Everyone knows that Handu Yishe's success secret is the 'small group system', but few can truly explain how it formed and operates. This article starts with the small group system to help readers understand the path and secrets of Handu Yishe's success. Contributed by | Blue Ocean Method 100 (ID: weilan_100) Regarding the small group system, Zhao Yingguang said: "The 'small group system' appeared at the beginning of the venture, but it was constantly being explored and improved. It's like the household contract responsibility system after agricultural cooperatives, where responsibility is assigned to individuals, increasing employee enthusiasm and thus productivity." Part 1: Theoretical Basis of the Small Group System Zhao Yingguang: Our entire company is a pure internet enterprise, and we referenced Japan's Amoeba model. In the process of business model innovation, we selected the three core keywords from the Amoeba model: first, freedom; second, repeated division; third, passion. Using these three words as the core of our business model innovation, we reconstructed the entire business model. When we started this, Taobao had been around for nearly 6 years, and competition was fierce. In the women's clothing industry, conventional approaches definitely wouldn't work, so we kept thinking about how to differentiate. We summarized three 'fasts': fast learning, fast trial and error, and fast iteration. What form could best realize these three 'fasts'? So we thought in reverse: why are traditional enterprises slow? Because the bigger the enterprise, the more like a dinosaur. The product is fine, but where's the problem? In strategic and tactical decisions. In traditional bureaucracy, it's often the core executives and chairman who handle strategic decisions. Strategic decisions rarely go wrong; if you lack the level, you couldn't reach the entrepreneur level. But the problem lies in tactical decisions made by middle management and grassroots executors. The bigger the enterprise, the worse the grassroots execution, and the more powerless the middle management. Traditional clothing companies have three core departments: R&D, sales, and procurement. Usually, sales leads; once the sales strategy is set, products are coordinated with sales, and procurement implements it. What's the problem? If performance is good this year, whose credit is it? It's hard to say. So it's even more limited in mobilizing grassroots initiative. Traditional enterprise management structure is top-down, from boss to executives to middle management to the most basic employees—a positive triangle. We tried to manage with an inverted triangle model. We named the whole model the 'single-product full-process operation system with the small group system at its core.' The core is the product small group, consisting of three people. These three people manage product R&D and sales entirely, while all other public departments provide support. All small groups are placed at the top, and all public departments below support them. When designing our small group system, we had five considerations: first, to achieve full participation in management as much as possible; second, to calculate finely down to each employee; third, to have highly transparent operations; fourth, to combine top-down and bottom-up approaches; fifth, to cultivate more leaders within the enterprise. I think the key difference between leaders and non-leaders is decision-making ability. Those with decision-making ability are leaders; those with execution ability are more like employees and management. Part 2: Growth Trajectory of the Small Group System ♢ Small Group System 1.0: From Buyers to Buyer Small Groups Before 2007, Zhao Yingguang had been dabbling in e-commerce for several years and had opened physical stores, but he was still confused about the future. He kept exploring how to differentiate. It wasn't until mid-2007 that Zhao Yingguang came into contact with South Korea's largest fast-fashion company. Ordinary companies give manufacturers three or four styles, each produced in tens of thousands of units. This company directly gave manufacturers 700 styles. Although there were many styles, the order quantity per style was small—ranging from a few hundred to over a thousand units. If it sold well, they would reorder. Zhao Yingguang suddenly understood and immediately tried this multi-style, small-quantity model. Initially, the company had limited resources and could only do purchasing on behalf of customers. He focused on cultivating buyers, recruiting a group of students, pairing Korean language majors with fashion design majors. From 3,000 Korean clothing brands, they selected 1,000 and distributed them among 40 people. Each person picked 8 new items from the official websites of 25 brands daily, meaning 200 new styles per day. At that time, Taobao search was sorted by refresh time. Zhao Yingguang originally just wanted to make products abundant and fresh, but unexpectedly gained traffic. This quickly distinguished Handu from countless other Korean clothing stores online, and competitiveness suddenly increased. But this competitiveness mainly showed in competing for customers at the front end. Zhao Yingguang soon discovered problems with this model in the backend: First, purchasing on behalf had several major drawbacks, such as long waiting