Inspect an example that changed
Keep the original file and compare one affected row with its imported result. A value displayed as a date may be stored as a spreadsheet serial, text or a timestamp. Do not repair a whole column until you know which representation the export actually contains.
| Input | Question | Preparation |
|---|---|---|
| 03/04/2026 | Does it mean 3 April or 4 March? | Confirm the intended value and the target’s accepted format |
| 2026-04-03 | Is the target a date-only field or a timestamp? | Preserve date-only meaning where appropriate |
| A numeric spreadsheet value | Which date system created the number? | Export and inspect a known reference date |
Use the documented target rules
PostgreSQL’s COPY documentation recommends an ISO DateStyle for portable output and explains that input interpretation can depend on DateStyle. That does not make every external import accept the same format: inspect the actual target’s field type and import documentation. Test ambiguous dates and date-only values separately from timestamps.
Run a small test import, compare counts and values, then verify a repeat does not duplicate records. Keep a rollback or correction path for the full migration. When the wrong values are already in production, identify the affected records before attempting a bulk correction.