The Data Landscape: Connect, Sync, and Where Datasets Fit

Before you build anything, you need a mental map of how data moves through CRM Analytics. Get this picture right and every tool in the platform suddenly has an obvious place. Get it wrong and you'll spend hours wondering why a dataset is stale or a sync failed.

The staging layer: Connect and Sync

Most of your data starts in your local Salesforce org, and the platform has a native connection to it. You bring objects in through Connect (also called Sync) exactly as they are — Account, Opportunity, User — each landing as its own synced object.

Think of this as a free staging area. You're not building finished datasets here; you're pulling raw objects into a holding zone. Later, a dataflow or recipe reads from this staging layer to assemble the datasets you'll actually use in dashboards, lenses, and stories.

Where data can come from

Salesforce is the common source, but it's far from the only one. The Connect/Sync layer can ingest from:

• Your local Salesforce org — native, the most common path.
• Other Salesforce clouds or connected orgs — pull data across environments.
• External sources like Snowflake — including a live connector that reads at query time.
• CSV files — a simple manual upload, which we'll do in the next lesson.
• Your own ETL tool or API — any ETL that has a connector to CRM Analytics can push data in, or you can call the ingestion API directly (documented under the Help link).

Data can flow out, too

In recent releases the platform has been extending its output connectors. Once you've built a dataset — brought it from Salesforce, combined it, enriched it — you can push it back out to Snowflake, Amazon S3, or a Tableau hyper file to build dashboards there. Analytics isn't a one-way street anymore.

One feature to know about: Trend Report Data

If you have the CRM Analytics Plus license plus the Trend Report Data in Analytics permission, every operational report gets a small "Trend in Analytics" button. Click it and you can schedule that report — daily, weekly, or monthly — to store its results directly in a dataset.

It's handy if you're a heavy Salesforce report user who just wants historical snapshots of counts, activities, or cases. Be aware, though, that it's limited: it goes straight to a dataset with little control over the transformation. In practice you'll more often snapshot data through an app or a recipe.

The engines: dataflows and recipes

Between the staging layer and your finished datasets sit the two engines that execute your instructions:

The key insight that trips people up: these run in an order. Sync refreshes the base objects first, then a dataflow or recipe runs to update the finished datasets. If your dashboard shows stale numbers, the culprit is almost always something out of order in that chain — a sync that hasn't run, or a recipe scheduled before its source finished.

Throughout this section you'll hear me distinguish baseline datasets (the raw building blocks) from final resulting datasets (what dashboards consume). Hold onto that distinction — it's the difference between a maintainable analytics app and a tangle nobody can debug. Next, the simplest way to make a dataset: a CSV upload.