Duplicate detection takes place at multiple stages in the processing workflow:

1. On submission: the platform checks whether an e-invoice (XML) has already been submitted electronically before.
2. On receipt: checks the combination of supplier, invoice date and invoice number.
3. During PDF conversion: additional check by the conversion partner that converts the PDF to an e-invoice.
4. Cross-channel: detection when the same invoice arrives both as XML and as PDF, regardless of the channel.
5. In the ERP package: check on supplier and invoice number when downloading to your accounting software.
6. On payment: final check during payment approval.

For genuine e-invoices (XML) the duplicate detection is 100% reliable, because the invoice number and all details are compared exactly.

For PDF invoices the detection depends on recognition by the IDR (Intelligent Document Recogniser). In virtually all cases duplicates are correctly identified, but there is a small margin for invoices with an unusual layout or when the invoice number is read slightly differently by the OCR.

The check in step 5 (in the ERP package) is a separate check performed by your accounting software itself, independent of the platform detection above. The platform only marks an invoice as a duplicate on an exact match of supplier, invoice date and invoice number together. Your accounting software can also reject an invoice based on supplier and invoice number alone, even if the invoice date differs. That is then not a platform duplicate. See ERP troubleshooting for package-specific error messages, such as [U4BadRequest]: Invoice number already exists in Unit4 ERPx.

An invoice recognised as a duplicate receives the status "Duplicate" in the Inbox. The invoice is not automatically forwarded to your accounting software. You receive an email notification when a duplicate invoice is detected.

The original invoice retains its status and is processed normally.

In exceptional cases an invoice may be incorrectly marked as a duplicate. This can happen when two different invoices coincidentally have the same combination of supplier, date and number, or when the OCR recognition reads an invoice number slightly differently.

A known pattern here: the recognition reads a recurring header field on the PDF, such as a postal code, instead of the actual invoice number. If that incorrect number occurs on multiple invoices from the same supplier, the platform correctly flags them based on the key, but incorrectly as a duplicate in terms of content. This occurs, for example, with invoices from Germany, where a postal code (PLZ) is sometimes read as the invoice number.

To release an incorrectly marked invoice:

1. Open the invoice in the Inbox.
2. Click the menu (three dots).
3. Choose "Mark as original".

The invoice is then processed and, if you have a software connection, forwarded to your accounting software.

PDF conversion always removes all spaces from the invoice number. Supplier XML (Peppol or an XML upload) keeps the invoice number as supplied, including spaces.

The platform check matches exactly on supplier, invoice date and invoice number. A space difference alone between the PDF conversion and the supplier XML produces two originals. That is not a duplicate and not a fault.

XML is leading. If the PDF conversion differs from the XML, submit only the XML. Adjust a differing file name with the sender.