How we sample revenue without drowning in invoices

Revenue testing fails when teams either pick random invoices with no plan or try to vouch every line. Neither approach respects the risk profile of a trading company with seasonal peaks.

We begin by reading the revenue streams: product sales, service retainers, and any government grants. Each stream gets its own risk note. High-volume retail sales may need analytical reviews plus a smaller voucher sample. Large project invoices need deeper contract reading.

Stratification matters. We sort the ledger by amount and isolate items above a coverage threshold. Those large items are usually tested in full. The remainder becomes a sample pool. Selection is documented so another auditor can re-perform the same picks.

For each selected invoice we want the customer order, shipping evidence or service acceptance, and cash receipt or receivable ageing support. If the company uses electronic invoicing common in Taiwan, we also check that the e-invoice number ties to the ledger entry.

Findings are written as facts first: missing delivery note, post-year-end cash without a receivable, or revenue recorded before performance. Recommendations come after the facts, not instead of them.