times, inability to return or exchange, frequent stockouts of colors and sizes, and poor cost-effectiveness. Second, style selectors lacked business and competitive awareness. After selectors uploaded new styles, whether customers ordered and how many sold had nothing to do with them. So Zhao Yingguang made adjustments: First, they shifted from 'purchasing products' to 'purchasing styles.' Buyers selected styles as before, then handed them to the production department to purchase samples, make prototypes, select materials, and find domestic factories for mass production. Second, they no longer required each person to track 25 brands; instead, they mixed everything up, creating competition among buyers and cultivating their independent business awareness. But new problems emerged: each buyer wanted to list more products but didn't pay attention to inventory issues. They only selected images and uploaded them, with little consideration for the supply chain. So Zhao Yingguang, on a trial basis, gave one buyer 20,000 yuan and let her decide production quantity, colors, and sizes. If profitable, the company and buyer would share the profits. This attempt also had problems: First, buyers came from design backgrounds; asking them to handle operations wasn't feasible. Second, even if a buyer had business talent, they had to select styles and handle business matters, making it hard to do both well. So Zhao Yingguang separated business affairs, not to the company's production department as before, but by assigning each buyer a visual person and an operations person. The prototype of the 'buyer small group' emerged. A few months later, the advantages of this small group + profit-sharing system began to show. Buyer small groups became more proactive. They could not only find the latest Korean fashion styles but also find relatively reliable contract factories to produce, reducing costs and controlling quality. Inventory turnover also sped up. Zhao Yingguang simply ran an internal experiment, setting up two teams: One team followed the traditional clothing company structure with three departments: design department, product page team, and a department for production coordination and order management. The other system broke up the three departments, taking one person from each to form a small group of three, totaling 10 groups. Both teams started simultaneously. After three months, the traditional team was shut down, and the company began using the more efficient and better-performing small group production model. Thus, the small group model took shape: buyer + visual person + operations person. ♢ Small Group System 2.0: Internal Resources Marketized, Everyone Is a Second Boss In 2011, Handu had 70 small groups. With more groups, resources that could previously be allocated couldn't be managed. For example, how to allocate internal promotion resources? Which group's products go on the store's homepage? Zhao Yingguang simply gave each group more autonomy: style selection, pricing, production quantity, and promotions were all decided by the group itself. Group commissions were calculated based on gross margin or capital turnover rate. Gross profit and inventory became the two indicators each group cared about most. Therefore, in Handu's Taobao store, there were no unified discounts or promotions; instead, each group made promotion decisions based on its own product situation to ensure gross margin and capital turnover. For homepage resources, they had an internal resource marketization mechanism: groups established for more than 6 months could bid for positions; those established for less than 6 months had dedicated homepage slots where they competed, first come, first served. Most importantly, financial power was fully delegated. Each group had free control over its capital quota, which was directly linked to sales: the more sold, the larger the quota. At Handu, this month's capital quota was 70% of last month's sales. For example, if a group sold 5 million yuan last month, 70% is 3.5 million yuan, so this month the group could use 3.5 million yuan for new orders. Therefore, each group had to have a strong sense of crisis. Suppose a group started with 50,000 yuan; it wouldn't use all 50,000 for orders because if products didn't sell, there would be no more quota, and the group would have to start selling inventory. If inventory never sold, the group would never get a quota again and might even die. What if it dies? Then it 'goes bankrupt' and 'restructures'. They ranked groups by category, with the top three receiving rewards and the bottom three being disbanded and restructured. Thus, each group was a competitive unit, almost a small company. This concept of making the company small was attempted by Kazuo Inamori and Zhang Ruimin, but Handu, leveraging its internet genes, moved further with light assets. This stage's mission was to solve internal resource allocation issues and marked the phase of comprehensive small-group restructuring of Handu's entire company architecture. If a product group felt the corresponding photography group wasn't good enough, they could switch; if they felt a certain production department group coordinated well, they would assign more tasks, giving that group more income and motivation. The entire organizational structure was like standard parts, freely connectable, ensuring