Unify Logo Footer.svg
Unify Data
Logo
Golden Records & Audit

Golden Records & Audit

Logo

3 mins read READ

Records is where you browse, search, and inspect the trusted, unified version of every entity in your model — the golden records that have already passed data quality checks and match resolution. Each record opens into three tabs so you can see what it contains, how it connects to other records, and everything that has changed on it over time.

Overview

Every model has its own Records screen, and you pick the entity you want to browse before searching. The screen lets you find unified records using any known information — a name, an ID, an email, or any other field the entity carries. Each row in the results can carry small pills next to it: a potential duplicate count, and a data quality insight count. Opening a record takes you into its detail view, where the Attributes, Relationships, and Audit tabs sit across the top.

Searching Golden Records

To find a record, go to Ontology › [Model] › Golden Records › Records, choose the entity you want from the entity picker, and search by any field you already know — you don't need to know the exact record ID to find what you're looking for. As you scan the results list, watch for pills on each row: they surface signals about that record without your having to open it first.

Opening a record also gives you access to its Data Quality Insights drawer, which lists the insights that were applied by data quality rules during record resolution. This is the quickest way to understand how the record in front of you was assembled from its underlying sources.

Record Tabs — Attributes, Relationships, Audit

Every golden record opens into the same three tabs:

  • Attributes — the record's field values, exactly as they stand today after data quality rules and survivorship have resolved them.

  • Relationships — the other records connected to this one through the model's defined relationships, plus a full relationship-details view for digging into a specific connection.

  • Audit — the record's history, so you can see what changed and when.

Record-Level Relationships

The Relationships tab shows every other golden record connected to the one you're viewing, based on the relationships defined for the model. Each connected record appears as a node; clicking into a connection opens a full relationship-details view with more context on that specific link.

Because each node has to look up its connected record's display name separately, you may briefly see a loading placeholder before the name settles in. If a connected record has no resolvable name, its node shows the record ID instead, so you always have something to identify it by. You should only see "Name not found" if the underlying reference itself is broken — this is not expected in normal use, so treat it as a signal something needs attention.

Audit Trail

The Audit tab is the record's history log — open it whenever you need to see what changed on a record and confirm it reflects the latest data. This is also where updates from merges show up: if an incoming record was merged into an existing golden record rather than creating a new one, you won't see a new row in Records, but the existing record's Audit tab will show the change.

Duplicate Indicators

A golden record can display a potential duplicate pill in the Records list, hinting that a match rule has flagged another incoming record as a likely duplicate of this one. The pill itself doesn't let you resolve anything from here — it's a pointer. To review and act on the flagged candidate, head to the Quarantined Records › Potential Matches screen, covered in the separate Quarantine Records page.

Notes

  • Records only ever reflect data that has already passed data quality checks and match resolution — anything still under review lives in the quarantine queues, not here.

  • A merged record never produces a new row in Records; check the existing record's Audit tab to see the merge reflected as a change.

  • "Name not found" on a Relationships node is unusual — if you see it, the underlying reference is likely broken rather than just slow to load.

  • Data quality insight counts and potential duplicate counts are separate signals; a record can carry either, both, or neither.