With wheels at the bottom and a camera on top, a cold machine nearly two meters tall roams around the store. When I stood in front of the ordering area at Bianlifeng, this robot walked straight toward me, 'forcing' me to make way, appearing somewhat 'arrogant.' 'This is our "director"; it's the one that gives me the most work,' a Bianlifeng clerk replied with slight irritation when asked about the machine's purpose. Rather than a director, it's more like a heartless monitor, watching everything in the store from the moment it opens. If you think it's just an ordinary surveillance camera, you're wrong—behind it lies Bianlifeng's algorithm-driven capability. As a representative of Bianlifeng's proud algorithm-driven approach, the application of big data and AI technology is mainly to help stores achieve standardized management, breaking through the constraints of talent management and refined supply chain operations that hinder the expansion of traditional convenience stores. Effectively using digitalization to transform convenience stores has also made Bianlifeng a 'new species' in the eyes of many investors. However, the algorithm-driven approach is a double-edged sword: while improving efficiency and refined management, it has also sparked considerable controversy, especially as the algorithm machine runs endlessly, intensifying conflicts with employees, ultimately leading to people fleeing, and Bianlifeng, which was supposed to provide convenience to consumers, instead faces a lack of 'convenience.' Around the Spring Festival holiday, many consumers complained about Bianlifeng's sloppy operations, which is a magnification of this contradiction. In this game between humans and algorithms, how should Bianlifeng make trade-offs?
Bianlifeng Employees Trapped in the Algorithm System Around the Spring Festival holiday, Bianlifeng's widespread out-of-stock items and sloppy operations were criticized by many consumers on social media. One netizen posted that the Bianlifeng store downstairs from their office first had a note on the glass door saying, 'One person working today, slow!' and 'If you're in a hurry, leave,' and later changed to opening at 11 a.m., with the note reading, 'One person working, no goods, preparing.'
When Leopard Change visited several Beijing Bianlifeng stores in mid-to-late February, it found that many drinks, breads, and snacks on the beverage shelves were out of stock, items were displayed messily, and price tags were misplaced under many products. Stores that should be open 24 hours either closed early or opened very late. These sloppy stores are mostly in a state of severe staff shortage. Stores that should have 3 to 4 employees **only have one employee. One clerk told Leopard Change that they had to take care of the store and had no energy to restock shelves. Many of the working employees are part-timers, including those from other stores on temporary assignment, who are not directly responsible for the store's operations, leading to a sharp decline in service quality. Although some Bianlifeng stores returned to normal service in March, these sloppy operations are not isolated incidents but occur from time to time. A very important reason behind this is that under algorithm control, Bianlifeng employees face high work intensity, having to handle numerous complex tasks assigned by the system and face an extremely high penalty rate, leading to persistently high employee turnover. The first conflict between algorithms and employees is a different perception of work saturation. Robots can work without rest, so Bianlifeng's algorithm system also requires people to work tirelessly like machines. 'The algorithm system was installed to make me work non-stop for these 12 hours, squeezing every bit of labor,' Li Hao, who works at Bianlifeng, told Leopard Change. 'The robot exists to supervise work.' Li Hao has worked at Bianlifeng for less than a month, managing a store of about 100 square meters, responsible for daily display and restocking, cleaning, inventory, and making meals. Due to a shortage of staff, the work that should be done by three people falls entirely on him. With insufficient staff, Li Hao's tasks become particularly heavy. Even when he's already busy, the Bianlifeng robot roams around, assigning tasks like restocking, removing items, and changing labels at any time. Employees are required to place products according to the automatic display equipment regulations and take photos to upload to the system. Facing the trivial tasks assigned by the algorithm, Li Hao said helplessly, 'I'm alone, having to get cigarettes, sell meals, and clean. How can I have time to put labels one by one? The system is well-intentioned, but it's not practical.'
