Check a bank CSV before opening it in Excel

A CSV can be structurally valid and still open badly in your spreadsheet. One column, altered dates, missing leading zeros or amounts stored as text are import problems to investigate before changing the underlying transactions.

Keep a raw copy and test a small, understood sample. Our fictional CSV download uses UTF-8, comma-separated fields, ISO dates and a decimal point. It is a demonstrator, not a universal import template for every accounting package or locale.

Separate the file’s structure from its display

In a plain-text view, identify the header, delimiter and quoted descriptions. A comma inside a quoted description belongs to that field. Counting every visible comma is not a reliable way to count columns. A line break inside a quoted field also needs to stay with its description.

If Excel puts everything in one column, investigate the delimiter chosen by its importer. Do not fix the file by globally replacing commas: some may be decimal marks or part of a description. Microsoft’s CSV import guidance describes explicit import options and delimiter selection. Menu names and available options depend on your Excel version.

Decide each column’s meaning first

Dates need a known convention. The string 03/04 is ambiguous without context; changing the display format after a mistaken import does not prove that the stored date is now correct. Prefer a deliberate date interpretation and compare against the source.

Amounts must be numeric for sums, with a known decimal convention and debit/credit policy. Preserve account identifiers and long references as text, even when they contain digits. Leading zeros are information in an identifier. A numeric display that looks neat can conceal a reference that was altered during import.

For the sample, opening 1,000.00 plus credits 1,523.40 minus debits 313.30 equals 2,210.10. This independent check can catch a decimal or sign problem, but it will not detect every changed reference or date. See the balance-review protocol for its limits.

Compare the imported result before using it

Check the five sample rows, the first and last date, numeric totals and every description. Then test an accent, a quoted comma and a leading-zero identifier in a separate synthetic fixture if those occur in your real workflow. Use fictional data for experiments that might be shared.

Do not overwrite the only raw file while testing. Save the reviewed workbook separately and record the import choices. If another system receives the CSV, its schema wins: column names, signed amount versus separate debit/credit, date representation and required reference fields must be verified with its documentation.

Know what the example does not promise

BankStatementScan is in preparation. There is no real file upload or generated Excel workbook here, and the sample CSV does not implement a payment, provider or confidential file-processing policy. The features page describes the intended workflow; the pricing page is a non-billable hypothesis. Review is part of the task, not a guarantee of perfect data.

You can request access through this website’s form. It sends your email and visit context to BankStatementLab for BankStatementScan; a recorded request does not create an account or provide immediate access. See privacy.