Why a scanned statement needs OCR first
A scanned bank statement may look identical to a normal PDF on screen while containing no usable text layer underneath. A transaction extractor cannot reliably identify dates, descriptions and amounts from text that does not exist in the PDF.
OCR adds searchable text to the page images. Once the statement has a usable text layer, the extracted text can be passed to the bank-statement workflow.
Make the statement searchable
Open the OCR tool, choose the scanned PDF and create a searchable copy. Review several names, dates and amounts in the OCR result before moving on.
Poor scans, skewed pages, unusual fonts, stamps and low-resolution images can all reduce recognition accuracy. If critical digits are visibly wrong after OCR, do not rely on the extracted spreadsheet.
Extract the transaction table after OCR
Open the searchable result in Bank Statement to Excel. The workflow extracts table-style transaction data into CSV or TSV that can be opened in Excel, Google Sheets or another spreadsheet application.
Different banks use different statement layouts, so column detection should always be reviewed rather than assumed to be perfect.
Verify numbers before accounting use
OCR can confuse characters such as 0 and O, 1 and I, or misread punctuation in amounts. Compare important dates, debits, credits and balances against the original scanned statement.
For financial records, checking a few representative transactions is not enough when the spreadsheet will be used for reconciliation or reporting. Review the rows and totals that matter to your workflow.
Keep both copies
Keep the original scan as the source record and the OCR/searchable copy as a working document. Keep the extracted CSV or TSV separately so corrections do not overwrite the source.
This gives you a clear path back to the original page whenever a spreadsheet value looks suspicious.