invoice OCR structured JSON

Here’s a practical structured JSON schema for invoice OCR output. You can use it as a target format for an OCR/parsing pipeline.

{
  "document_type": "invoice",
  "confidence": 0.0,
  "source_file": "",
  "language": "",
  "page_count": 0,
  "invoice": {
    "invoice_number": "",
    "invoice_date": "",
    "due_date": "",
    "purchase_order_number": "",
    "currency": "",
    "payment_terms": "",
    "status": ""
  },
  "seller": {
    "name": "",
    "tax_id": "",
    "vat_id": "",
    "address": {
      "line1": "",
      "line2": "",
      "city": "",
      "state": "",
      "postal_code": "",
      "country": ""
    },
    "phone": "",
    "email": "",
    "website": ""
  },
  "buyer": {
    "name": "",
    "tax_id": "",
    "vat_id": "",
    "address": {
      "line1": "",
      "line2": "",
      "city": "",
      "state": "",
      "postal_code": "",
      "country": ""
    }
  },
  "shipping": {
    "name": "",
    "address": {
      "line1": "",
      "line2": "",
      "city": "",
      "state": "",
      "postal_code": "",
      "country": ""
    },
    "method": "",
    "tracking_number": ""
  },
  "line_items": [
    {
      "line_number": 1,
      "description": "",
      "quantity": 0,
      "unit_of_measure": "",
      "unit_price": 0.0,
      "discount": 0.0,
      "tax_rate": 0.0,
      "tax_amount": 0.0,
      "line_total": 0.0
    }
  ],
  "summary": {
    "subtotal": 0.0,
    "discount_total": 0.0,
    "tax_total": 0.0,
    "shipping_total": 0.0,
    "other_charges": 0.0,
    "grand_total": 0.0,
    "amount_paid": 0.0,
    "amount_due": 0.0
  },
  "payment": {
    "method": "",
    "bank_name": "",
    "account_name": "",
    "account_number": "",
    "iban": "",
    "swift_bic": "",
    "reference": ""
  },
  "metadata": {
    "ocr_engine": "",
    "parse_version": "",
    "extraction_timestamp": "",
    "warnings": []
  }
}

Example filled JSON

{
  "document_type": "invoice",
  "confidence": 0.98,
  "source_file": "invoice_001.pdf",
  "language": "en",
  "page_count": 1,
  "invoice": {
    "invoice_number": "INV-10482",
    "invoice_date": "2026-01-15",
    "due_date": "2026-02-14",
    "purchase_order_number": "PO-77821",
    "currency": "USD",
    "payment_terms": "Net 30",
    "status": "unpaid"
  },
  "seller": {
    "name": "Acme Supplies Inc.",
    "tax_id": "12-3456789",
    "vat_id": "",
    "address": {
      "line1": "100 Market Street",
      "line2": "Suite 500",
      "city": "San Francisco",
      "state": "CA",
      "postal_code": "94105",
      "country": "US"
    },
    "phone": "+1-415-555-0100",
    "email": "billing@acme.com",
    "website": "https://acme.com"
  },
  "buyer": {
    "name": "Globex Corporation",
    "tax_id": "",
    "vat_id": "",
    "address": {
      "line1": "250 Park Avenue",
      "line2": "",
      "city": "New York",
      "state": "NY",
      "postal_code": "10177",
      "country": "US"
    }
  },
  "shipping": {
    "name": "Globex Corporation Warehouse",
    "address": {
      "line1": "88 Warehouse Road",
      "line2": "",
      "city": "Jersey City",
      "state": "NJ",
      "postal_code": "07305",
      "country": "US"
    },
    "method": "Ground",
    "tracking_number": "1Z999AA10123456784"
  },
  "line_items": [
    {
      "line_number": 1,
      "description": "Printer Paper A4",
      "quantity": 10,
      "unit_of_measure": "box",
      "unit_price": 12.5,
      "discount": 0.0,
      "tax_rate": 8.25,
      "tax_amount": 10.31,
      "line_total": 135.31
    }
  ],
  "summary": {
    "subtotal": 125.0,
    "discount_total": 0.0,
    "tax_total": 10.31,
    "shipping_total": 0.0,
    "other_charges": 0.0,
    "grand_total": 135.31,
    "amount_paid": 0.0,
    "amount_due": 135.31
  },
  "payment": {
    "method": "ACH",
    "bank_name": "First National Bank",
    "account_name": "Acme Supplies Inc.",
    "account_number": "****1234",
    "iban": "",
    "swift_bic": "",
    "reference": "INV-10482"
  },
  "metadata": {
    "ocr_engine": "example-ocr-v1",
    "parse_version": "1.0",
    "extraction_timestamp": "2026-01-15T10:30:00Z",
    "warnings": []
  }
}

If you want, I can also provide:

1. a minimal JSON schema
2. a JSON Schema (draft 2020-12) for validation
3. a Python/Pydantic model
4. a multi-invoice OCR output format