A readable source makes review easier
Scanned invoice data extraction reads a document image and identifies fields such as the supplier, invoice number, dates, and amounts. A scan can look acceptable at page size while a small decimal point or final invoice digit remains unreadable. Inspect those details before uploading.
Parzable’s upload interface accepts PDFs and image files. For an invoice with several pages, keep the pages together in one PDF and in their original order. Keep separate invoices in separate files. Test an unusual image format with a small sample before relying on it for a batch.
When the original digital PDF is available, use it instead of a screenshot of an email attachment. Each extra capture or compression step can make small print harder to inspect.
A capture checklist for paper invoices
- Include the whole page. Check the corners, footer, page numbers, and any totals printed near an edge.
- Keep the page flat and upright. Avoid curved pages, steep camera angles, and rotated text.
- Use even light. Move glare off the page and keep phone shadows away from figures.
- Check focus at the smallest text. Zoom into the invoice number, decimal marks, and item codes.
- Include continuation pages. A page with only totals can still be essential.
- Retain the original. Improve a copy rather than overwriting your only source.
The Tesseract image-quality guide discusses how factors such as resolution, skew, borders, and background noise affect OCR. Those are useful capture considerations; they are not a claim that Parzable uses Tesseract or achieves a particular accuracy.
Match the correction to the reading problem
| Observed problem | Inspect in the source | Next action |
|---|---|---|
| Invoice number ends in O instead of 0 | The original digit at full size | Compare against the document; do not guess from the numbering pattern. |
| Decimal mark missing | Focus, glare, compression | Rescan or request the original PDF. |
| Last item missing | Bottom margin and next page | Restore the complete page set. |
| Subtotal captured as total | Labels around both amounts | Clarify the schema description and review the rerun. |
Repeatedly submitting the same unreadable image will not restore information absent from the file. Request a better copy when the source itself cannot resolve a critical value.
Run a small, varied sample first
Create a schema, select it in Extract, and try a clear scan, a typical phone photo, and a multi-page invoice from your own workflow. Review the returned fields against the originals. Record which issues come from capture quality and which come from unclear field definitions.
A successful upload only establishes that the file was accepted. A completed extraction still requires checking. Pay particular attention to similar characters, negative amounts, page-spanning item tables, and multiple totals. Use the validation checklist before the data moves into a final purchase register.
If you are comparing scanning methods, keep the invoice and schema fixed and change one capture setting at a time. That gives you useful evidence about your documents without turning a few successful examples into a general accuracy claim.
Put the workflow to work
Try your own invoice in Parzable
Create an account, define a schema, and test a small sample. Review extracted values before using them. See current plans and page allowances.
Try Parzable