How to convert a bank statement PDF to CSV

First identify the kind of PDF

A digital PDF usually contains selectable text and may preserve columns. A scanned PDF is a set of page images and needs optical character recognition (OCR). A password-protected PDF must be opened with an authorised password before either method can work. These inputs should not be treated as interchangeable because their error patterns differ.

Define the output columns before extracting

For most accounting workflows the useful fields are transaction date, description or reference, debit, credit and running balance. Some banks combine debit and credit into one signed amount; others use separate columns. Decide how the target CSV represents signs, decimals, dates and opening balances before generating hundreds of rows.

Preserve statement context

Page number, account period and source row references make review much easier. A converter should flag text it cannot place confidently rather than drop it. Multi-line descriptions, repeated page headers and brought-forward balances are common sources of accidental rows.

Validate the result

• Recalculate balance movement where the statement provides running balances.
• Compare extracted row counts with the visible statement sections.
• Look for duplicate date-description-amount combinations across page boundaries.
• Review OCR warnings, unusual characters and missing values.
• Compare opening and closing balances with the source document.

Review before importing

CSV is a transport format, not proof of correctness. Keep the original statement available to an authorised reviewer, resolve warnings and test a small import against the accounting system's expected fields.