workflows

CSV dates or amounts changed? Check locale before automation

Validate delimiters, encoding and per-column types before a cross-border CSV drives your workflow. A release card catches changed IDs, dates and amounts.

A CSV can show the expected columns while changing the meaning of a date, amount or identifier. If the file drives a CRM, invoice or scheduling workflow, the preview alone is not a release check. Compare the imported values with the raw file and the sender’s format specification.

Short answer

Separate delimiter and encoding, column types, and the locale used to parse dates and decimals. Excel lets you select a delimiter during import, and Power Query uses locale when converting text to dates and numbers. A date such as 11/12/2026 is ambiguous without its source convention. Do not pass it to an automated action until the parsed result matches a documented expectation.

An input release card

Column Fictional raw example Expected interpretation Block if
order_id 000123 Text, leading zeros preserved Converted to 123 or scientific notation
delivery_date 11/12/2026 Date in the sender’s stated convention Day and month could be reversed
amount 1234,50 Decimal under sender’s stated locale; currency separate Parsed as 123450, unexpected text or missing currency

This is Pulse’s original release card. The values are fictional, not customer data. A numeric match does not establish a currency match: the same number could be GBP, USD or CAD. Do not infer exchange rates or convert currency silently.

Check before the workflow runs

  1. Keep an untouched original. Ask the sender for delimiter, encoding, date format, decimal symbol and currency. The .csv extension does not specify all of these.
  2. Import into a staging step, select the delimiter and inspect the preview. Zapier’s Formatter help documents delimiter and encoding errors for its own CSV transform; that behavior is product-specific.
  3. Keep identifiers as text. Convert dates and decimals using the file’s locale, not the machine’s default. For a UK file consumed by a US team, day/month dates may need an explicit UK locale. A Canadian file may use different conventions by source; verify instead of assuming one national format.
  4. Compare sentinel rows with the raw file and check row counts and control totals. A matching total cannot prove that an ID or date survived.
  5. Release to production only after the checks pass. If a saved conversion destroyed digits or leading zeros and no original remains, stop rather than promising recovery.

A US operations team receives a UK supplier CSV and keeps IDs as text, checks ambiguous dates and reconciles amounts in the declared currency before updating its order board. This is an editorial example; Pulse did not run the import.

Stop conditions

Stop when an ambiguous date lacks a source specification, the only file has truncated identifiers, currency is missing or control counts differ. Before a contact import, choose a match key. Before external effects, test the workflow. If an active flow loses a field, trace the mapping.

Sources