Why do dates change when a towing CSV opens in a spreadsheet?

Establish what the records mean

Inspect a small raw sample and record the source’s date definition. Values such as 03/04/26 are ambiguous without context, and leading zeroes may disappear when treated as numbers. Distinguish dates from identifiers that happen to look numeric. The reconciliation lab accepts explicit calendar dates in YYYY-MM-DD form.

A worked example

Fictional fixtures: 2026-03-04 clearly identifies March 4; 03/04/26 needs a source-format decision; 2026-02-30 is invalid; a blank date is missing. The lab should reject the invalid or missing required date rather than silently rolling February 30 into March. A corrected value must come from source evidence, not a convenient guess.

Choose the next action

Import the original as text where the spreadsheet workflow allows it, then create a separate normalized column using a documented method. Keep time zones if timestamps are involved; a date-only field cannot recover a missing time. Do not overwrite the original while trying several formats. Verify exact interface steps against the spreadsheet version being used.

Check whether it worked

Check a known day/month boundary, an invalid-date case and the before/after row count. Reopen the exported CSV as text to ensure the intended values survived. If only a software library was used, describe that verification accurately instead of claiming the file was tested in live Excel or Google Sheets.

A repeatable review order

1. Original raw date. Record the evidence and mark any missing answer as unknown.
2. Source format and event meaning. Record the evidence and mark any missing answer as unknown.
3. Normalized date or unresolved status. Record the evidence and mark any missing answer as unknown.
4. Boundary and invalid-date fixtures. Record the evidence and mark any missing answer as unknown.
5. Reopened export verification. Record the evidence and mark any missing answer as unknown.