The wiring arrives before the lights
A Marketing Cloud Next org with a licence and no data kits is a marketing platform with nothing to segment on. Consent objects are missing. Email engagement events are missing. Flow telemetry is missing. The features exist in the menu and the tables behind them are empty.
The remedy sits in Setup: Quick Find “Basic”, open Basic Settings, choose Install Marketing Data Kits, then Install or Update. It takes a few minutes. The first run deploys everything required; later runs reinstall only the kits that changed. If an ingestion stream fails to deploy on its own, you deploy it by hand from the Data Streams tab.
That button is the least glamorous and most consequential difference from Braze in this series. It is also the one most teams underestimate, because installing schema feels like configuration and turns out to be a commitment.
Seven kits and the objects they create
A data kit is a packaged bundle of DMOs, fields, relationships and data streams. You are installing structure, and the structure then dictates what the platform can do.
| Setup label | Package | What it wires up |
|---|---|---|
| Marketing Setup Objects Data Kit | MarketingSetup_General plus SMSAddOn_General | Base marketing objects |
| Consent Objects Data Kit | UnifiedMessagingConsent | ConsentAuditTrail and MessagingConsent streams |
| Flows Integration Data Kit | Salesforce Data Cloud Flow Integration | Flow data bundle plus two Flow Run ingestion streams |
| Email Channel Data Kit | MessagingEventsEmailEngagement | MessagingEventsEmail |
| SMS Channel Data Kit | MessagingEventsSms | MessagingEventsSMS, add-on only |
| WhatsApp Channel Data Kit | UnifiedWhatsAppPackage | MessagingEventsWhatsApp, add-on only |
| Sales Data Kit | Deployed separately from Data Cloud Setup | Accounts, leads, contacts, prospects and their mappings |
The consent kit is the one that surprises people. It ships an audit trail object your compliance team will later ask for by name, and until the kit is installed there is no such object to query.
The Sales Data Kit is optional on paper and load-bearing in practice. Skip it and reporting dashboards need additional configuration. Skip it and Opportunity Influence cannot work at all, with no error message explaining why.
What a data stream actually does
A data stream is a connector plus a mapping definition. It moves records from a source system into a data lake object, where they land exactly as the source sent them.
The interesting part is the ingestion mechanism inside each stream, because each one carries a different latency budget. The CRM connector uses Change Data Capture, which pushes record changes in roughly three minutes, and falls back to a batch cycle of around ten minutes for objects and formula fields that capture does not support. The Ingestion API runs scheduled batch cycles that can lag by several hours. Cloud storage and packaged connectors follow their own schedules, which you inherit rather than choose.
The first full extraction can take up to 24 hours. Nothing in the platform tells a marketer that their audience is thin because the initial load is still running.
Braze ships pipes and no package
Braze has no kit concept at all. It ships four independent ingestion paths, each configured per workspace, and none of them installs schema.
| Braze pipe | What it does | Nearest Marketing Cloud Next analogue |
|---|---|---|
| SDK, client side | Real-time writes from web, iOS, Android and other platforms through changeUser, logCustomEvent and setCustomAttribute | Web tracking plus mobile engagement streams, though Braze is SDK-first by design |
REST /users/track | Server-side batched writes of attributes, events and purchases, up to 75 objects per request in the standard shape | Ingestion API |
| Cloud Data Ingestion | Braze reads changed rows from your warehouse tables on a schedule | A cloud storage stream, inverted in direction |
| Connected Sources | Live query of Snowflake, BigQuery, Redshift or Databricks at segment or send time | Zero copy |
The closest thing to a kit is the Recommended Events catalogue for eCommerce. It gives you a standardised event taxonomy (product viewed, cart updated, checkout started, order placed), prebuilt Canvas templates for abandoned browse, abandoned cart, abandoned checkout and order confirmation, and calculated profile fields for total revenue and total orders. That is the shape of a data kit delivered as a convention.
The difference matters at audit time. A data kit has a version, an owner and an upgrade path. A convention has a wiki page and a Slack channel.
Two directions, two failure modes
Marketing Cloud Next pushes from the source. The CRM tells the platform that a record changed. Braze’s Cloud Data Ingestion pulls from the warehouse, which means Braze owns the scheduler and your warehouse owns the bill.
Direction determines who notices when something breaks. A push integration fails at the source, and the source team sees it. A pull integration fails quietly in Braze’s sync log, and the marketer sees an audience that stopped growing.
Cloud Data Ingestion carries two constraints worth memorising. Only one sync runs per integration at a time, so an on-demand trigger fired during a scheduled run waits its turn. And on-demand syncs do not replace the recurring schedule; you can run both, and they behave as separate jobs.
| Stage | Marketing Cloud Next | Braze |
|---|---|---|
| Fastest ingest path | CDC at roughly 3 minutes | SDK or REST, effectively instant |
| Batch ingest | Ingestion API, hours | CDI from 15 minutes by default, 5 minutes with support approval, up to monthly |
| First full load | Up to 24 hours | Depends on table size |
| External read at send time | Zero copy with a scheduled refresh | Connected Content over HTTP, with caching and retries |
Braze’s default 15-minute CDI interval exists to protect your warehouse bill. Every sync is a query your warehouse charges for, which is also why the 5-minute floor is a support exception.
The marketer's Tuesday changes shape
In Marketing Cloud Next, the admin installs kits in Setup and verifies each one’s status. They optionally deploy the Sales Data Kit. They map any fields the kit did not map. They confirm that the email, SMS and WhatsApp streams are receiving events, then move on to identity resolution and the data graph. Step one is a button, and step four usually needs a data team.
In Braze, an engineer installs the SDK per app instance with its own API key, then implements the event taxonomy. A marketer defines custom attributes and events in settings, or lets /users/track create them implicitly. A data engineer configures Cloud Data Ingestion for warehouse-resident data. Verification happens in the Event User Log and the sync logs.
The asymmetry is worth naming precisely. Braze’s step one is a sprint of SDK work, and its later steps need no ticket. Marketing Cloud Next front-loads the button and defers the ticket.
What the convenience costs
Rigidity is the first cost. Installing Marketing Cloud Consent Objects commits you to Salesforce’s consent schema, and there is no uninstall followed by a fresh attempt without migrating data first.
Schema debt is the second, and it lands differently on each side. Braze lets a typo in a payload create a permanent custom attribute, and blocklisting and deleting it is manual cleanup. Marketing Cloud Next prevents the typo by giving you a fixed schema, then charges you for the fields you cannot remove from a data graph.
There is a third cost specific to Braze that no kit will solve: no packaged consent stream exists. The equivalent of UnifiedMessagingConsent is something your team builds and keeps authoritative.
A quick check for understanding
Your CRM team adds a field on Thursday. On which platform can a marketer personalise with it on Friday morning, and what has to happen in between?
Part four follows one record through the three layers that Marketing Cloud Next maintains and the single layer Braze keeps.