If the ultra-high-intensity tasks arranged by the algorithm make employees feel exhausted, then the fines imposed by the algorithm are the last straw that breaks the camel's back. Speaking of his experience with fines, Li Hao couldn't help complaining, 'Sometimes I clearly uploaded the photo, but because the system had a bug and didn't upload successfully, I was fined.' Later, he tried to appeal to the system to get his money back, but the system didn't handle it, and he couldn't get his money back. 'In Beijing, you can earn 22 yuan per hour for a full work hour, with a night shift allowance of 10 yuan per hour. If you rest only four days a month, you can earn about 7,000 yuan a month. Employees sign labor contracts with third-party companies and don't have social insurance or housing fund,' Li Hao told Leopard Change. It seems like a decent income, but due to severe algorithm fines, his take-home pay often shrinks significantly. To reduce fines, Li Hao unplugged the robot and put it in the storage room. Although the system kept prompting him to charge and turn on the robot, he turned a deaf ear. 'This is the supervisor sent by the company. As long as it's there, I'll lose half of my daily wages.' But even so, Li Hao couldn't escape the system's surveillance. More than a dozen cameras in the store would capture whether his operations were compliant at any time. Once, when he was cooking oden and was a few minutes late, he was fined an hour's wages.
The algorithm, which was supposed to improve management efficiency and service quality, instead accelerated the departure of Bianlifeng employees. During the Spring Festival holiday, many stores had particularly severe staff shortages, and service quality plummeted. Stores that should have been managed by three people were left with only one person to maintain operations. To alleviate the labor shortage, Bianlifeng hired many part-time employees. The remaining employees gradually found ways to survive, taking advantage of the company's labor shortage to quietly resist the system. Li Hao's store manager was fined over 2,000 yuan in one month and directly resigned. The company, at a critical juncture of staff shortage, had to agree not to deduct the fines. Part-time employee Luo Qi also learned to slack off when filling in at other stores. Even when he saw empty shelves, he didn't restock; when price tags were misplaced, he was too lazy to fix them. Luo Qi said bluntly, 'It's not that I didn't see it; it's just not my store.' When asked what he would do if the robot deducted his pay, he said angrily, 'If it dares to deduct my money, I'll resign. Can they find anyone to replace me?'
Behind the Algorithm Supremacy 'When founding Bianlifeng, Zhuang Chenchao still sought answers in the details. He didn't follow trends but looked at logic. When he discovered that the convenience store business model could be driven by algorithms and no one had done it yet, he started his second venture,' Zhang Lei of Hillhouse Capital described Bianlifeng's founder Zhuang Chenchao in his book Value. Zhuang Chenchao is a math genius. When founding Qunar, he witnessed the magic of algorithm-driven approaches and also tried to use algorithms to transform the retail industry. In an interview with Qingteng's One Question, Zhuang Chenchao expressed his preference for algorithms. In 2018, the company spent a whole year moving toward a fully systematic model that completely trusts the system. Reflecting later, he felt 'this decision could have been made earlier... For Bianlifeng, no one can defeat the system.' Regarding its positioning, Bianlifeng, which relies on algorithms, publicly responded, 'On the surface, Bianlifeng is a chain convenience store, but in fact, we are a data technology company.' As a convenience store in traditional retail, it inherently has the defect of high labor costs, and the introduction of algorithms is to solve the 'people' problem. Through algorithm-driven management, standardizing every aspect of production and operation, reducing the interference of human factors, thereby achieving cost reduction, efficiency improvement, and accelerated expansion. According to third-party data, Bianlifeng currently has fewer than 3,000 stores. From its initial 5 stores in Beijing's Zhongguancun to 1,000 stores nationwide, Bianlifeng, which operates on a self-operated model, took only 2 years. In comparison, the established convenience store chain Lawson took 20 years to open 2,000 stores in mainland China through a franchise model. Relying on its algorithm system, Bianlifeng has carved out a new oasis in Beijing, the 'convenience store desert.' Product selection, store replenishment, pricing, and customer ordering and checkout at Bianlifeng stores are all intelligent, minimizing human involvement and relying more on algorithms to make decisions, making operations more online and systematic. This also makes employee training faster and more efficient. Zhuang Chenchao once said, a clerk can fully master a store in about 5 days, and a store manager takes about a month. This efficiency is much higher than the industry average. To optimize this algorithm system, Bianlifeng has poached many people from internet companies like Meituan and Baidu over the past few years, and even requires its employees to understand algorithms. In 2019, Bianlifeng made headlines for using math exams as a regular internal capability test. Zhuang Chenchao said, 'Convenience store business requires a large number of small decisions based on mathematical logic evaluation every day. If mathematical logic is good, performance may not exceed expectations, but if mathematical logic is poor, it's hard to achieve expected performance.' Another manifestation of his advocacy for algorithms is that Bianlifeng also leaves store location decisions to the algorithm system. When opening new stores, Bianlifeng completes site selection based on data analysis results from the algorithm system. The premise for algorithm operation is having massive data. To collect data, Bianlifeng has installed multiple surveillance cameras in stores, from entrance monitors to robots, all are tools for Bianlifeng to collect data, recording all actions of employees and consumers. In addition to inside the store, Bianlifeng has also been collecting data outside the store. According to Southern Weekly, since 2017, Bianlifeng has recruited a large number of information collectors to place cameras in different locations for recording, including residential communities, office buildings, and neighborhoods.
