Two PDFs can look identical on screen and be completely different inside. A text PDF, like the e-statements banks generate, stores every character as text, so converters can read it exactly. A scanned PDF is a picture of a page. Software has to guess the characters with optical character recognition (OCR), and with statements one misread digit throws off every balance after it.
Three quick tests
- Search for a word. Open the PDF, press Ctrl+F (⌘+F on a Mac) and search for a word you can see on the page, such as your bank's name or a payee. If it's found and highlighted, the PDF has text. If nothing is found, it is almost certainly scanned.
- Try to select a line. Drag across a transaction. In a text PDF you can highlight individual words. In a scanned PDF you either select nothing or the whole page as one image.
- Zoom in to 400%. Text PDFs stay sharp at any zoom. Scans turn blurry or blocky, and you may see a slight tilt, specks or the edge of the paper.
One edge case: some PDFs pass the search test but paste as gibberish, because the PDF uses an unusual font encoding. Converters may struggle with these too. Downloading a fresh copy of the statement sometimes helps.
Where scanned statements come from
- Printed statements or passbook pages that were scanned or photographed.
- Statements that were printed and then "printed to PDF" or faxed. Printing a page as an image loses its text.
- Statements from some branches or older accounts that are issued as scanned copies.
Statements you download from net banking, or receive as an e-statement by email, are normally text PDFs.
If your statement is scanned
- Look for a text version first. Download the same period from net banking or the bank's app. This is nearly always quicker and more accurate than OCR.
- If you must use OCR, desktop PDF editors with OCR keep the file on your computer. Cloud services also work: for example, opening a PDF with Google Docs from Google Drive runs OCR on it. But the statement is then stored in that account, so treat it like any other upload of financial data.
- Always check OCR output against the balances. OCR commonly confuses 1 and 7, 5 and 6, 0 and 8, and drops decimal points. A running-balance check catches these. Here is how to do it in Excel.
What StatementSieve does with each type
StatementSieve reads text PDFs only. If you choose a scanned statement, it tells you that no text was found instead of guessing. It doesn't do OCR. Because the conversion runs in your browser, the file isn't uploaded in either case.