most people's income was linked to market performance. ♢ Small Group System 3.0: For an Abnormal Sell-Through Rate From 2012 to 2013, Handu had over 200 small groups, 7 brands, and nearly 20,000 styles per year. What was the biggest headache at this stage? The supply chain! This required overall planning and precise single-product management. So the small group system evolved again. They created a single-product full-process operation system, and at the company level, they established a planning center to use the sell-through rate to push all chains to achieve single-style lifecycle management and coordinate the overall situation. So-called single-product operation means considering each style individually. From design to sales, all data is controlled. Each product's lifecycle is carefully maintained by dedicated personnel. On average, each group manages seven or eight styles per month, deciding what position each gets, what matching, how far to push for a hit, and at what inventory level to discount. With long-term practice, they become adept. The planning center, based on historical data, sets goals at the beginning of the year, referencing the annual peak and trough rhythm, then breaks them down to each group. Each group has detailed assessment indicators on monthly, quarterly, and annual bases. The planning department acts as Handu's NDRC and data center, coordinating competition among groups. The planning center's rhythm control is crucial for Handu's supply chain, allowing the production department and its factories to predict next steps in advance and prepare materials. Ordering tens of thousands of styles without rhythm control is pure suicide. Now Handu's sell-through rate can reach 95%, which is abnormal in the clothing industry, especially with 20,000 styles per year. According to Liu Jinggang, head of Handu's distribution department, achieving this indicator isn't stressful. To achieve this, Handu classifies products into: hot, popular, average, and slow. Hot and popular styles can be reordered (Handu's hot styles aren't like traditional companies' tens of thousands; selling 2,000 units is a hot style at Handu). Average and slow styles must be discounted immediately, and during peak sales times, a slight discount sells them out. By season's end, there's naturally little bad inventory needing clearance. This makes the supply chain more responsive and quality easier to control. Of course, this process was gradually explored and improved; without historical data accumulation, prediction wouldn't be possible. ♢ Small Group System X: Incubating 'Little Ants' Now, Zhao Yingguang is thinking about several questions: Can Handu's sub-brand array keep growing indefinitely? Where are its boundaries? How to break through when that day comes? In Handu's evolution, the core driving force has come from successive upgrades of the 'small group system.' But previous upgrades were 'inward.' Next, can they open up outward? Zhao Yingguang has always believed that many people have design ideas, but they often stay as ideas because testing the market is too costly. Who would find a factory to make samples and produce for a creative idea? Can Handu allow these 'external' designers to become Handu groups? Take T-shirt design as an example: users first choose a style, then design a pattern, then choose production quantity, such as 200 pieces at 20 yuan each, or 200-500 pieces at 19 yuan each, produced at Handu's partner factories. Next, Handu helps with sales, and the team provides daily sales reports. Throughout the process, Handu has two major profit points: one is service fees, and the other is identifying promising designers through operational data, allowing them to create 'small but beautiful' fashion brands through Handu's existing operation platform. If so, Handu might gradually transform from a clothing manufacturer into a fashion brand incubation platform. Zhao Yingguang calls this model the 'Fashion Cloud' platform. It's worth noting that because designers can choose their sales platforms, this model also involves e-commerce platforms like Tmall, JD.com, and Vipshop. From another angle, Handu might use one clothing brand after another to connect the entire internet industry. Coincidentally, Taobao has also opened a dedicated section for independent designers. Zhao Yingguang describes these small brands that will emerge in large numbers as little ants, which will eventually nibble away at the big players' markets. Part 3: Supporting Forces Behind the Small Group System ♢ The Small Group System Requires Clear Responsibilities, Rights, and Interests First, responsibilities: Each October, management sets the next year's production and sales plans with each group. They discuss with each group to determine sales targets, desired gross margins, and inventory turnover, then finalize them. That's responsibility. Second, rights: First, styles—the three members decide which styles to launch. Second, colors and sizes—they decide the colors, sizes, and inventory for each. Third, pricing—they decide together, with the company only providing a minimum markup standard to prevent unreasonably low prices. Third, interests: How bonuses are calculated. It's simple: bonus = sales × gross margin × commission coefficient. So each group can calculate daily how much they'll earn. Their profit and bonus