Revenue in a China Due Diligence: What to Test and What to Ask For

Revenue carries most of the value in the multiple being paid, and it is the line most sensitive to a change of ownership. Cost adjustments move the number gradually. A customer relationship that does not transfer can remove a large share of earnings at once.

So the question is not whether the sales are real. It is whether they are complete, whether they are recorded at the right time, and whether they continue after the founder leaves.

Start with the three-way reconciliation

Every transaction leaves three marks: it is shipped, it is invoiced, and it is recognised as revenue. In a clean business the three line up within a few days. Where they do not, the gap is the interesting part.

Record	What it proves	What it does not
Delivery and logistics records	That goods moved, and when	Whether the customer accepted them
Sales invoices	What was billed, and in which period	Whether the amount will be collected
Revenue ledger	What management recognised	Whether recognition matched delivery or billing

Run the reconciliation on the last two months of the period rather than the year. Cut-off judgements are concentrated at period end, and a two-month window makes the pattern visible without requiring a full-year dataset.

Contract terms that move revenue

Four clauses deserve reading in the original contract, not in a summary:

• Acceptance. Whether revenue is contingent on the customer commissioning, testing, or signing off the product.
• Return rights. Whether returns are permitted, on what grounds, and how they have been exercised historically.
• Consignment. Whether goods held by the customer remain the company's inventory until sold on.
• Bill-and-hold. Whether goods are invoiced but stored on the company's premises — a legitimate arrangement, and one that needs both a substantive reason and evidence of control transfer.

Where returns exist, the historical return rate should be applied against the current period, not accepted as immaterial. A return provision that has stayed flat while the customer base has changed is worth questioning.

Channel: sell-in versus sell-through

This is the test most often skipped, and it changes what the revenue number means.

Where a target sells to distributors, reported revenue measures what was shipped to the channel, not what the channel sold to end users. If channel inventory is building, the reported figure is running ahead of demand, and the correction arrives after closing.

Three practical questions resolve it: how many distributors are there, what are their payment terms, and can the target produce sell-through data. Where a distributor takes title and carries its own risk on normal commercial terms, sell-in is defensible. Where the distributor returns unsold stock, or is paid a guaranteed margin, the economics are closer to consignment than to a sale.

Customer concentration and retention

Ask for the top twenty customers by revenue for each of the last three years, then look at the movement rather than the totals. A base that is stable because the same customers keep buying is different from one that is stable because new customers replace churned ones at similar value.

The follow-up question is why each major customer buys from this company. If the answer is price, the relationship is exposed to a competitor's pricing. If the answer is a personal relationship with the founder, it is exposed to the founder's departure. Both are answerable, but they price differently.

Price realisation

Headline price is rarely realised price. Discounts, volume rebates, year-end incentives and side letters all reduce what is collected, and some of them are only documented in correspondence.

The test is arithmetic: take gross invoiced value for the top customers, subtract every credit note and rebate in the period, and compare the result with the revenue recorded. A gap that is not explained by returns should be explained by something.

Turning revenue into cash

Matching bank receipts to invoices for the top customers over twelve months is unglamorous, and it resolves more questions than any analytical procedure — existence, timing and price realisation at once.

Two patterns are worth measuring: the lag between invoice and receipt, and whether the lag is longer at period end than during the year. A long cycle is normal in many sectors; what matters is whether it is stable or lengthening. A lag that extends materially is the pattern worth asking about, because it points to a change in payment behaviour rather than to the cycle itself.

When revenue may be understated

Revenue can also be recorded below the economic level. Where sales are routed through a group company that sits outside the perimeter being acquired, or where part of the activity is invoiced by another entity, the accounts understate what is actually being bought. This matters as much as overstatement: it changes the scope of the transaction, and it usually means the perimeter itself needs to be redefined before closing.

What to request

• Sales sub-ledger for three years, by customer and product
• Contracts and side letters for the top twenty customers
• Delivery and logistics records for the last two months of each period
• Credit note and rebate listings, reconciled to revenue
• Tax filing summaries, for comparison with recorded revenue
• Bank statements covering customer receipts
• Channel inventory data, where distributors exist

The constraint in most processes is not the analysis. It is whether these records are produced before exclusivity expires, and in a form that can be tied to the ledger. Establishing that in the first week is worth more than any single test.

This article describes a general approach. It is not accounting, tax, or legal advice for a specific transaction. Figures and examples are illustrative. Views are my own and do not represent any firm.