If a transaction description on your The Bank of New York Mellon statement is long enough to wrap across two or three lines, it still needs to land in a single row once converted. Transaction descriptions in a statement document can sometimes grow long: a single entry may carry a multi-line payer name, a full remittance advice or a concatenation of reference codes that push the row well past the usual cell width. Convert·Into reads the entire block of text that belongs to one transaction, even when it wraps across several physical lines. The parser groups the continuation text with its parent row so that nothing is left behind, which means the exported file preserves the complete description and does not split one entry into fragments spread across multiple cells.
When your The Bank of New York Mellon statement arrives as a scan or a photo of the page rather than a clean export, the numbers on it are just pixels until they are extracted. The engine is designed to accept statement documents that have been scanned or photographed, not just those generated as clean digital files. It passes the image through OCR and reconstructs the transaction table from the result. Because the running balance is then verified line by line against the printed closing figure, any character misread as a wrong digit creates a mismatch, and that specific row is flagged. You can correct the one entry in the review step instead of rekeying the entire file.