The pattern: download monthly, identify gaps and drift, follow up with field owners, upload corrections. The cycle is the discipline that keeps Data Classification meaningful; without it, classifications go stale fast.
- Open Setup, Data Classification Settings
The page hosts both Download and Upload utilities for Data Classification metadata.
- Click Download to generate the CSV
Choose Include All (every field) or Classified Only (fields with at least one classification dimension populated). Include All is the right pick for compliance audit.
- Open the CSV in your analysis tool
Excel, Google Sheets, or a data tool. Filter, pivot, and analyze. Pivot by Sensitivity Level to see distribution; pivot by object to see per-object coverage.
- Identify gaps: fields without classification
Filter for rows where Sensitivity Level is empty or Compliance Categorization is empty. The list is the compliance-debt backlog.
- Assign default classifications for the gaps
Use a default (Internal sensitivity, no compliance categorization) or coordinate with field owners for accurate classification. Update the CSV in place.
- Upload the updated CSV via Data Classification Upload
The Upload utility ingests the corrected CSV. Classifications update on the matching fields without code deployment.
- Schedule the next monthly download
Diff against this month's download to identify drift. New fields added without classification are the most common drift signal.
Whether the CSV includes every field or only fields with at least one classification dimension populated.
Sensitivity Level, Compliance Categorization, Data Owner, Description. Each captures a different aspect.
Monthly or quarterly download for compliance-heavy orgs. Annual is too infrequent for actively-evolving schemas.
Excel, Sheets, or external data catalog. The tool consumes the CSV.
Data Classification Upload utility ingests the corrected CSV without code deployment.
- Download is not automated by default. Manual cadence is the standard pattern; automation requires Metadata API scripting.
- Include All produces large CSVs for big orgs (thousands of rows). Filter and pivot tools are required for analysis.
- Standard field defaults vary by release. Salesforce ships PII defaults on common fields; the defaults may not match the org's specific policy.
- Field owners are required for accurate classification. Defaults work as placeholders; accurate classification requires the owning team's input.
- Classifications go stale as fields are added or repurposed. The monthly cadence is the only reliable freshness mechanism.