Data
Data and freshness
Understand file snapshots, connected sources, datasets, and last successful updates.
Choose the right source
Use a file upload for a one-time snapshot. Use a configured connection when the source must be read again. Both produce workspace resources; neither makes an external system authoritative merely because it was added.
- Before uploading, remove columns you do not need and confirm that the workspace may store the data.
- For a connection, check its owner, scope, credentials, and refresh policy before relying on scheduled updates.
- A dataset's most recent successful update is more useful than a page's current browser time. Treat missing freshness information as unknown.
Inspect a dataset
- Open Datasets in the workspace and select the dataset.
- Check schema, row count, update state, and available history before building on it.
- If a downstream result looks wrong, trace its input dataset and last successful run before changing a rule or action.
Replacement and recovery
Replacing a snapshot is a new data event; it does not retroactively make an earlier review current. Recheck results and evidence after replacement. A failed refresh should keep the last successful result distinguishable from the failed attempt.