Keep 0017 as 0017.

Try this small file

Copy the following synthetic text into a UTF-8 file. Treat the first record as a header and use comma as the separator.

id,date_label,team
0017,03/04/2026,fern
"0018",04/03/2026,moss

Expected data field strings: ["0017", "03/04/2026", "fern"] and ["0018", "04/03/2026", "moss"]. The quotes around 0018 delimit a field; they are not part of its value and do not declare a text type.

Check the importer, not just the file

1. Keep an unchanged copy of the source.
2. Use your destination’s explicit CSV import settings. Where supported, select text/string handling for IDs and ambiguous dates instead of automatic conversion.
3. Inspect the imported values: do both IDs still contain four characters? Is date text still unchanged?
4. After exporting, inspect the raw field values again. A visually formatted spreadsheet cell is not proof of the saved value.

If an importer already discarded leading zeros, the resulting number cannot tell you how many zeros the original contained. Return to the source rather than guessing.

The boundary

CSV does not settle whether 03/04/2026 means March 4 or April 3. Resolve date meaning from the producer’s schema, not a locale guess. Quoting alone also does not prevent downstream formula interpretation; read that separate boundary.