cottonwood-development/synthetic-invoice-sample

Invoice 500 — Free Sample (50 records)

This is a free 50-record sample of the full 500-record commercial dataset. Every record in this sample is a real, valid invoice record — real records, just the data. you'd build by hand.

What's in this sample

• 50 records in records.jsonl (NDJSON, one record per line)
• Schema-validated: every record parses against the real industry-standard schema before publishing
• Ready to use: import directly into any invoice pipeline — no transformation step needed
• Format: standard JSON Lines (.jsonl), invoice-native structure

Why these records hold up

Most synthetic datasets are "good enough" — they look like the right shape, but fail against real-world parsers. Cottonwood Development's datasets are generated and validated against the actual schema:

• Healthcare records validate as valid FHIR R4 Patient resources
• Banking messages parse against the ISO 20022 pacs.008.001.10 XSD
• CRM records maintain referential integrity (customer → orders → line items)
• Invoice records have internally-consistent totals (subtotal + tax = total)

Every record passes schema validation before publishing. If it doesn't parse against the real spec, it doesn't ship.

Sample record

{
 "invoice_id": "INV-2025-000001",
 "vendor": {
 "name": "Tech Solutions Inc.",
 "address": "456 Tech Lane, San Francisco, CA 94107",
 "tax_id": "CA-12345678"
 },
 "customer": {
 "name": "ABC Corp",
 "billing_address": "123 Business Rd, Los Angeles, CA 90001"
 },
 "invoice_date": "2025-03-15",
 "due_date": "2025-04-01",
 "payment_terms": "Net 15",
 "line_items": [
 {
 "description": "Consulting Services - Strategy Planning",
 "quantity": 2,
 "unit_price": 300.0,
 "total": 600.0
 },
 {
 "description": "Project Management Support",
 "quantity": 5,
 "unit_price": 150.0,
 "total": 750.0
 }
 ],
 "subtotal": 1350.0,
 "tax_rate": 0.0825,
 "tax_amount": 111.38,
 "total_amount": 1461.38,
 "currency": "USD",
 "status": "pending"
}

Schema

The full dataset uses the following field structure:

Field	Type	Description
invoice_id	string	Sequential invoice ID (e.g. INV-2025-000001)
vendor	object	name, address, tax_id
customer	object	name, billing_address
invoice_date	date	ISO 8601 date issued
due_date	date	ISO 8601 date due
payment_terms	string	"Net 15", "Net 30", "Due on receipt", etc.
line_items	array	description, quantity, unit_price, total
subtotal	decimal	Sum of line items before tax
tax_rate	decimal	Tax rate as decimal (0.0825 = 8.25%)
tax_amount	decimal	subtotal * tax_rate
total_amount	decimal	subtotal + tax_amount
currency	string	ISO 4217 currency code
status	string	"paid", "pending", "overdue", etc.

Use cases

• Test fixtures for accounts payable / invoice processing systems
• Demo data for AP automation platforms
• OCR training data for invoice extraction
• ETL pipeline testing for invoice ingestion
• ML training for invoice categorization, vendor matching

Get the full 500-record commercial dataset

This is a 50-record sample. The full commercial dataset contains 500 records for $17, Schema-validated, with full Vendor invoices with internal-consistency validation (subtotal+tax=total).

Buy on Gumroad: https://bluecore8.gumroad.com/l/cottonwood-invoice-500-records

About Cottonwood

A one-person data shop generating Schema-validated synthetic datasets for AI training, ML eval, and application testing. Every product is generated against an industry-standard schema and validated against the real parser before publishing. 100% synthetic, no real customer data, no legal overhead.

License

This sample is released under CC0 1.0 (public domain dedication). You can use it for any purpose, including commercial use, without attribution.

The full commercial dataset is sold under a separate license (see Gumroad product page for terms).