Marketing Cloud Next architecture
A reference architecture for Marketing Cloud Next on the Salesforce Platform and Data 360, including consent, agents, personalisation and coexistence with Engagement.
Marketing Cloud Next is Salesforce's agentic marketing product built on the core Salesforce Platform and Data 360. Because it shares the platform with Sales Cloud and Service Cloud, its architecture looks different from Marketing Cloud Engagement: customer data, consent and segments live in Data 360, and marketing work happens alongside CRM rather than in a separate account. This page sets out a reference architecture, labelling feature status exactly as Salesforce states it.
This is a reference design. We validate the edition, Data 360 provisioning, licences and permissions for each implementation, and GeekMidst can design and implement it through Marketing Cloud Next implementation.
Business context
New customers start with Marketing Cloud Growth Edition or Advanced Edition. Existing customers get Next capabilities by renewing into "+" versions: Engagement+, Account Engagement+, Personalization+ and Intelligence+. Salesforce says Marketing Cloud Next "doesn't replace Marketing Cloud Engagement—it complements and extends it", and existing campaigns keep running. Many estates will therefore run both for a time, and the architecture has to show where each one owns data and sends.
Marketing Cloud Next reference architecture
Reference architecture: a reusable pattern, not a live client integration
Select a component to see what it is responsible for and how data moves to and from it.
Components and data flows (text version)
| Component | Type | Responsibility |
|---|---|---|
| Salesforce CRMSales and Service Cloud | Salesforce product | System of record for accounts, contacts, leads and cases on the same platform as Marketing Cloud Next. |
| Web and mobileInteractions SDK | Channel or touchpoint | Captures behaviour through the Salesforce Interactions SDK, which sends events only after the visitor consents. |
| External datawarehouse, files, APIs | External or custom system | Orders, products and other data brought in through Data 360 connectors or zero copy federation. |
| Data streams and DMOsingest and map | Data and analytics | Ingests sources into data lake objects and maps them to data model objects for marketing use. |
| Unified profilesidentity resolution | Data and analytics | Match and reconciliation rules build unified individuals without altering source data. |
| Consent DMOscontact points, subscriptions | Identity, policy and governance | Communication Subscription Consent records opt-in or opt-out per contact point, subscription and channel; checked at send time. |
| Segments and audiencesbuilt on Data 360 | Data and analytics | Audiences built on unified profiles and published for campaigns. |
| Campaigns and flowson the Salesforce Platform | Engagement platform | Campaign execution and Flow-based automation, alongside CRM records and processes. |
| Marketing agentsstatuses vary | AI and agents | Campaign Agent and Marketing Goals Agent are pilot per the Winter '27 newsroom; Content Agent status not stated. |
| Salesforce Personalizationadd-on | Engagement platform | Data 360-native decisioning for web, mobile and email. An add-on for Growth and Advanced Edition customers. |
| EmailEinstein features per channel | Channel or touchpoint | Email sends with Einstein Send Time Optimization, Metrics Guard, Engagement Frequency and Engagement Scoring enabled in Setup. |
| SMS and WhatsAppmessaging credits | Channel or touchpoint | Messaging channels whose credit spend is tracked in the Digital Wallet. |
| Mobile app messagingAdvanced Edition | Channel or touchpoint | App messaging through the Mobile App Messaging module of the Unified Mobile Engagement SDK, for Marketing Cloud Advanced Edition. |
| Preference pagesconsent capture | Channel or touchpoint | Preference Pages and the Consent Status component let people change subscriptions. |
| Marketing Cloud Engagementif retained | Engagement platform | Keeps running existing journeys and high-volume sends while work moves to Next in stages. |
| Marketing analyticsIntelligence+ if licensed | Data and analytics | Reporting and attribution. Intelligence+ adds Paid Media Optimization, Segment Intelligence and Flexible Attribution. |
Data flows
- Salesforce CRM → Data streams and DMOs: CRM connector. Data 360 CRM connectors ingest selected objects. Refresh failures show in data stream status.