However, the extensive use of algorithms to replace people is a double-edged sword. In the view of new retail expert Bao Yuezhong, retail convenience store management is very complex. Although algorithms can save labor costs, they must also ensure basic operational needs. Generally, a store should have at least two employees to ensure loss prevention; one person is not appropriate from a configuration perspective. Algorithms are indeed a good thing for retail, but how to set the parameters and formulas of algorithms still requires people to do it.
Finding the Optimal Solution The idea of reshaping convenience store competitiveness through algorithms makes Bianlifeng unique in the convenience store industry.
Bianlifeng once set a '10,000-store plan' to be achieved by 2023, but under the 'hibernation plan' in 2022, there is still a long way to go to reach 10,000 stores. In addition, the 'Bumianhai' coffee and 'Wuqiongxiang' front warehouse, which Bianlifeng regarded as new growth points, also quickly declined after a high-profile launch. Bianlifeng explained that it was due to the pandemic, but the pandemic was only the trigger. During the company's rapid development, these system problems could be masked, but as Bianlifeng slowed down, the contradiction between employees and algorithms became more prominent. Employees even constantly escape from the system to get rid of algorithm control, and the system's operational efficiency is actually greatly reduced. Returning to the traditional sense, the main function of a convenience store is to provide shopping convenience, allowing hungry pedestrians to enjoy warm meals more quickly. One of its charms is the enthusiastic service provided by clerks. However, over-reliance on algorithms makes Bianlifeng appear colder compared to other convenience stores. Facing the lack of service warmth, many customers vote with their feet. On the Black Cat complaint platform, there are 1,559 complaints about Bianlifeng, a large part of which are about the system and service attitude, including 'poor service attitude' and 'pending orders cannot be canceled.' Employees who have been trapped in the algorithm cage for a long time become more emotional, which not only 'hurts' customers but also accelerates their departure. The empty shelves, chaotic store management, and even closures due to staff shortages around the Spring Festival this year are a concentrated resistance by employees against the algorithm.
On one hand, consumers are not buying it; on the other hand, employees are leaving. Under this dual pressure, the revenue of many stores has dropped significantly, and the decline in revenue will ultimately affect Bianlifeng's development. 'Actually, this store should have four employees. Previously, daily sales could reach 20,000 yuan. Now, with one person unable to restock, sales have dropped to over 10,000 yuan,' Li Hao told Leopard Change. The staff shortage caused by algorithms has already led to a significant decline in revenue for his store. The introduction of algorithms not only failed to help solve store operation problems but also created conflicts between employees and algorithms, and between stores and customers. For Bianlifeng, this algorithm system is not doing addition but has been doing subtraction that harms the company's interests. The algorithm system created by people should be used by people to empower work and improve efficiency. If people are made to obey algorithms, be dominated and exploited by them, then such algorithms will eventually be abandoned by people. Algorithms without people cannot independently complete business operations, and people without algorithms cannot make more scientific and efficient decisions. Only when the two cooperate can they produce a 1+1>2 effect. Now, for Bianlifeng, finding the optimal balance between algorithms and people has become more urgent. (At the request of the interviewees, the names in the article are pseudonyms.)