Scanned Bank Statement to Excel

Text PDFs versus scanned PDFs

There are two kinds of PDF bank statements. A text PDF is created by the bank's systems and contains the characters as text, which is why you can select and search it. A scanned PDF is a picture: the paper statement went through a scanner or a phone camera, and the PDF contains an image of each page. There is no text for software to read, so converters that only extract text return nothing.

You can tell the difference in seconds. Open the PDF and try to select a line of text or use the search shortcut. If nothing highlights, it is a scan. Our converter checks this automatically. If a file has almost no extractable text, it says so and offers OCR instead of showing an empty table.

How the OCR works here

Optical character recognition turns the pixels of the page image into text with positions. We run the Tesseract engine inside your browser using WebAssembly, with the English language model served from this website. Each page is drawn to a canvas, recognised, and the words with their positions are fed into the same statement reader used for text PDFs. That reader works from the positions of the words, so the table logic (dates on the left, amounts aligned to the right) is the same.

Nothing is sent to a server for recognition. That matters for scanned statements in particular, because scans are often the documents people are least comfortable emailing around.

Steps for a scanned statement

1. Drop the scanned PDF into the converter above. It will report that no text was found.
2. Press Run OCR. Expect 10 to 40 seconds per page on a modern laptop, longer on a phone.
3. Review the table carefully. OCR commonly confuses 5 and 6, 1 and 7, 0 and 8, comma and period, and drops faint minus signs.
4. Use the balance banner to find the errors. A red banner and highlighted rows tell you where the running balance breaks, and you correct the cell.
5. Choose Excel or another format and download.

How to get the best OCR result

• Scan at 300 dpi in black and white or greyscale. Higher resolution rarely helps, and colour backgrounds hurt.
• Keep pages straight. A skew of more than a degree or two causes lines to merge. Use the scanner's deskew or a scanning app that detects page edges.
• Avoid shadows and folds. Phone photos should be taken flat, from directly above, in even light.
• One statement page per image page. Do not scan two A5 pages side by side.
• Download the digital statement if you can. Most banks keep several years of statements as text PDFs in online banking, and a text PDF is faster and more accurate than any OCR.

Why the balance check is essential for OCR

OCR can look perfect and still be wrong by a digit. A reader scanning a page of figures will not notice that 1,284.50 became 1,234.50. The reconciliation check will, because the opening balance plus all transactions will no longer equal the closing balance, and the running balance on the affected row will not follow from the row above. The banner shows the difference and the rows involved, so you compare only those rows against the paper original instead of re-reading the whole page.

If the statement does not print an opening and closing balance or a running balance, there is nothing to check against. In that case compare the total of your money in and money out with the statement summary yourself before importing anything into your books.

Limits to be aware of

• English-language statements work best. Other languages with accented characters may be read less accurately.
• Handwriting and heavily stamped pages are not supported.
• Very long scans are slow in a browser. The limit is 80 pages per file; split large scans.
• A password-protected PDF must be unlocked before upload.

Pricing for scanned pages

Scanned pages count like any other page: each page of the PDF uses one page of your allowance when you download, and you are never charged for a failed OCR. The free tier of 3 pages a day is enough to try a short scan.

More converters