- Web and mobile → Data streams and DMOs: consented events. Events are sent only after consent. Standard web streams load engagement every 15 minutes; the real-time layer is faster.
- External data → Data streams and DMOs: connectors and zero copy. Files and APIs are ingested; warehouse tables can be federated without copying.
- Data streams and DMOs → Unified profiles: mapped DMOs. Identity resolution runs on mapped DMOs. Turning off automatic runs can reduce cost but delays updates.
- Preference pages → Consent DMOs: preference changes. Preference changes write consent records in Data 360. A failed write must block the next send, not be skipped.
- Unified profiles → Segments and audiences: unified individuals. Segments are built on unified individuals and published on a schedule or on demand.
- Consent DMOs → Campaigns and flows: send-time consent check. Consent DMOs are checked at send time, so an opt-out recorded before the send is honoured.
- Segments and audiences → Campaigns and flows: audiences. Published segments feed campaigns and flows.
- Marketing agents → Campaigns and flows: drafts for review. Agents propose briefs, content or audiences. A marketer reviews and approves before anything sends.
- Salesforce Personalization → Email: recommendations. For Advanced Edition email, the recommender is called server-side per recipient at send time.
- Campaigns and flows → Email: email sends. Email sends with the Einstein features enabled for that channel.
- Campaigns and flows → SMS and WhatsApp: messaging sends. SMS and WhatsApp sends draw down messaging credits.
- Campaigns and flows → Mobile app messaging: app messages. Messages to app users on Advanced Edition through the Mobile App Messaging module.
- Segments and audiences → Marketing Cloud Engagement: batch activation. Batch Personalizations and Data 360 activations can send decisions or segments to Engagement for outbound sends.
- Email → Marketing analytics: engagement data. Engagement results feed reporting and attribution.
- Email → Data streams and DMOs: engagement data. Engagement results return to Data 360 for segments and scoring. Which engagement data is stored there depends on edition and channel; confirm per implementation.
Component responsibilities
| Component | Responsible for | Status or licence note |
|---|---|---|
| Data 360 | Ingestion, mapping, identity resolution, segments | Must be provisioned |
| Consent DMOs | Opt-in and opt-out per contact point and channel | Checked at send time |
| Campaigns and Flow | Campaign execution and automation | Completion actions via Flow reported for Winter '27 (third-party summary) |
| Marketing agents | Campaign, goal and content assistance | Marketing Goals Agent and Campaign Agent: pilot; Content Agent: status not stated |
| Salesforce Personalization | Recommendations and decisions | Add-on for Growth and Advanced Edition |
| Einstein email features | Send Time Optimization, Metrics Guard, Engagement Frequency, Engagement Scoring | Enabled per channel in Setup |
| Digital Wallet | Messaging-credit spend for email, SMS and WhatsApp | Engagement+ and Next |
| Marketing Cloud Engagement | Existing journeys and high-volume sends | Coexists; no forced replacement |
Data flow, step by step
- Data lands in Data 360. CRM objects arrive through CRM connectors, web and app behaviour through the Interactions SDK, and other data through connectors or zero copy federation.
- Profiles are unified. Identity resolution builds unified individuals; see designing unified individual profiles.
- Consent is recorded. Preference Pages, CSV consent imports and Data Cloud-triggered flows write consent records. Salesforce's Trailhead warns against using the MessagingConsent actions for this.
- Audiences are built. Segments on unified individuals are published to campaigns.
- Campaigns run and agents assist. Marketers build campaigns on the platform. Agents in pilot can draft and suggest, and a person approves.
- Messages go out. Each send checks consent at send time. Personalised email can call the recommender per recipient.
- Results return. Engagement flows back into Data 360 and reporting.
