Your customer data lives in Sales Cloud. Your campaigns live in Salesforce Marketing Cloud. Marketing Cloud Connect is the bridge between them — and it is usually where integration projects quietly go wrong: too many objects synced, queries that blow past their automation window, send results that never make it back to the CRM. This guide walks through the decisions that actually matter: which objects to sync, how to query Synchronized Data Extensions in SQL, how to personalize safely in AMPscript, and how to diagnose a sync that has stalled. Code included.
What Marketing Cloud Connect really does
Marketing Cloud Connect (MCC) has two halves: a managed package installed in Sales Cloud or Service Cloud, and a connected user on the Marketing Cloud side. Once the link is live, you get two capabilities that teams routinely conflate.
The first is sending from the CRM: a rep triggers an email straight from a Contact record, or a Journey Builder journey starts when a record changes stage in Sales Cloud.
The second is data synchronization: CRM records are copied into Marketing Cloud as Synchronized Data Extensions (SDEs). These tables carry the _Salesforce suffix and are read-only. They are not sendable as they stand — and that is the first trap.
Three limits are worth agreeing on before anyone opens Contact Builder:
- The sync is not real time. It polls on a schedule, with a fifteen-minute minimum interval.
- Records deleted in the CRM do not always disappear on the Marketing Cloud side if the recycle bin is not accounted for.
- MCC is not a substitute for Data Cloud. It syncs Salesforce objects, not external sources.
Choosing what to sync: less is more
The hidden cost of wide objects
Every synced object consumes storage and processing time. Syncing the entire Contact object with its two hundred custom fields because someone said we might need it later is the decision that costs you most six months in: slower queries, automation windows that overrun, and a storage line item with nothing to show for it.
The three-question test
Before you add a field to the sync, ask three questions. Does this field segment? Does it personalize a message? Does it trigger a journey? If the answer is no three times over, it does not belong in Marketing Cloud.
In practice, a sensible starting footprint looks like this.
| Object | Primary use | Typical fields |
|---|---|---|
| Contact | Subscriber key and personalization | Id, Email, FirstName, LastName, HasOptedOutOfEmail |
| Lead | Pre-conversion nurture | Id, Email, Company, LeadSource, Status |
| Account | B2B segmentation | Id, Name, Industry, BillingCountry |
| Opportunity | Sales-driven triggers | Id, AccountId, StageName, CloseDate, IsClosed |
| Campaign Member | Linking sends to CRM campaigns | Id, CampaignId, ContactId, Status |
Querying Synchronized Data Extensions in SQL
SDEs behave like any other Data Extension in Automation Studio, with one wrinkle: the table name must be wrapped in square brackets and carries the _Salesforce suffix. Here is a sanity-check query to confirm your sync is delivering records you can actually use.
SELECT
c.Id AS ContactId,
c.Email,
c.FirstName,
c.LastName,
a.Name AS AccountName,
a.Industry
FROM [Contact_Salesforce] c
INNER JOIN [Account_Salesforce] a
ON a.Id = c.AccountId
WHERE c.Email IS NOT NULL
AND c.HasOptedOutOfEmail = 0
AND c.IsDeleted = 0
Three habits are worth forming on your very first query. Always filter on IsDeleted = 0, because deleted records linger in the SDE for the retention window. Avoid joining on text fields — join on the 18-character Salesforce Id, which is indexed and stable. And never run SELECT * against a synced object; you pay for every unnecessary column in execution time.
Building a clean sendable Data Extension
An SDE cannot be sent to. So you materialize your audience into a local Data Extension, flagged as sendable, populated by a Query Activity. That query is also the right place to apply consent rules and suppression lists, rather than repeating them inside every journey.