aren't decided by the company; they earn and calculate it themselves. That's the entire responsibility, rights, and interests. ♢ Without an 'Elimination Mechanism,' How to Drive Group Optimization Since Handu Yishe's founding, there has been no designed elimination mechanism. Why not? Our groups update automatically. Every morning at 10, we publish yesterday's sales ranking. For example, if a brand has 20 groups, we publish the previous day's sales ranking every morning. With daily rankings, each group receives a strong stimulus, a very strong one. What results? The first place is excited and wants to maintain its position. The second place thinks, 'If I try a little, I can surpass the first,' and is also excited, constantly thinking about how to overtake. The last place wants to work hard to move up one spot at a time and not stay at the bottom. Every group works hard to move up in ranking. There's no overtime system; they voluntarily work overtime. That's one effect. Second, the group leader decides how to distribute the group's bonus. If the bonus is 20,000 yuan this month, how does the leader divide it? They might keep 10,000 and give 5,000 to each of the two members. After a year or two, the members might not want to work with the leader because they feel exploited. After a year or two, they mature and want to start their own group. In poorly performing groups, the leader might only get 2,000 yuan in commission and typically takes nothing, giving 1,000 to each member, which aligns with human nature—feeling sorry for the brothers, seeing others get 20,000 while they get 2,000, so they're embarrassed to take it. But how do members getting 1,000 yuan think? 'The leader is too stupid; I only get 1,000 with you,' and they don't want to work with them either. So one-person groups are allowed. If someone doesn't want to work with a leader, they can say so immediately and leave. If they can't find a partner, they can become a one-person group and do everything alone. But one person doing everything is too tiring, so they naturally find helpers to form a new group. This process continues; division and combination are the norm in our company. One more point: The company stipulates that if you leave your original leader, 10% of your commission for one year is automatically transferred to the original leader's account as a token of appreciation. This also makes the leaver feel justified: 'Even though I left, the company gives you 10% of my bonus, so I don't owe you.' The leader also feels better: 'I exploited you for two years and am embarrassed to continue, so go out and earn, and give me a small cut.' They then exploit newcomers, who don't mind. This cycle repeats—that's the ecosystem of small groups. ♢ Using 'Human Nature' to Drive Public Departments Our public departments are also very efficient. Why? How do they get motivated? Because human nature has two aspects: seeking benefit and avoiding harm. We incorporated avoiding harm into our system design. Since each group's responsibilities, rights, and interests are clear, if any public department infringes on a group's interests, the group will continuously complain. For example, if a driver is late picking up goods, the group leader will complain about him. So to reduce complaints, he works actively. This fully uses avoiding harm as a key source of motivation for our administrative departments. Now the company has over 280 product groups. How are they divided? All non-standardized processes are handled by groups; all standardized processes are handled by the company. We cultivate operations personnel with business thinking on the company's public platform. ♢ Disadvantages of the Small Group System First, cultivating buyers takes time. The master-apprentice model typically takes 2 years to mature, experiencing a full spring-summer-autumn-winter cycle to get a feel, and two cycles to become relatively mature, gradually improving. Initially, with only a few groups and small order quantities, no factory was willing to take orders, so cultivating the first batch of buyers was difficult. Second, buyer levels vary, leading to unstable product quality. A chief quality control officer is needed to strictly control it. Third, the multi-batch ordering model places high demands on the supply chain. Data integration with factories is necessary, requiring a management system, but none of the current ERP systems on the market suit the buyer small group system. Editor's note: The above is the behind-the-scenes story of Handu Yishe's 'small group system' that everyone knows. Below is Zhao Yingguang's ambition for the future. Zhao Yingguang: Times have changed. Ten years later, the era of 'crossing the river by feeling the stones' is gone. For e-commerce enterprises to develop, they must follow internet laws and deduce forward with logic that fits the times. The future has arrived, and it's already wildly popular. What should the next decade look like? Over the past decade, Handu Yishe has been upgrading and innovating. In the next decade, the challenge of innovation and upgrading is even greater. It's a big future full of temptation and imagination. We hope to contribute to the big e-commerce ecosystem, find the best solution for 'small but beautiful,' participate in creating a new business civilization, and enjoy it. 121~ I've already started running.