One pastry, one price, one honest claim

In July 2024 the agency representing one of the country's largest livestream hosts offered our classic twelve-piece mooncake box a place in a Mid-Autumn session. The terms were RMB99 a box, against a shop price of RMB168 and a member price of RMB148, together with a commitment that no channel would sell the box for less during the season. The agency projected 220,000 boxes, roughly RMB21.8 million in one evening. Our franchise partners bought the same box from us at RMB104.

Our head of e-commerce argued for acceptance, and the minority investor's representative saw an audience we could not otherwise reach. I took the opposite view. Members would watch strangers pay a third less than they did, franchise partners would face a public price below their cost, and the clause would constrain our own pricing for the rest of the season. Jingrou Pan accepted my recommendation after I set out exactly what we would give up.

We declined. Instead, we designed an eight-piece box with smaller pastries and two flavours not sold in shops, priced at RMB98, and offered it only through our own hosts and three mid-sized creators in Jiangxi, Hunan and Fujian. Every creator session carried advertising disclosure. Franchise partners received the pricing plan two weeks before it was announced.

The eight-piece box sold 96,000 units, about RMB9.4 million, well under half the agency's projection. The classic box held its member price throughout the season, and franchise sell-through of that box reached 91 percent against 84 percent a year earlier. Season revenue rose 4 percent, below the 9 percent plan the board had approved.

The episode produced written channel-pricing rules, which the board now reviews each June. I would do two things differently. I would design the live-selling assortment in the spring rather than in response to an offer. I would also test a smaller external session with different contents, because I still do not know how many lasting customers that audience might have brought us.

Early in 2023 our product team completed a walnut-cake and flaky-pastry range containing 18 percent less sugar than our classic recipes. The working name was Light Sweet, the packaging carried a reduced-sugar statement, and our sales team had agreed spring listings for five items with a regional supermarket chain.

Our quality and regulatory manager advised that a comparative sugar claim under China's nutrition-labelling rules requires a larger difference from the reference food than we had achieved, and that the working name would be read as the same claim. The product director wanted to launch as planned and correct the wording later. Delay meant losing the listing slot and the buyer's goodwill.

I recommended launching on schedule with no sugar claim at all. We renamed the range after its osmanthus and walnut flavours, removed the reduced-sugar statement and reprinted packaging at a cost of RMB380,000. Our hosts and contracted creators received a short list of phrases they could not use, including low sugar, sugar-free and any suggestion that the pastries suited people managing blood sugar, because an ordinary food may not claim a health function. Shop colleagues received the same guidance with a one-page answer for customers who asked.

The range sold 58 percent of its six-month forecast, and the supermarket chain delisted two of the five items. Part of the shortfall probably reflected the missing claim; part reflected flavours that were less popular than we had expected. The product team reformulated, and the 2024 version contained 27 percent less sugar than the classic recipes. It now carries a comparative claim that names the reference recipe, and its first-year sales exceeded the original forecast.

The lasting change was procedural. Claims now pass regulatory review at concept stage, before names, listings or packaging are agreed. I would have insisted on that review before the sales team discussed listings in 2023, sparing the buyer, the bakery and our own colleagues a correction that should never have been needed.

Since 2019 every active member had received a monthly coupon worth RMB15 on purchases of RMB80 or more. By late 2024 coupons were used in 43 percent of member transactions, and purchases clustered in the week after each issue. Regional chains and new discount bakery outlets had made price cuts a weekly event, and our coupon had become part of that contest rather than a reward for loyalty. Its annual cost had reached RMB31 million, and franchise partners funded half the value in their own shops.

I proposed ending the coupon from April 2025 and replacing it with benefits suited to fresh pastry: items reserved online and held for collection until eight in the evening, a birthday pastry, points exchanged for products, and two member price days before the Spring and Mid-Autumn festivals. The operations director expected footfall to fall, and four franchise partners wrote to oppose the change. The chief financial officer supported it. Jingrou Pan approved it on condition that we reported monthly to the executive committee and could reverse it within two quarters.

We gave members six weeks' notice in shops and online and kept paper vouchers for those who had not moved to the app. Franchise partners now fund only the birthday pastry. Member transactions fell 8 percent in the second quarter of 2025 and had recovered to 2 percent below the prior year by the fourth. Active members declined from 1.24 million to 1.10 million by June 2026. Annualised coupon expense fell to about RMB12 million, and gross margin on member sales improved by 2.6 percentage points.

The board now sees member health reported as active members, repeat rate and margin together rather than as registrations. I would change two things. I would trial the change first in two county towns, where partners carried more of the risk and their objections deserved evidence. And I would have spoken with older members before deciding, rather than learning of their difficulties through complaints.