Marketing Cloud Next - Consent Management: Consent at Scale and Data Ingestion

Note: This article is a companion article for the Consent Management: Frequently Asked Questions resource. Check that article out for additional details and other frequently asked questions.

How do I handle consent at scale (millions of records) for initial onboarding?

The Marketing Cloud Next CSV import tool is limited to 50,000 rows per file, so it is not suited to enterprise-scale loads. The recommended approach for large initial loads is the following 2 step process:

1. Complete the initial data load into Data Cloud. A Batch Data Transform can be used to create records in the consent DLO from contact or lead records.

2. Then use a Data Cloud-Triggered Flow with the Create Consent action to write the consent so it's honored at send time.

Do not use the MessagingConsent or MessagingConsentV2 actions — they are not supported. 

My primary data is in Snowflake (or Databricks, Azure, another non-Salesforce system). What's the supported path?

You can drive consent from a non-Salesforce primary data source. The send-time consent check doesn't require a CRM Lead or Contact — what's required is an Individual, a Contact Point value, a Communication Subscription Consent record, and a PartyId, all produced upstream. What does not change is that consent must be written through a supported method.

Supported end-to-end pattern at scale

1. Land the source data in Data Cloud. Bring person records and consent rows in via a Data Stream or the Data Cloud Bulk Ingestion API. (This populates Data Cloud but does not yet write consent in Marketing Cloud Next.)

2. Stage the consent rows for a Data Cloud Record-Triggered Flow. Use a Batch Data Transform to project consent rows from your source into a target that fires the flow.

3. Run the Data Cloud Record-Triggered Flow with the Create Consent action. The flow iterates the staged rows and calls Create Consent for each — the supported write path.

Things to avoid (see companion knowledge article on supported actions)

• Don't map the source system directly to the Consent DLO via a Data Stream.

• Don't use the Data Cloud Bulk Ingestion API to write directly into the Consent DMO.

• Don't use the MessagingConsent / MessagingConsentV2 Flow actions.

• Don't rely on the CSV Consent Import tool for sustained high-volume sync (it's capped at 50,000 rows per file and intended for one-time loads).

Alternate write path: For programmatic ingestion driven by an external system rather than a Data Cloud event, an On-Demand Flow can be invoked via REST and supports the Create Consent/Consent Request action.

How do bulk consent imports handle bad rows, and why do imports sometimes stay "In Progress"?

For loads of 50,000 rows or fewer, use the Consent tab import wizard, which updates consent correctly and reflects mailable status within about two minutes. The wizard supports CSV files only and is practically capped by spreadsheet limits, so it doesn't scale to very large customers. 

Imports with validation errors (for example, "contactPointValue provided value cannot be formatted") can stay "In Progress" indefinitely, and there's no option to skip bad rows — a clean file is required.

For enterprise-scale initial loads, use the Data Stream + Batch Data Transform + Data Cloud-Triggered Flow (Create Consent) pattern described above. 

Important: don't mix a direct-to-DLO bulk load with out-of-the-box tools (Preference Page, Create Consent action, one-click unsubscribe, LWC) for the same records, as this can cause consent values to become inconsistent.