How to Convert Scanned Bank Statements to Excel

By Syed Hassan, creator of SafeStatement · Last updated September 29, 2026

PDF bank statements come in two very different kinds. Digital statements, downloaded from online banking, contain a real text layer, so you can highlight and copy the words. Scanned statements are photos of paper. To a computer they're just pixels with no text in them at all. Text-based converters, including ours, can read the first kind directly. The second kind needs OCR (optical character recognition) first, to turn the picture of text back into actual characters.

This guide explains how to tell which kind you have, how to avoid OCR where possible, the free OCR tools that work well, and how to catch the errors OCR always introduces.

Step 0: Check which kind of PDF you have

Open the PDF and try to drag your mouse across a line of transactions.

• Text highlights word by word: it's digital. You don't need OCR. Drop it into the statement converter.
• A blue box covers the whole page, or nothing highlights: it's a scan or image.
• Text highlights but copies as gibberish: the PDF has a broken or custom font encoding. Treat it like a scan and run OCR.

Other signs of a scan are slightly tilted pages, grey speckles, a visible paper edge, or a PDF that came from a phone scanning app or a photocopier.

First, try to avoid OCR entirely

OCR always loses some accuracy, and on financial documents a single misread digit matters. Before starting, check these:

• Online banking archives. Most banks keep several years of digital statements (often 7 in the US and UK). If the period you need is there, download it and skip OCR completely.
• Transaction exports. Many banks let you download transaction history as CSV directly, which is cleaner than any statement.
• Ask the bank. Banks can usually regenerate older statements, sometimes for a small fee. Ten minutes on support chat often beats an hour of OCR cleanup.

Free OCR option 1: Google Drive

1. Upload the scanned PDF (or JPG/PNG) to Google Drive.
2. Right-click it → Open with → Google Docs.
3. Google runs OCR and opens a document with the page image at the top and the recognised text below it.
4. Copy the transaction lines into Google Sheets or Excel. Use Data → Split text to columns (Sheets) or Data → Text to Columns (Excel) to separate date, description and amount.

Drive works best on clean, straight, high-contrast scans. Tables sometimes come out as one long column that you'll need to rearrange. A privacy note, given what this site is about: Drive is still an upload to a server. Google's terms are clearer than those of an unknown converter site, but if that bothers you, use one of the offline options below.

Free OCR option 2: Excel "Data from Picture"

Excel 365 (Windows, Mac and the mobile apps) can turn a photo of a table straight into cells: Data → From Picture → Picture From File. It's aimed at tables, so it keeps rows and columns better than document OCR does, and it shows a review pane where you can correct flagged cells before inserting. It works one image at a time, so export each PDF page as an image first, or take a screenshot of it. The image is processed by Microsoft's cloud service.

Free OCR option 3: offline, on your own computer

• Windows: OneNote can extract text from images. Paste a page image into a note, right-click it, and choose Copy Text from Picture. The Snipping Tool in Windows 11 also has a Text actions button that copies text from a screenshot.
• macOS / iPhone: Live Text lets you select text directly in images and in scanned PDFs opened in Preview or Photos. Drag across the table and copy.
• Linux / power users: the open-source ocrmypdf tool adds a text layer to a scanned PDF locally (ocrmypdf scan.pdf searchable.pdf). You can then feed the result into a normal text-based converter.

The offline options keep the statement on your own machine, which is the same principle our converter is built on.

How to get better OCR results

• Scan at 300 DPI in black-and-white or greyscale. Higher resolutions rarely help. Lower ones noticeably hurt accuracy on small digits.
• Keep pages straight. Skewed scans confuse column detection. Most scanner apps have an auto-straighten option.
• Crop out the margins, logos and advertising panels so the OCR only sees the transaction table.
• Phone photos: use a scanning app (Google Drive's scan feature, iOS Notes' document scanner or Microsoft Lens) rather than the camera. They flatten, crop and increase contrast automatically.

The errors OCR makes on bank statements

OCR misreads specific characters in predictable ways. Watch for these:

A quick way to catch text-that-should-be-numbers in Excel: add a helper column with =ISNUMBER(C2) and filter for FALSE. Every row that shows up has a letter hiding in the amount.

Always reconcile after OCR

The reliable test is the statement's own arithmetic: opening balance + credits − debits = closing balance. If your extracted numbers reproduce the closing balance exactly, the amounts are almost certainly right. If they're off, the size of the difference is a clue. A difference that is an exact multiple of 9 often means two digits were swapped, and a difference 99× an amount means a lost decimal point. Our Excel reconciliation guide walks through the formulas.

When OCR isn't worth it

For a page or two, OCR plus cleanup is fine. For 50 pages of faint photocopies, the checking can take longer than asking the bank for digital copies, or longer than typing the transactions yourself with a second person reading them back. For tax or legal work, keep the original scans alongside your spreadsheet. The spreadsheet is your working copy and the scan is the record.

Once you have real text

If your OCR tool produced a searchable PDF (with a text layer), run it through the free statement converter to get Excel or CSV. If you pasted the text into a spreadsheet yourself, tidy it into Date / Description / Amount columns. From there, follow our QuickBooks import guide or budgeting guide depending on where the data is going.