Why Excel Dates Look Wrong After Exporting to CSV

Excel stores dates internally as a serial number — the count of days since a fixed reference point — and only applies date formatting for display. CSV has no formatting layer at all, so if that underlying number gets exported or read without the date format attached, you see the raw serial number (like 45678) instead of a readable date.

Why this happens

A CSV file is plain text with no type information — no column can be marked "this is a date, display it that way." When you double-click a CSV to open it, Excel re-guesses every column's type from scratch based on its content, and if a date-like string doesn't match the format Excel expects for your system's locale, it can either misinterpret it (swapping day and month) or fall back to showing the literal text instead of formatting it.

The locale trap: MM/DD vs. DD/MM

US Excel installations assume MM/DD/YYYY by default; UK and many other locales assume DD/MM/YYYY. A date like 03/04/2026 is unambiguous to a human only if they know which convention was intended — Excel doesn't know either, and will guess based on your system's regional settings, which may not match the settings the file was originally created under. The safest date format for CSV interchange is ISO 8601 (YYYY-MM-DD), since it's locale-independent and every reasonable parser reads it correctly.

Fixing it after the fact

If a CSV column already shows raw serial numbers after import, select the column, open Format Cells, and choose Date — Excel will reinterpret the same underlying number using the format you pick. If dates are a mix of true serial numbers and plain text strings that only look like dates, a helper column with a formula that checks whether each cell is numeric before converting it is more reliable than trying to reformat the column directly, since reformatting alone can't fix a value that was never stored as a real date in the first place.