How to Troubleshoot Common Issues
Resolve document processing problems, extraction failures, and confidence score issues.
Occasionally you may encounter issues with document uploads, extraction quality, or processing. This guide covers the most common problems and how to resolve them.
Before you start
- You have documents that aren’t processing correctly
- You’ve checked the document file format and size
- You have access to your original documents for re-scanning or re-uploading
Steps
1. Document stuck in “Processing”
Problem: A document has shown “Processing” status for more than 30 minutes.
Solution:
- Refresh your browser to see if processing has completed
- Check your internet connection—a lost connection can pause processing
- If still processing after an hour, delete the document and re-upload it
- Try uploading from a different browser or device to isolate the issue
- Contact support if the problem persists; it may indicate a server issue
2. Low confidence on clean documents
Problem: A clear, well-scanned document shows confidence below 70% on most fields.
Solution:
- Check the document type—is it actually an invoice, or is it an unusual format? CargoLint is optimized for standard customs documents.
- Verify scan quality: Did you scan at 300 DPI or higher? Is the contrast good? (See Scanning Tips)
- Look for handwritten sections or annotations that may confuse the extraction
- Try uploading a different page of the document if it’s multi-page
- If confidence remains low, the document may use an uncommon format or language variant—plan for manual data entry on this type
3. Upload failures or “Unsupported file type” errors
Problem: You see an error saying your file can’t be uploaded.
Solution:
- Verify your file format is supported: PDF, PNG, JPEG, or TIFF
- Check file size—is it under 5MB? If it exceeds 5MB, compress the PDF (use online PDF compression tools) or split multi-page documents
- Confirm the file isn’t corrupted by trying to open it on your computer first
- If it’s a PDF, try converting it to PNG using an online converter, then upload the image
- Ensure the file extension matches its actual format (if it says .pdf, it must be a real PDF, not a renamed image)
4. Missing or incomplete extracted fields
Problem: CargoLint didn’t extract certain fields (invoice number, date, totals) that are clearly visible in the document.
Solution:
- This usually indicates the field is in an unusual location or format. Check that critical fields are in standard positions on your document
- If your invoices use a custom template, the field may not match CargoLint’s expected layout—review a few manually to confirm the pattern
- Check if the field uses special characters or formatting (for example, a date written as “14-Mar-2026” instead of “2026-03-14”) that may confuse extraction
- Upload another document of the same type to see if the problem is consistent
- If multiple documents from the same vendor are missing the same field, that vendor’s document format may not be optimal for extraction—consider asking them to use a standard layout
5. Batch upload slow or stalling
Problem: You’re uploading 50 documents, but the process is slow or stops partway through.
Solution:
- Upload in smaller batches (10 documents at a time) rather than all at once
- Check your internet connection speed—large batch uploads require stable bandwidth
- Close unnecessary browser tabs or apps to free up memory
- Try uploading from a wired internet connection instead of WiFi if possible
- Stagger uploads throughout the day rather than uploading 100 documents in one hour
6. All documents showing low confidence suddenly
Problem: Yesterday your documents averaged 85% confidence, but today’s uploads average 55%.
Solution:
- Check if you’re using different scanning equipment (new scanner, phone camera, or different phone)
- Verify scan quality hasn’t changed—lighting, DPI, or contrast issues can appear suddenly
- Confirm you’re still uploading the same document types (mixing invoices with contracts, for example)
- If multiple team members are uploading, ask if anyone changed scanning procedures
- Check your scan settings if using a scanner—some devices reset to lower quality after firmware updates
7. Document appears twice or is missing after upload
Problem: A document you uploaded isn’t showing in your Documents list, or you see duplicate uploads.
Solution:
- If missing: Refresh the Documents page. If it’s still not there, check the upload progress—it may still be processing in the background
- Use search to find the document by name or date
- Check if your browser filter is hiding it (apply a narrow date range filter by mistake, for example)
- If you see duplicates: This can happen if you upload the same file twice. Delete the unwanted duplicate from the Documents page
- Contact support if a document genuinely disappears—backups are maintained
8. Can’t approve document due to required fields
Problem: You’re trying to approve a document in the review queue, but the Approve button is greyed out.
Solution:
- You may have unfilled required fields. Scroll through the document to find empty fields (shown in red or highlighted)
- Fill in all empty fields or correct all fields below 70% confidence
- Some documents require certain information (like HS codes for invoices) before approval—verify you’ve filled these in
- If a field is impossible to determine from the document, contact your admin—they may need to adjust field requirements for your document type
Tip: When troubleshooting extraction quality, start with scan quality. Most issues trace back to unclear or poorly formatted source documents.
Tip: Keep a sample of successful uploads to compare against problematic ones. This helps identify what changed in your scanning process.
Tip: If you experience repeated issues with one document type, reach out to CargoLint support with examples. Your feedback helps improve the platform.
What’s next
- Scanning Tips – Improve document quality to avoid extraction issues
- Understanding Confidence – Learn why confidence scores vary
- Reviewing Documents – Correct extracted data for problematic fields