Convert a Scanned Bank Statement to Excel with OCR
Updated
You can convert a scanned bank statement to Excel by running it through OCR (optical character recognition), which turns the image of each page into text, and then rebuilding the transaction table from that text. It works well on clean, straight scans and less well on blurry or skewed ones, so the two things that matter most are making a good scan and checking the output against the statement’s running balance. This guide explains how to do both.
Text-based PDFs vs scanned PDFs
Bank statements come in two very different kinds of PDF, even though they look the same on screen.
- Text-based (digital) PDFs are generated by your bank’s systems, usually when you download a statement from online banking. The characters are stored as real text with exact positions, so a converter can read them precisely.
- Scanned PDFs are pictures of paper pages, made with a scanner, a phone camera or a fax. There’s no text inside, only pixels, so software has to recognise each character visually.
If you have the choice, the digital version always converts more accurately and much faster. Many banks let you download past statements through online banking, so it’s worth checking there before you scan a pile of paper.
How to tell which kind you have
Open the PDF in any reader and try these checks:
- Select a word. Click and drag across one word in a description. If individual words highlight, it’s text-based. If a whole rectangle or the entire page highlights, it’s an image.
- Search for a word. Use Ctrl+F (Cmd+F on a Mac) to look for a word you can see on the page. No results usually means a scan.
- Zoom right in. Text in a digital PDF stays sharp at any zoom level; scanned text goes fuzzy or blocky.
- Consider where it came from. Downloaded from your bank: almost certainly text. Received as a photo or from a scanner: an image.
One caveat: some scanning software adds a hidden OCR text layer, so the file is searchable but may still contain recognition errors. Treat those files like scans and check them carefully.
How OCR works
Most OCR engines work through similar stages for each page:
- Render the page as an image at a resolution high enough to show small characters.
- Clean it up, for example by converting to black and white and correcting slight rotation.
- Find lines and words by looking for blocks of dark pixels separated by white space.
- Recognise each character by comparing its shape with patterns the engine has learned.
- Output text with positions, so the converter knows where each word sat on the page.
From there, the converter does the same job it does with a digital PDF: it uses the positions to work out which text belongs in the date, description, debit, credit and balance columns.
Every stage can introduce errors. A shadow can hide a digit, a slight tilt can merge two lines, and low resolution can make a 3 look like an 8. That’s why scan quality matters so much.
Converting a scanned statement step by step
- Open the free bank statement converter and add your scanned PDF.
- Switch on OCR. It’s optional because it’s slower than normal text extraction. It runs in your browser, so the file stays on your device, and it currently reads English text.
- Wait while each page is recognised. Long statements take a while, and older or low-powered devices are slower.
- Review the preview and the balance check (more on that below).
- Correct flagged cells, delete stray rows, choose a date format and export to Excel or CSV.
Getting a better scan
Most OCR errors start with the image. These habits make the biggest difference.
Resolution
Scan at 300 DPI, a widely used setting for OCR. Below that, small digits and decimal points start to blur. Going much higher mostly makes files bigger.
Straight, flat pages
Line the page up with the edge of the scanner glass. Skewed pages make rows drift into each other, which confuses both line detection and column alignment. Flatten creased or folded statements before scanning.
Contrast and colour
For printed statements, grayscale or black and white usually works better than colour. Avoid settings that lighten the image, since faint grey text on a pale background is hard to read, and switch off any photo enhancement that smooths detail.
Phone photos
If you have to use a phone:
- Use a document-scanning app or your phone’s built-in document scanner, which crops and flattens the page.
- Shoot in bright, even light, with no shadow from your hand or phone.
- Hold the phone directly above the page rather than at an angle.
- Capture one page per image and fill the frame.
Compression
Heavily compressed JPEGs create blotchy artefacts around characters. If your scanner offers a quality setting, choose high quality, or save as PDF or PNG.
Checking OCR output with the balance check
Even a good scan produces the odd misread. Rather than proofreading every line, let the numbers do the work. The converter re-adds the running balance row by row: each row’s balance should equal the previous balance plus money in, minus money out. When a row doesn’t add up, it’s flagged.
A flag usually points to one of these:
| What you see | Likely OCR misread |
|---|---|
| Amount off by a factor of 100 | Decimal point dropped (12.50 read as 1250) |
| Small, round difference | Similar digits confused: 3 and 8, 5 and 6, 1 and 7 |
| Money in treated as money out | Minus sign or CR/DR marker missed |
| Balance jumps unexpectedly | A faint or cut-off transaction line was skipped |
| Amounts right but balance wrong | A digit misread in the printed balance itself |
To fix one, compare the flagged row with the original page, correct the cell in the editable preview and confirm the flag clears. A single misread balance can flag two neighbouring rows, so work from the top down and re-check as you go.
It’s also worth comparing total money in and money out with the statement’s summary box, if it has one. For a deeper check in a spreadsheet, see how to reconcile a bank statement in Excel.
Limitations to know about
- Handwriting isn’t supported. Handwritten notes or amounts won’t be read reliably.
- Unusual layouts may need small fixes. Rotated tables, heavy shading or several transaction columns per page can need manual tidying.
- OCR reads English only. For statements in other languages, convert the digital PDF if you can.
- It’s slower. OCR takes noticeably longer than extracting text from a digital PDF.
Key takeaways
- Check whether your PDF is digital or scanned first, and download the digital version from online banking if you can.
- Scan at around 300 DPI, keep pages straight and flat, and use good contrast.
- OCR turns page images into text, and every blur or shadow raises the chance of a misread.
- Use the balance check to find errors instead of proofreading every row.
- Your scanned statement never leaves your device. For more on that, see the privacy page or our guide on whether it’s safe to convert bank statements online.
Frequently asked questions
Can you convert a scanned bank statement to Excel?
Yes. A converter with OCR reads the characters from each page image and rebuilds the transaction table. Accuracy depends heavily on the quality of the scan, so always check the result.
How do I know if my bank statement PDF is scanned?
Try selecting a single word, or search for a word you can see with Ctrl+F (Cmd+F on a Mac). If you can't select or find individual words, the PDF is almost certainly a scanned image.
What resolution should I scan a bank statement at?
300 DPI is a widely used setting for OCR. Lower resolutions blur small digits and decimal points, while much higher settings mostly make the file larger.
Is the OCR on this site done on a server?
No. The optional OCR runs in your browser, so the scanned statement stays on your device. It is slower than converting a digital PDF, especially for long statements or on older computers.
Can OCR read handwritten notes on a statement?
No, handwriting isn't supported. Printed text converts best, so check any handwritten amounts or notes by hand.