Scanned vs text-based bank statement PDFs: how to tell and what to do
Two PDFs can look identical on screen and behave completely differently when you try to extract data. Here is how to tell which kind you have, and what your options are if it is a scan.
By Muhammad Sharif · October 2026 · 6 min read
"PDF" describes a container, not a content type. Inside that container, a bank statement can be real text, a picture of text, or a mixture of both. Whether a converter can read your statement depends almost entirely on which one you have, so it is worth knowing how to check before you waste time on a tool.
Text-based PDFs
When a bank's system generates a statement, it writes each word to the PDF as characters with a font, a size and a position. These are text-based, sometimes called "digital-native" or "born-digital" PDFs. They have these properties:
- You can click and drag to highlight individual words and copy them.
- The search function (Ctrl+F or Cmd+F) finds merchant names and amounts.
- The file is small, usually 50 to 300 KB per statement.
- Text stays sharp however far you zoom in, because it is rendered from font outlines.
Anything downloaded from online banking or a banking app is almost always text-based. These are the files that converters handle well.
Scanned PDFs
A scanner or a phone camera app produces an image of the page and wraps it in a PDF. The page looks the same to you, but the computer sees pixels, not words. You will recognise these by:
- Clicking and dragging selects the whole page, or a rectangle, rather than words.
- Search finds nothing, even for words you can see.
- The file is large, often 1 MB or more per page.
- Zooming in far enough shows pixels, and edges may look slightly soft or skewed.
- Slight tilt, shadows, a visible paper edge or a phone's finger in the corner.
The ten-second test
- Open the PDF in any reader, such as your browser, Adobe Acrobat Reader or Preview.
- Press Ctrl+F (Cmd+F on a Mac) and search for the word "balance" or any merchant name you can see on the page.
- If it finds and highlights the word, the file is text-based. If it reports no matches, it is an image.
There is also a hybrid case. Some PDFs carry the scanned image with an invisible text layer added by OCR software. Those appear text-based to search and selection, but the text came from recognition rather than the bank, so mistakes in digits can be hiding in the layer. Treat these files as scans for accuracy purposes and always run a balance check.
Why it matters for conversion
A text extraction library, such as the open-source pdf.js engine used by this site's converter, reads the characters and their positions from the file. If there are no characters, it has nothing to work with. There is no hidden data to find, and no setting that makes a picture into words. A converter that handles scans must add an OCR step, and OCR is probabilistic. It can misread 5 as 6, 8 as 3, a comma as a full stop, or drop a minus sign, and each of those changes the money.
| Text-based PDF | Scanned PDF | |
|---|---|---|
| Source | Bank's system or online banking download | Scanner, phone camera, photographed paper |
| Selectable text | Yes | No (unless OCR was applied afterwards) |
| Typical accuracy of conversion | Very high; errors are layout problems | Depends on image quality; digit errors possible |
| Extra steps | None | OCR, then verification against the image |
| Best check | Opening plus transactions equals closing | The same check, plus a visual comparison of large amounts |
How to get a text-based copy instead
Before investing in OCR, try to replace the scan with the real thing:
- Download it again. Log in to online banking and look for statements or documents. Most banks keep several years online, and the PDF you download there is text-based.
- Ask for an electronic copy. If the statement is older than the online archive, call or message the bank and request a digital copy, which they can usually email or make available in secure messages.
- Use the bank's export instead. Many banks offer a CSV or spreadsheet download of recent transactions. It will not carry opening and closing balances, so combine it with the PDF for reconciliation.
- Ask the client. If you are a bookkeeper receiving statements from clients, ask them to download the PDFs from their online banking rather than photographing paper copies.
If you must use a scan
- Re-scan at 300 dpi or higher, flat and straight, with even lighting and no shadows.
- Run OCR in your PDF software or scanner app. Choose a mode that keeps layout if offered.
- Convert the OCR'd file, or copy the recognised text into a spreadsheet.
- Verify every figure against the image. At minimum, recompute the closing balance and compare it with the printed figure, and check the largest ten amounts by eye.
Remember that StatementHarbor does not currently include OCR and says so openly. It is on the roadmap, and until it ships we would rather you know the limit than lose time. The same applies to password-protected files, which must be unlocked in a PDF reader first.
Frequently asked questions
Does a statement saved from a bank's mobile app count as text-based?
Usually yes, if the app downloads the bank's own PDF. A screenshot of the same screen, however, is an image, and converting a screenshot needs OCR.
My PDF is text-based but the converter still finds nothing. Why?
It may have an unusual layout, a password, or dates in a style the converter does not recognise. Work through the troubleshooting guide.
Can a phone camera scan be accurate enough for bookkeeping?
Only with care, and only with verification. Treat any scanned statement as a document that must be reconciled before it is trusted.
