Importer
Drop PDF, image, XML or ZIP files into a company file, follow their automatic analysis and understand the import statuses.
The importer turns a file into an accounting document ready to validate. You drop invoices, credit notes, receipts or archives; NovaFisko reads each document, identifies the third party, extracts the lines and suggests an account and a VAT code. Nothing is booked without your validation.
Dropping files
- Open the Documents tab of the company file.
- Drag your files into the "Drop your PDF, image and XML files" area, or click Import then Choose files.
- An "Importing … documents" panel opens and shows the progress of each file.
- At the end, the panel shows "Import finished" with the number of documents processed and the number of errors. You can close it: the documents stay in the To validate filter.
Formats and limits
| Item | Rule |
|---|---|
| Accepted formats | PDF, JPG, PNG, HEIC, WEBP, GIF, XML (UBL Peppol BIS 3 or CII), ZIP |
| Size per file | 25 MB at most |
| Files per upload | 20 at most |
| ZIP archive | Extracted on arrival; at most 50 files handled per archive |
| 100 pages at most; a password-protected PDF is refused | |
| Scanned PDF | 20 MB at most for automatic reading |
A file whose format is not supported is set aside as soon as it is dropped, with the message "… file(s) ignored: unsupported format."
How a document is analysed
The actual type of the file is recognised from its content, not only from its extension. The analysis then depends on what the file is.
| File | Method |
|---|---|
| XML UBL or CII | Direct reading of the electronic invoice: no interpretation, the data is exact. The embedded PDF is used as the preview. |
| PDF with text | Reading of the text layer, then extraction by artificial intelligence. |
| Scanned PDF, photo | Visual reading by artificial intelligence. |
| ZIP | Each supported file becomes a separate import; the archive takes the status "Archive extracted". |
The steps are shown live: Uploading file, Queued, Reading text, AI analysis, Extracting lines, Matching third party, Analysis finished.
After the extraction, NovaFisko matches the document against the company file:
- Third party. Search by VAT number, then by name (strong similarity, legal forms ignored). If nothing matches, the creation of a third party is suggested with the details read, completed from the company directory when it is available.
- Account. Default account of the third party, otherwise a match on keywords in the line description, otherwise a generic account (services for a purchase, default sales account for a sale).
- VAT code. Default code of the third party if its rate matches, otherwise according to the direction, the rate and the country of the third party: intra-Community code for an EU third party at 0%, import or export code outside the EU.
- Nature. A till receipt, a restaurant VAT stub or a parking ticket is flagged as such, because its deductibility rules are specific.
If the artificial intelligence analysis is not configured on your environment, the application shows "AI analysis is not configured: only XML files are analysed." Electronic invoices can still be imported.
Import statuses
| Status | Meaning | Next step |
|---|---|---|
| Uploading, Queued, Analysing | Processing in progress. | Wait; an import left in the queue is picked up again automatically every minute. |
| Ready | Analysis complete and consistent. | To validate. |
| Needs review | One point requires your attention. | Open the record with Review. |
| Duplicate | The file or the document already exists. | Delete the import, or confirm the import anyway. |
| Error | The document could not be read. | Read the message, correct, then run the analysis again. |
| Validated | The document is created in the company file. | Nothing more to do in the importer. |
An import moves to Needs review when the third party, the date, the totals or the lines were not found, when the lines do not match the document total (gap above 2%), when the document number is missing or when the analysis confidence is below 60%.
Duplicate detection
Two checks prevent double entry:
- same file: a strictly identical file already imported into the company file is recognised immediately, without a new analysis;
- same document: a document with the same number, the same direction and the same third party (or the same VAT number) already exists in the company file.
Imports left in Error or Duplicate are deleted automatically after 30 days, together with their file. A validated import is never deleted this way: its document belongs to the company file.
Common error messages
| Message | What to do |
|---|---|
| This PDF is password-protected | Remove the protection, then upload it again. |
| This PDF has … pages, which is too many for an automatic analysis | Split the file. |
| This XML file is not a UBL invoice or credit note nor a CII invoice | Upload the corresponding PDF. |
| HEIC photos cannot be converted on this server | Upload a JPG or a PNG. |
| This ZIP archive contains no supported document | Check the content of the archive. |
| The automatic analysis failed | Try again in a moment with Retry. |
Always prefer the XML file of an electronic invoice to its PDF: the amounts, the rates and the VAT number are taken over as they are, with no risk of a reading error.