OCR Invoice Scanning
How Marginly's OCR invoice scanning works — photographing invoices to extract data automatically.
Overview
OCR invoice scanning is available on Pro plans and above. It allows you to photograph an invoice and have Marginly automatically extract the key data — supplier name, invoice date, invoice number, line items, and total amount — saving you the time of manual data entry.
How to scan an invoice
- In the Budget & Spend module, click "Scan Invoice" (the camera icon)
- On mobile, this opens your camera. Take a clear photo of the invoice — make sure the full invoice is in frame and there is adequate lighting.
- On desktop, you can upload a PDF or image file (JPEG, PNG, or PDF)
- Marginly's OCR engine extracts the data and pre-fills the invoice form
- Review the extracted data for accuracy and correct any errors before saving
Accuracy
OCR accuracy depends on the quality of the invoice and the scan. Printed invoices from accounting software typically have very high accuracy (95%+). Handwritten invoices or poor-quality photographs may extract with lower accuracy — always review before saving.
Supported formats
- JPEG and PNG image files
- PDF documents (including multi-page PDFs — each page is scanned separately)
- Camera photos taken directly in the app (mobile)
Privacy
Scanned invoice images are processed by our OCR service and then stored encrypted in your project document vault. They are never shared with third parties. You can delete scanned invoices at any time from the document vault.
Tips for best results
- Ensure the full invoice is in frame with no cropping
- Good, even lighting — avoid glare from a desk lamp
- For paper invoices, lay them flat rather than holding them
- Contrast is important — light paper with dark print scans best
- PDF invoices emailed by suppliers will always give better accuracy than photos of printed invoices