Where a dairy's revenue and its cash part company

Chilled milk and yoghurt last between seven and twenty-one days, and most of our distributors also carry ambient products that keep for six months. When I became CFO, both were sold on the same forty-five to sixty days of credit, set by regional sales heads. I now run distributor credit as a lending book. A credit committee that I chair meets monthly, scores each distributor on payment history, sell-out data, stock checks and the owner's other businesses, and sets a limit and a term. Chilled products carry fifteen days, ambient thirty, and distributors who prepay receive a 1.5 per cent settlement discount. Expected credit losses are estimated by distributor score rather than by ageing alone. I also track channel stock in days of remaining shelf life, not days of sales, because a warehouse full of milk with four days left is a write-off already in progress.

E-commerce earned RMB 1.21 billion of net revenue for us in 2025, about 14 per cent of the total. Most of it now comes through our own livestream studio in Hangzhou and our stores on the large marketplaces; a smaller share comes from anchors paid on commission for product launches. I measure the channel from gross merchandise value down to contribution: refunds, coupons we fund, platform commissions, anchor commissions, fulfilment and the price-support claims our distributors make when an online price undercuts their shops. Refunds are estimated by product and session type from our own history and recognised when goods ship, as the revenue standard expects, rather than when customers ask for their money back. For chilled products a refund is usually a total loss, because the goods cannot travel back through the cold chain, so we price that risk before agreeing to any promotion.

A dairy company pays its customers in many ways: entry and display fees to supermarket chains, volume rebates to distributors, price support when a promotion runs, and free goods with larger orders. Under the revenue standard, money paid to a customer reduces revenue unless we receive a distinct service at fair value in return, so classification is a judgement made contract by contract. In 2021 we tightened that test with our auditors and moved RMB 214 million of supermarket fees from selling expenses into revenue deductions. Reported revenue and gross margin fell; profit did not change. Rebates are estimated as variable consideration each month from distributor volumes, and year-end accruals are reviewed with the Audit Committee before the results are approved. I report trade spend as a share of gross sales by channel, so the board can see when growth is being bought.

Raw milk is our largest cost. In 2025 about 45 per cent came from our eleven ranches, 42 per cent from contract farms and the rest, in milk-equivalent terms, from imported whole milk powder used in some ambient drinks and yoghurt bases. Contract farms sell to us under multi-year agreements; since 2025 most price their milk against a published regional raw-milk price, within a band. Imported powder is bought in US dollars, and we hedge committed purchases with forward contracts under a board-approved policy covering between 50 and 80 per cent of the next six months. When the renminbi weakened sharply against the dollar in 2022, the hedges protected that year's budget; they could not protect the following year's prices. Our dollar purchases fell from about USD 71 million in 2022 to USD 20 million in 2025, mostly because we used our own surplus milk first.

Chilled products sell only where there is cold space, so for many years we placed branded cabinets in convenience stores and community shops and paid part of their electricity. By 2019 we had 31,400 of them. Each cost about RMB 5,200 to buy and install, and with servicing, the electricity subsidy and write-offs of expired product, a cabinet needed roughly RMB 30,000 of annual sales to pay for itself. A quarter did not reach it. I built a model for every cabinet from distributor sell-out data, service records and the shop's location, and the sales teams used it to move 4,600 cabinets and withdraw 3,300. Annual sales per cabinet rose from RMB 38,000 to RMB 52,000 by 2021. The same discipline now applies to our doorstep-delivery depots, each of which carries its own profit and loss, including the cost of the bottles that never come back.