Scanned statements are a different problem
A digital PDF from your bank contains real, selectable text — a converter can read it directly. A scannedstatement doesn't: it's a picture of a page, whether it came off a flatbed scanner, a phone camera, or a fax saved as a PDF. To a normal tool there's no text there at all, just pixels. That's why a scan defeats copy-paste and most converters — and why getting one into Excel needs OCR first.
This page is about exactly that case. If your statement is digital, the standard bank statement converteralready handles it; if it's a scan or photo, read on.
What counts as a "scanned" statement
| Source | Selectable text? | Needs OCR? |
|---|---|---|
| Digital PDF from online banking | Yes | No |
| Flatbed / sheet-feed scan (PDF) | No — it's an image | Yes |
| Phone photo of a statement | No — it's an image | Yes |
| Faxed copy saved as PDF | No — it's an image | Yes |
| Image-only PDF (looks normal, can't select) | No | Yes |
A quick test: open the PDF and try to select a line of text with your cursor. If you can't highlight it, it's an image and needs OCR — which FlowParse runs automatically when it detects one.
How OCR turns a picture into transactions
There are two distinct jobs, and you need both. OCRreads the characters off the image — it's the part that "sees" the numbers and words. Then AI extraction understands them: which text is a date, which is an amount, which is the running balance, and how to rejoin a description that wrapped across two lines into a single transaction. OCR alone gives you a wall of loose text; the AI layer turns it into rows you can use.
For the deeper distinction between raw OCR and structured, validated extraction, see OCR vs AI document extraction and the OCR engine behind it.
Why copy-paste and ordinary tools fail on scans
Copy-paste
There's no text to copy — you select the image and get nothing.
Generic PDF table tools
They look for embedded text; an image PDF has none, so they return empty.
OCR + AI converter
Reads the pixels, structures the transactions, validates the balance.
Editable preview
Flags any character the OCR was unsure about so you fix it in seconds.
Step-by-step: scan → Excel
Step 1 — Upload the scan or photo
Drop your scanned PDF, multi-page scan or phone photo. FlowParse detects that it's an image and routes it through OCR.
Step 2 — OCR reads the page
The OCR engine recognises the characters across every page, including continued tables and wrapped descriptions.
Step 3 — AI structures the data
Transactions are assembled — date, description, signed amount, balance — and the running balance is validated end to end.
Step 4 — Review confidence flags
The editable preview highlights any low-confidence field so you can confirm a number the OCR wasn't sure about.
Step 5 — Export to Excel
Download a clean .xlsx (or CSV / .QBO) with all original columns preserved and a summary tab if you want one.
Accuracy and confidence scoring
Scans introduce a quality variable a digital PDF doesn't have, so trust comes from two safeguards. First, every statement is balance-validated: opening + transactions = closing, so an OCR misread on an amount breaks the balance and is caught immediately rather than entering your books. Second, the validation engine assigns a confidence score to each field, surfacing the handful the OCR was unsure about so you review those, not the whole statement.
On a clean scan, accuracy sits around 98% on standard layouts; on a poor one it's lower, but the flags tell you exactly where to look — and correcting a few cells still beats retyping hundreds.
Getting the best results from a scan
- Scan at 300 DPI or higher — more detail means cleaner character recognition.
- Keep the page flat and the whole statement in frame; avoid cut-off edges.
- Good, even lighting and high contrast beat a dark or glare-heavy photo.
- Straighten skewed pages — a level page reads far more accurately than a tilted one.
- Prefer a multi-page PDF scan over many separate images so pages stay in order.
- If a digital PDF exists, use it instead — text is always more accurate than a scan.
Exports and next steps
Once a scan is structured, it behaves like any other converted statement — you choose the output your next tool needs:
| You want… | Export |
|---|---|
| A spreadsheet for analysis | Excel (.xlsx) with a summary tab |
| A universal import file | CSV |
| QuickBooks / Quicken import | .QBO / .QFX / .OFX bank-feed file |
| Xero import | Xero-ready CSV |
| A year of scans combined | One workbook via Smart Merge |
Related: bank statement to CSV, PDF to QBO, combining many statements with Smart Merge, and the step-by-step convert bank statement PDF to Excel guide.
Convert your scanned statement to Excel
Upload a scan, photo or image-only PDF — OCR reads it, AI structures it, and you get clean, balance-validated Excel.