential to avoid inflating the contact count and bill. Only sync records you intend to market to.

Sync latency assumptions

MC Connect “near-real-time” sync latency is near-real-time (exact latency varies; typically minutes, not seconds). Do not design journeys that depend on instant CRM data availability. For true real-time triggers, use API Events or Data Cloud real-time segments.

Subscriber Key migration is irreversible

Migrating your Subscriber Key strategy (e.g., from email address to CRM ID) is a one-time, irreversible operation that affects every contact and data extension in your MC account. Plan this at project inception, not after go-live.

---

Related Topics

• Integration Patterns: the six canonical patterns apply to MC integration touchpoints
• Event-Driven Architecture: Platform Events and CDC can trigger MC journeys through MC Connect
• Salesforce API Landscape: MC has its own API surface, distinct from Core Salesforce APIs
• Middleware Architecture: MuleSoft as an alternative to MC Connect for complex multi-system flows
• Error Handling Patterns: retry and idempotency for MC API integrations
• Data Modeling: contrast the relational CRM model with MC’s flat DE model

---

Sources

• Marketing Cloud Connect (Salesforce Help)
• Synchronized Data Sources (Salesforce Help)
• Synchronized Data Sources Best Practices (Salesforce Help)
• Contact Model Relationships & Data Model (Trailhead)
• Synchronized Data Source Creation (Trailhead)
• MC Connect Setup (Trailhead)
• Effective Data Model Design (Trailhead)
• API Integration Best Practices (Trailhead)
• Triggered Email Scenario Guide (Salesforce Developers)
• Pardot vs. Marketing Cloud (Salesforce Ben)
• How Salesforce Is Rebuilding Marketing Cloud for the Age of AI (Salesforce Ben)
• Salesforce Data 360: The 3 Pillars of Implementation (Salesforce Ben)