Handu Yishe ran an apparel brand as 300 three-person groups carrying their own losses

Zhao Yingguang spent ten years on trade assignment in Seoul, watching Korean e-commerce turn women's wear into its biggest online category, and came home to Jinan to found Handu Yishe in 2008. The obstacle was China's apparel supply chain, built for export runs of tens of thousands per style: when a co-founder brought a dozen sample garments to a large factory proposing orders of a few hundred pieces, he was shown the door before the box opened. Traditional retail worked the other way — pick styles at an October trade fair and gamble on a full-size run, absorbing the misses.

Handu's answer, later written up as a Harvard Business School case, was organizational: three-person groups in which one member selects styles, one builds the page and one manages inventory. The group keeps its profits and carries its losses, and its internal quota grows with results; roughly 300 groups ran like small bosses on one shared platform. That meant about 100 new styles a day — some 30,000 a year, with design to shelf in about 30 days. Every style started at 300–500 pieces; winners were restocked repeatedly off click, favorite and sales data, and misses were discounted in season.

The numbers compounded. Handu Yishe lost ¥37.5 million in 2014, turned a ¥33.9 million profit in 2015, and took the 2014 Singles' Day women's-wear crown with ¥198 million in a day, ¥90 million ahead of runner-up Uniqlo. By its 2016 NEEQ listing it had 2,600 employees, revenue of ¥1.43 billion and net profit of ¥88 million, with IDG, Yunfeng Capital and Star VC on the register — and its factories had learned the lesson too, shattering 50-person production lines into three-to-five-person cells to chase its small, fast orders.

Split buying into 300 three-person P&L groups

Losses are capped at the group level, so a wrong style costs one team's bonus rather than the brand's balance sheet — which is what makes 300 concurrent experiments affordable.

The reorder rule turns fashion from forecasting into measurement: small first runs priced to learn, then winners restocked within days, pushing sell-through to 95% against an industry norm of 50–60%.

The shared platform keeps the scale economies a three-person team could never reach — fabric, photography, warehousing, service — so tiny groups still trade like a national brand.

Quotas grow with demonstrated results, so internal capital flows to proven merchandisers instead of seniority; the mechanism is a stock market of shopkeepers.

Put inventory decisions where the market signal is: small teams carrying their own stock risk inside a shared platform replace one buyer's forecast with thousands of cheap bets, measured in days.

After the 2016 listing the company pushed for an A-share IPO, delisted from the NEEQ in 2018 to prepare, then had its listing tutoring quietly terminated in April 2020. Out of the capital markets' sightlines it dropped the 'internet brand incubator' story and its dozen-sub-brand ecosystem, retreating to being a design-capable, supply-chain-strong online apparel firm whose flagship stores still operate across the platforms — still owned by its founder, untroubled by distress, and no longer claiming to be anything more.