SELECT
c.Id AS SubscriberKey,
c.Email AS EmailAddress,
c.FirstName,
c.LastName,
a.Name AS AccountName,
a.Industry,
o.StageName,
CONVERT(DATE, o.CloseDate) AS CloseDate
FROM [Contact_Salesforce] c
INNER JOIN [Account_Salesforce] a
ON a.Id = c.AccountId
LEFT JOIN [Opportunity_Salesforce] o
ON o.AccountId = a.Id
AND o.IsClosed = 0
WHERE c.Email IS NOT NULL
AND c.HasOptedOutOfEmail = 0
AND c.IsDeleted = 0
AND c.Id NOT IN (
SELECT SubscriberKey FROM [Global_Suppression]
)
One design decision deserves more attention than it usually gets: the Subscriber Key. Use the 18-character Salesforce Id, never the email address. An address changes, gets shared between two people, or reappears on a duplicate record; the Id stays stable for the life of the contact. Migrating Subscriber Keys after the fact on a live tenant is one of the most painful projects you can sign yourself up for.
Personalizing in AMPscript without breaking the send
CRM data is rarely tidy. All-caps first names, accounts with no name, contacts too new to have an opportunity — your code has to absorb all of it without shipping a stray Hi , to production.
%%[
VAR @firstName, @accountName, @stage, @greeting
SET @firstName = AttributeValue("FirstName")
SET @accountName = AttributeValue("AccountName")
SET @stage = AttributeValue("StageName")
IF EMPTY(@firstName) THEN
SET @greeting = "Hi there"
ELSE
SET @greeting = Concat("Hi ", ProperCase(@firstName))
ENDIF
IF EMPTY(@accountName) THEN
SET @accountName = "your team"
ENDIF
]%%
<p>%%=v(@greeting)=%%,</p>
%%[ IF @stage == "Negotiation/Review" THEN ]%%
<p>Your project at %%=v(@accountName)=%% is entering its final stage.</p>
%%[ ELSEIF @stage == "Prospecting" THEN ]%%
<p>A few resources to help you scope things out at %%=v(@accountName)=%%.</p>
%%[ ELSE ]%%
<p>Here is what we picked out for %%=v(@accountName)=%% this month.</p>
%%[ ENDIF ]%%
Note the consistent use of AttributeValue() instead of the %%Field%% shorthand. It lets you test a value before rendering it, and stops a single missing field from failing the entire send.
Pushing engagement back into the CRM
This is the part teams discover too late. Marketing Cloud Connect writes send results into the Individual Email Result object in Sales Cloud: opens, clicks, bounces, unsubscribes. Your sales team then sees marketing activity right on the contact record, without leaving the tool they live in.
Marketing Cloud Connect only reports engagement back to the CRM for sends that go through it. A standard send against a local Data Extension is completely invisible in Sales Cloud — and that is the number one reason attribution reports do not add up.
In practice, for a send to be tracked in the CRM it must either be triggered from the CRM, or run through a journey whose entry source is a Salesforce object. If your reporting strategy depends on that write-back, settle it while you are designing the journeys — not after the first steering committee.
Diagnosing a stalled sync
The sync has quietly paused
A sync can pause without announcing itself, usually after the integration user's password expires or their permissions change. Check the Synchronized Data Sources screen in Contact Builder first, then verify the credentials on the Sales Cloud side.
Missing or empty fields
A field added in Salesforce does not appear in the SDE automatically. You have to select it explicitly, then run a full resync. Likewise, a field the integration user cannot read will arrive empty with no error at all — always check field-level security before blaming the platform.
Volumes getting out of hand
If your automation windows are overrunning, shrink the synced footprint first, then add filters at the source rather than in the query. A smaller SDE speeds up every downstream query at once.
Key takeaways
Two halves, not one. Keep CRM-initiated sending and data synchronization separate in your design; they carry different constraints and different failure modes.
Sync only what earns its place. A field that does not segment, personalize, or trigger has no business in Marketing Cloud.
Never send from an SDE. Materialize audiences into a local sendable Data Extension through a Query Activity that applies your suppressions.
Subscriber Key is an Id, not an email. Use the 18-character Salesforce Id from day one and save yourself a migration.
CRM tracking is a design decision. Only sends routed through Marketing Cloud Connect populate the Individual Email Result object.
Planning a Sales Cloud and Salesforce Marketing Cloud integration, or trying to rescue a sync that has drifted? Let's talk about your project.
