Unify Logo Footer.svg
Unify Data
Logo
Sources & Data Sync

Sources & Data Sync

Logo

4 mins read READ

Sources connect a model to the systems that hold your data, and Data Sync is where records from those sources actually flow into the model and become golden records.

Overview

Every model needs at least one source before it can hold real data. A source is a live link to an application — a CRM, a database, a spreadsheet system — through one of your existing connections. Once a source is attached, Data Sync governs how records move from that source into the model: through data quality checks, matching, and survivorship, until they emerge as trusted golden records. Sources and Data Sync live as two tabs on the model, and you'll typically set up the former before the latter.

Connecting a Source

Open the model's Sources tab and select New Source. You'll move through two steps: first Applications, where you search for and pick the app you want to pull data from (for example Salesforce or a database); then Source, where you either choose one of your existing connections to that app or add a new one. Each source links out to View in Connections Manager if you need to inspect or change the underlying connection.

Removing a source asks for confirmation first, and the action can't be undone, so double-check before you proceed. Adding a source requires Edit access to the model — if you don't have it, the New Source button is hidden entirely, both in the tab header and in the empty state, rather than appearing and then failing when you click it.

Note: if the Sources tab shows an access-denied message instead of the usual "add your first source" prompt, it means sources already exist on the model but you don't have permission to view the connections they're built on. This is intentional — it stops you from being told to add "the first source" when sources are already there and simply hidden from you.

Setting Up Data Sync

Open the Data Sync tab to define and manage how data moves from source to model. If the model isn't ready yet, you'll see a Data Sync Setup checklist instead of the sync screen itself, walking you through what's still missing:

  • Create an Entity — define the entity the incoming records will populate.

  • Define Entity Fields — set up the fields that records will be mapped into.

  • Configure Quality & Matching Rules (optional) — set the rules that will validate and de-duplicate records as they sync.

Each item links straight to the screen where you complete it and checks itself off once done. When every required step is complete, Start Sync becomes available and opens the sync-method chooser.

Pipelines vs. Automations

Selecting Start Sync opens New Sync, which asks how you want to fetch data. You have two methods to choose from:

MethodHow it worksWhere it's managed
PipelinesFetches data from the source into the model through a managed data pipeline — best for scheduled, bulk movement of records.Pipelines tab
AutomationsUses event-driven workflows to write records into the model as things happen in the source — best for near-real-time updates.Automations tab

Each tab lists everything you've built with that method, and you can delete any item from the list, though you'll be asked to confirm first since the action can't be undone.

The Record Pipeline

Records don't jump straight from a source into a finished golden record. Once a sync brings a record in, it lands in a staging layer and works its way through the entity's rules in order. First it's checked against your data quality rules — a record that fails can be rejected and routed into Rejected Records rather than moving forward. Next it goes through match rules, which look for duplicates of the same real-world record; matches are either merged automatically or flagged into Potential Matches for you to review, depending on your merge policy. Finally, survivorship decides which value wins for each field when a record has multiple sources of truth.

A record that clears all three stages becomes — or updates — a golden record, visible on the model's Records screen. In short, every synced record ends up in one of four places: a golden record is created or updated, the record is flagged for review, the record is rejected, or nothing changes because the incoming data matched what was already there.

Notes

  • You need Edit access to the model to add a source; without it, the option to add one is hidden rather than disabled.

  • Removing a source or deleting a pipeline/automation is permanent — both actions require confirmation and can't be reversed.

  • An access-denied message on the Sources tab doesn't always mean the model has no sources — it can mean sources exist but you can't see the connections behind them.

  • Data Sync stays gated behind the setup checklist until an entity, its fields, and (optionally) quality and matching rules are in place.

  • Quality and matching rules are optional to start a sync, but skipping them means records skip straight from staging to golden records(without any deduplication or validation.