What "real time" means here
Data 360's real-time layer documents that 95% of events complete ingestion to activation within about 500 milliseconds under the Sub-Second Real-Time Profile, from the Web SDK, Mobile SDK or server-to-server ingestion. Standard web and mobile streams load engagement data every 15 minutes and profile data every hour. Latency therefore depends on the path and features you license.
Trust boundaries and authentication
Marketing Cloud Next runs inside your Salesforce org, so access is governed by the same users, permission sets and sharing as CRM. The main boundaries are the org boundary, the public boundary (preference pages and SDK capture) and the Engagement boundary if you keep an Engagement account. Salesforce has published no MCP server specific to Marketing Cloud Next; AI access through MCP could use the Salesforce Hosted MCP Servers or Data 360 MCP Server, under their own licences.
Failure scenarios and how to handle them
| Failure | Effect | How to handle it |
|---|---|---|
| Consent write fails | Send to someone who opted out | Treat consent writes as blocking; alert and retry |
| Identity rules over-merge | Wrong person gets a message | Test match rules on sample data before go-live |
| Segment publishes late | Campaign uses stale audience | Monitor publish jobs; schedule sends after refresh |
| Agent drafts poor content | Off-brand or wrong message | Human review before any send |
| Engagement and Next both message a person | Duplicate or conflicting messages | One owner per programme during coexistence |
Operational monitoring
Monitor data stream refreshes, identity resolution runs and segment publish times in Data 360, and send results per channel. Watch messaging-credit spend in the Digital Wallet. Our article on monitoring Data Cloud jobs and refresh schedules explains what to watch.
Alternative approaches and trade-offs
Stay on Engagement if your programmes are high-volume and stable and you do not need platform-native data. Move programmes in stages to Next as Data 360 data and consent are ready, keeping Engagement for what it does well; see planning a move to Marketing Cloud Next. Pair Engagement with Data 360 activation if you want unified segments without changing the sending platform.
Implementation checklist
- 01
Confirm edition and licences
Growth or Advanced Edition, or the "+" renewal path, and Data 360 provisioning
- 02
Map sources
Choose CRM, web, app and external sources and their refresh modes
- 03
Design identity
Match and reconciliation rules, tested on sample data
- 04
Design consent
Contact points, subscriptions and channel consent records
- 05
Plan coexistence
Decide which programmes stay on Engagement and which move
- 06
Enable Einstein features
Per channel in Setup, after checking edition scope
- 07
Pilot agents carefully
Use pilot agents only with human approval of outputs
- 08
Monitor
Refresh jobs, segment publishes, sends and credit spend
Related services and solutions
Marketing Cloud Next implementation
Setup, data and campaigns on the platform
→Consent and preferences for Next
Consent design on Data 360
→Next vs Engagement
A decision guide
→Marketing Cloud Engagement architecture
The Engagement reference design
→Data 360 reference architecture
Ingestion, identity and activation
→Marketing Cloud Next architecture guide
A longer read on the same topic
→Questions we hear
Is this design from a client implementation?
Do we have to leave Marketing Cloud Engagement?
Are the marketing agents generally available?
Is there an MCP server for Marketing Cloud Next?
Sources
- https://salesforce.com/blog/marketing-cloud-next-details/
- https://trailhead.salesforce.com/content/learn/modules/marketing-cloud-next-for-marketing-cloud-engagement-foundations/get-to-know-marketing-cloud-next-with-marketing-cloud-engagement
- https://trailhead.salesforce.com/content/learn/modules/consent-management-fundamentals-for-marketing-cloud-next/explore-consent-tools-and-concepts
- https://trailhead.salesforce.com/content/learn/modules/ai-in-marketing-cloud-next/boost-your-marketing-campaigns-with-einstein
- https://developer.salesforce.com/docs/data/data-cloud-ref/guide/c360a-real-time-ingestion-api.htm
- https://www.salesforce.com/news/stories/winter-2027-product-release-announcement/
Plan your Marketing Cloud Next architecture
Tell us about your current marketing stack and Data 360 plans. We can map a staged architecture and coexistence plan for your edition.
