The livestream price is heard in every shop

In the autumn of 2022 a regional sales manager in Anhui sent me a photograph. It showed a supermarket shelf in Bengbu with a twenty-four-carton case of our ambient milk priced at RMB 69. Beside it the shop owner was holding up his phone, which showed the same case selling for RMB 49.9 in a livestream the night before. He had bought his stock from our distributor at RMB 58. He wanted to know which price was real.

I kept that photograph on my desk for a year, because it explained something our e-commerce reports did not. Livestream selling has become one of the largest channels for Chinese consumer brands, and for most of us it began as marketing. The discount, the anchor's commission and the fixed fee for a place in the running order were budgeted as the cost of reaching younger households, usually under the marketing director. That framing made sense when online sales were small. It no longer does for any company that also sells through distributors and shops. A livestream price is a published price. The shop owner, the distributor and the supermarket buyer see it as quickly as the consumer does, and it resets what each of them believes the product is worth.

One evening, six provinces

Cuiling sells in five provinces and Shanghai. Most of our ambient milk and much of our yoghurt reach shops through distributors, who buy from us at a list price, add a margin and sell to supermarkets, convenience stores and community shops, each of which adds its own. Every link in that chain depends on a shared understanding of the shelf price, because their margins are carved out of it.

When a large anchor sold our products below that shelf price, the evening's audience was national, but the consequences were local. Shop owners asked distributors for price support. Distributors passed the request to our regional teams, who usually granted it, because a distributor who loses money on our products starts to favour someone else's. Supermarket buyers invoked the lowest-price clauses in their contracts. In 2022 we paid RMB 37 million in price-support claims that could be traced to specific livestream sessions, and in the month after a large session distributor orders for the promoted product typically fell by 10 to 15 per cent while shops sold through stock that had suddenly become expensive.

None of this appeared in the e-commerce budget. It appeared as trade spend in regional sales accounts, or as weaker orders that the sales teams explained by the weather or the season.

What the commission statement leaves out

The e-commerce report itself was flattering in a different way. It opened with gross merchandise value: RMB 1.62 billion in 2022. Below that sat a long series of deductions. Refunds on sessions with the large anchors ran at 21 per cent, against 6 per cent on our own sessions. Coupons we funded reduced the price further. Anchors' commissions of 18 to 25 per cent, fixed placement fees of up to RMB 300,000 a session, platform charges and fulfilment followed. After all of it, the channel lost RMB 96 million.

Refunds deserve particular attention in dairy. A refunded case of ambient milk can sometimes be resold. A refunded box of chilled yoghurt almost never can, because it cannot travel back through the cold chain, so the refund is a total loss of product, packaging and delivery. Many consumer companies recognise revenue when goods ship and adjust for refunds as customers claim them. The revenue standard expects expected refunds to be estimated at the point of sale. Recognising them later makes the quarter of the session look better and the following quarter worse, and it hides the real price of a promotion from the people deciding whether to run it.

When we rebuilt the report, we started with the lowest price any channel saw for each product in the period and worked outwards: what that price cost us online, what it cost us in price support offline and what it did to distributor orders. The view was uncomfortable for everyone, including finance, because we had approved most of those sessions.

A floor, a studio and a smaller number

From January 2023 we made three changes. First, a floor: no online price for a product sold in shops may fall below the lowest promotional shelf price we fund offline in the same month, after coupons. Second, no fixed placement fees. Anchors are paid on commission only, and we use them mainly for new products, where trial matters and there is no shelf price to disturb. Third, we built our own studio in Hangzhou, with fourteen hosts broadcasting six hours a day, selling at the same floor as everyone else.

The cost was immediate. Gross merchandise value fell to RMB 1.08 billion in 2023, and the share of first-time buyers in our online sales fell from 41 to 29 per cent. Our head of e-commerce had warned that we would lose younger city households who never visit the shops our distributors serve, and some of them we did lose. Two analysts asked whether we had abandoned the channel.

By 2025 the picture had changed. Gross merchandise value was RMB 1.43 billion and net revenue RMB 1.21 billion. The channel made a contribution of RMB 58 million. Price-support claims fell to RMB 4 million in 2024. Online and regional teams now plan promotions on one calendar, against one floor, and distributors no longer learn our prices from their customers' phones.

I do not argue that every company should leave the large anchors. A brand with no shops to protect, or one launching a product nobody yet knows, may find them the cheapest route to trial. My argument is narrower: when a company sells through more than one channel, the livestream price belongs to the pricing decision, not to the marketing budget, and the CFO should own its floor.

For directors the questions are simple. What was the lowest price any channel paid for our largest products last quarter? Who paid for the gap between that price and the shelf price, and in which line of the accounts? And are refunds estimated when the goods ship, or only when customers complain? A management team that can answer all three has a pricing policy. One that answers with gross merchandise value has a marketing campaign.