When your TD Canada Trust 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. Scanned or photographed statements add a layer of difficulty that clean digital files avoid entirely: the numbers and labels exist only as pixels on a page, not as a table the computer can read directly. Upload a scanned statement and the engine applies OCR to pull every character from the image. It does not stop there: the extracted transactions are verified against the printed closing balance line by line, and any row whose numbers do not add up is flagged for review rather than accepted silently. That means a smudged digit or a stray mark that tricks the OCR into misreading a figure will be caught before the data reaches your ledger.
If a transaction description on your TD Canada Trust statement is long enough to wrap across two or three lines, it still needs to land in a single row once converted. Long transaction descriptions that wrap across two or three lines inside the statement table are a common source of split rows in a simple copy-paste workflow. The engine reads the full transaction text as a single record, preserving it as one row in the extracted table even when the original layout wraps the description for space. The running-balance check still applies: every line's arithmetic is verified against the closing figure printed on the statement, so a row break that might otherwise split a debit across two phantom transactions is resolved cleanly in the export.