Turning a bank statement PDF into rows you can actually reconcile looks trivial right up until the closing balance does not match. The arithmetic is rarely the problem. The transcription is. Someone opens a PDF the bank already produced and retypes 200 rows by hand, which is where a digit slips and the month takes an extra afternoon to untangle.

The practical fix is removing the retyping step while keeping the checking step. To see the output shape before committing a whole month, this browser-based converter takes a statement PDF or a photo and exports Excel, CSV or JSON: https://bank-statement-converter.online/

Start with the cleanest source. A PDF downloaded from the online banking portal has a real text layer laid out as a table, so a converter reads structure rather than guesses. A PDF that is a scan of a scan has no text layer, and everything depends on optical recognition; if the tool reports confidence, treat a low value as a prompt to look. A phone photo sits between the two: lay the page flat, keep the camera parallel, and keep all four corners in frame, because perspective and shadows cluster errors near the edges.

Then pick the export that matches what happens next. Excel when a person will read it, CSV when it feeds an import routine or a script, JSON when rows merge with another source and numeric types must survive. The download shape matters too: separate files keep each statement distinct, a ZIP keeps a month together without merging, and one merged export is fastest for a period summary but loses the account boundary unless you add a source column first.

Keep the columns you will check against: transaction date, description, debit, credit, and a running or closing balance, plus currency and account identifier so one workbook cannot quietly mix two accounts. If the statement prints a running balance, keep it, because it turns the next step from a guess into arithmetic.

Then reconcile without sorting or tidying. Find the last exported