PennyweightPennyweight

CSV import says dates must be YYYY-MM-DD

Last updated October 13, 2026

On this page

If a CSV import stops with a message that dates must be YYYY-MM-DD, the file was uploaded directly and some of its dates are written another way, such as 09/03/2026. Nothing was imported. You can upload it again with its date order named, or let your AI convert the dates and import the rows itself.

What you'll see

The message counts the rows that "have a date this import can't use", shows up to five of those dates, and says Dates must be ISO YYYY-MM-DD. It adds "The import is blocked so no partial batch lands". It then suggests three fixes: upload again naming the file's date order, export the statement again with ISO dates, or send the rows with the dates already converted.

The upload step itself accepts the file. The message appears when your AI asks for the preview, and again if it tries to save.

If your AI sends the rows itself with a date it didn't convert, Pennyweight refuses that row by number with "is not a valid ISO date — use YYYY-MM-DD".

Why it happens

Pennyweight saves dates as YYYY-MM-DD. A date such as 09/03/2026 means September 3 in a file that puts the month first and 9 March in one that puts the day first, so the upload never guesses the order. Your AI has to name it when it uploads the file. These orders are supported:

  • YYYY-MM-DD
  • MM/DD/YYYY
  • DD/MM/YYYY
  • DD.MM.YYYY
  • YYYYMMDD

With an order named, single-digit days and months, such as 9/3/2026, are fine. Without one, every date stays exactly as the file writes it, and only YYYY-MM-DD dates pass. Even with an order named, these are refused:

  • a date that fits neither the named order nor YYYY-MM-DD, such as 2026/09/03
  • a two-digit year, such as 09/03/26
  • a date with a time after it
  • a date that doesn't exist, such as 2026-02-30
  • an empty date cell

A single refused date blocks the whole file, so you never end up with half an import.

How to fix it

  1. Share the file with your AI and name the account.
  2. Tell it whether the month or the day comes first. Your AI can upload the file again with that order, or convert the dates and import the rows itself.
  3. Read the preview. Check the row count and a date or two, then say yes.
Try asking

My CSV import keeps saying the dates are wrong. Can you import us-mmdd.csv into my Studio Checking account for me?

To see the upload read the dates once the order is named:

Try asking

Upload us-mmdd.csv straight to Pennyweight for my Studio Checking account and preview it. The dates are month first. Don't import anything yet.

What you'll see after

For the first prompt, your AI reads the dates as month first, shows a preview, and after you confirm, reports 14 transactions saved to Studio Checking. Rows marked Debit come in as money out and rows marked Credit as money in. See reading an import preview.

For the second, your AI shows a preview of the 14 rows with no date message and saves nothing.

The Type column on the same file

A file like this often has a column headed Type, with Debit or Credit next to amounts that have no sign. The upload reads that column as the direction: Debit is money out and Credit is money in. It never becomes the category.

If rows from an earlier import carry Debit or Credit as their category, your AI can recategorize them in bulk, or you can undo that import and run it again.

Good to know