Track Changes to Transformation Datasets with Version History

Transformation-type Datasets now include a Version History on the Dataset Details page. Every time someone saves a change to a transformation Dataset, a new version is recorded with the date of the edit and the user who made it — giving your team a clear, auditable record of how a Dataset has evolved over time.

For SQL transformations, each version also includes tools to review the exact query at that point in time and see what changed relative to the previous version.

Where to find Version History

1. Open Data Cloud and navigate to the Dataset you want to inspect.
2. Open the Dataset Details page for that Dataset.
3. Scroll to the Version History section.

Each row in Version History represents one saved version of the Dataset, ordered from most recent to oldest.

What's in each version

View the SQL for a past version

For SQL transformation Datasets, you can open the exact query that was in effect for any version:

1. In Version History, find the version you want to review.
2. Click View SQL on that row.
3. A read-only modal opens with the full SQL query.
4. Use the copy-to-clipboard button to copy the query — handy for pasting into a scratch workspace, sharing with a teammate, or adapting into a new transformation.

The viewer is read-only. Opening a past version's SQL does not change the Dataset or roll it back.

Compare SQL between versions with Diff

When you open the SQL viewer for a version, you'll also see a Diff tab. The Diff tab highlights line-by-line changes between the selected version and the version that came immediately before it.

• Added lines are highlighted to show what was introduced in this version.
• Removed lines are highlighted to show what was taken out from the previous version.
• Unchanged lines are shown for context so you can read the change in place.

Use Diff to quickly answer questions like:

• "What changed in this Dataset between last week and today?"
• "Which join or filter was added in the most recent edit?"
• "Did this refactor actually change the logic, or just the formatting?"

Publish status and republishing

Every transformation dataset shows its current publish state in a Published Status panel on the Dataset Details page.

Status values

Detail message

When a dataset is in an Error state — or when the platform returns any other status message — a Detail: line appears beneath the status indicator showing the specific error or status message. Use this to identify why a build did not complete without needing to dig through logs elsewhere.

Republish latest

The Published Status panel includes a Republish latest button that rebuilds and republishes the most recent saved version of the dataset in one click.

1. On the Dataset Details page, locate the Published Status panel.
2. Click Republish latest.
3. The button label changes to Republishing… while the operation is running.
4. A confirmation message appears: "Republish submitted. Processing may take a few minutes."

Rebuild Dataset

Transformation-backed datasets now have a Rebuild Dataset action in the More Actions menu on Dataset Details. Triggering it republishes the dataset and rebuilds downstream views — no more opening the SQL editor and saving a no-op edit just to force a rebuild.

How to trigger a rebuild

1. Open Data Cloud and navigate to the transformation dataset you want to rebuild.
2. Open the Dataset Details page.
3. Click the More Actions menu (⋯) in the top-right area of the page.
4. Select Rebuild Dataset.
5. A confirmation toast will appear to let you know the request was submitted. Processing may take a few minutes depending on the size of the dataset and its downstream dependencies.

What a rebuild does

• Republishes the transformation dataset using its current saved SQL.
• Rebuilds all downstream views that depend on this dataset.
• Does not modify the SQL or field mappings — it only re-executes what is already saved.

Recent Publish Events

Dataset Details for transformation datasets now shows a Recent Publish Events table, displayed under Edit History. This table gives you a live view of rebuild activity for the dataset.

What's in the table

• The table polls for live updates — new events appear automatically as they come in, without requiring a page refresh.
• Only transformation-type datasets show this table. Ingested and native-provider datasets do not have a Recent Publish Events section.

Using Recent Publish Events to troubleshoot

If a downstream report shows unexpected data after a rebuild, check Recent Publish Events to:

1. Confirm that a rebuild actually completed successfully (vs. silently failing).
2. Identify the timestamp of the last successful publish — useful for correlating data freshness.
3. Review error details on any failed events to determine the root cause.

Common use cases

• Audit changes. See who edited a transformation Dataset and when, without digging through Slack or change-management tickets.
• Troubleshoot downstream reports. If a report's numbers shifted, use Version History to identify when the upstream transformation changed and what changed in the SQL.
• Force a rebuild. Use Rebuild Dataset from More Actions to republish the dataset and its downstream views without touching the SQL.
• Monitor publish health. Check the Published Status panel for the current pipeline state, and Recent Publish Events to confirm rebuilds are succeeding and to diagnose failures.
• Onboard teammates. Walk new team members through how a Dataset has evolved, with the full SQL history as reference.
• Review before rolling forward. Compare the current SQL against prior versions before making your next edit.

Related articles

• Transform Data with SQL Transform Builder
• Monitor Materialization Refresh Status
• Dataset Exploration, Organization, & Transformation
• Managing Your Datasets in Data Cloud
• Dataset Lineage