The overtime partners of the wage earners are behind a 2-billion-yuan retail business | Zhiliao

Author｜Huang Nan

Editor｜Yuan Silai

Recently, a new unmanned vending cabinet placed at the elevator entrance has become a regular stop for tea lover Li Pei and her colleagues during tea breaks and overtime.

It is only the size of a regular freezer, without a high-definition display screen. Instead, there is a QR code for placing orders through a mini-program. Its interior is also full of surprises. In addition to mineral water, tea beverages, coffee and other drinks, there are also instant noodles, spicy strips, bread, pastries, dairy products, etc., equivalent to a mini unmanned convenience store.

Compared with the traditional vending machine, which requires clicking on the vending screen, selecting the product and pressing the button, scanning the code or inserting coins to pay, and finally the product is pushed by the shelf and dropped, with a long waiting time and complicated procedures; today, in many first- and second-tier cities in China, in parks, commercial buildings, CBD offices and other places, the previous generation of vending machines is being replaced by intelligent vending cabinets.

The products it provides are more diverse, and the transaction chain is more direct and simple. Occasionally, when there is an impromptu meeting and it is too late to wait for takeout or go downstairs to the store, Li Pei always chooses to get an energy bar or bread from the vending cabinet. By scanning the face or code to open the cabinet door through the mini-program, you can choose the products at will after opening the door, and the payment is automatically settled when closing the door. It can even intelligently push you coupons for your favorite products according to your consumption preferences, which greatly meets the immediate consumption needs of consumers.

According to reports from the China Research and Popularization Institute and others, in 2021, the market size of unmanned retail stores in China was approximately 28.27 billion yuan, with a year-on-year growth of 24.6%; it is expected that by 2025, the market size of unmanned retail may approach 2 trillion yuan, covering a consumer group of 250 million people.

In order to meet the needs and preferences of users, in the absence of manual participation such as salespersons, how to understand the purchasing behavior of users and accurately provide personalized services through a small machine, AI plays an important role behind this. Through data analysis, optimizing the product layout and recommendation strategy has become a key issue for the operation and development of the current smart retail cabinets.

The "Intelligent Brain" Behind the Small Cabinet

The unmanned vending cabinet aims at the demand of instant retail.

In the era when takeout was not yet widely popular, convenience stores were the best choice for people's short-term choices. It takes about 15 to 20 minutes to buy a bottle of drink from the upstairs to the downstairs. Later, with the rise of takeout, the problem of limited product selection in space was solved, but the waiting time of users was also quietly lengthened. It takes an average of 40 minutes for a meal to be delivered after placing an order.

Suppose you only have 10 minutes of rest time. When you want to eat something to fill your stomach, but it takes too long to wait for takeout and it is a distance to go to the convenience store, how to choose? The intelligent vending cabinet can fully meet your needs.

When many users' impression of unmanned retail still stays at the vending machine by the stairs, the blind box machine at the airport or train station, and the direct-sale cabinets of several beverage manufacturers, as the smallest unit of the offline retail channel, the unmanned vending cabinet is becoming more and more intelligent. It can understand the preferences of consumers, predict future sales, and even "know" that you are tired of eating the same things and it is time to launch new products.

Feng e Zu Shi Intelligent Cabinet (Image Source/Official)

Feng e Zu Shi is one of the enterprises in the industry that earlier applied artificial intelligence technology to empower the growth of retail performance, mainly to solve the problems of market supply and demand prediction and precise matching of user needs.

Its intelligence consists of two parts. One part is the front-end intelligence, that is, for users, they can intuitively feel the friendly user experience of the machine, including the smooth use of the mini-program, the accurate recognition of the camera, and the convenient transaction process.

The second part comes from the intelligence on the back-end operation side. With the help of AI algorithms, through the analysis and prediction of user consumption behavior, the product structure and quantity are adjusted in advance, and the local consumption scene needs are decided in real time through the algorithm to help enterprises achieve stronger terminal competitiveness at the lowest cost.

For example, in the replenishment link, a direct change is that it recovers the decision-making power of replenishment that originally belonged to the front-line logistics personnel, and the AI intelligent algorithm replaces the replenishment staff to complete the work. The offline manpower only needs to quickly deliver the products that customers need according to the already planned replenishment route, and the decision-making part can be left to the algorithm.

The application of AI technology in unmanned retail is mainly in optimizing product strategies and improving operational efficiency. When extended to the field of intelligent cabinets, more applications are reflected in achieving advance stocking by accurately predicting market supply and demand. For example, by analyzing the historical sales data of the vending cabinet, following the "purchase based on sales" (determining the purchase quantity according to the sales demand), and replacing the traditional "sales based on purchase" (the purchase quantity determines the quantity of goods placed at the front end) purchasing model.

 "The most essential difference between the two is that sales based on purchase is to push the demand and push the products you want to promote to the users. On the contrary, purchase based on demand is initiated by the users. Users can find products that can solve their current needs and match their personal preferences in my vending cabinet." Zhu Tao, COO of Feng e Zu Shi, told Yingke.

Algorithm decision-making relies more on data. Only with accurate data and correct model can the decision-making quality be high. However, there are some data that the algorithm cannot know, which is an inevitable problem. Therefore, in the actual application process, it is necessary to comprehensively analyze the commodity circulation at the actual offline points, the collection of temporary needs and the historical sales data.

The advantage of this model is that on the basis of ensuring the authenticity of the data, this part of the data is fed back in real time. Users can directly interact with the customer service through the system background to provide real demand feedback, and then feed back to the recommendation algorithm for iterative upgrade to provide users with more accurate and timely product services.

At the same time, for a vending machine to meet the users' immediate needs throughout the day, a sufficient variety of products is also extremely critical. From beverages, snacks to instant foods, they need to be flexibly and reasonably combined.

"In actual operation, many enterprises will consider more about maximizing the sales of high-margin products rather than how to satisfy the consumption experience to the greatest extent. Many users have feedback to us that the first feeling when opening the Feng e Zu Shi cabinet is that the products we provide are very rich, and there are all kinds of categories, which really gives a feeling of a small supermarket." Zhu Tao said, "What we are thinking about is to solve your more comprehensive and subtle needs such as being hungry, thirsty, sleepy, etc."

Feng e Zu Shi Intelligent Cabinet Entering the Office (Image Source/Official)

The space capacity of the intelligent cabinet is fixed. In order to improve the space utilization rate of the intelligent cabinet, the R & D team of Feng e Zu Shi introduced an expansion model to optimize the space layout by combining different types of products and adjusting the product display. Combined with the differences in user needs in different regions, the combination of almost every vending cabinet is different, and the carrying capacity of its goods is also quite different. By constantly changing and updating products, it can increase the freshness and purchase desire while meeting the needs of users.

For example, in an office with thirty or forty people, considering the occasional meetings, overtime and other irregular hours, in addition to beverages, the vending cabinet will also be equipped with bread, instant noodles and other meals that can fill the stomach; while in parks, sports fields and other areas, tea drinks, water drinks, snacks, etc. are more popular.

 "Many times, even if selling only one product of water in this venue may be more profitable, based on the diversity of the needs of the entire scene crowd, it is decided that we cannot only sell water. Effective algorithm decision-making should not be a cold and rational information decision, but should be able to integrate the care for people and take into account the needs of other consumers to make the global optimal decision." Zhu Tao said.

It can be seen that at this time, the unmanned vending machine is an intelligent with temperature, which truly gives full play to the value of people and the maximum value of the algorithm. Based on the development of software algorithms under the integration of AI, the development of smart retail business is being pushed to a new stage.

Lower Costs Accelerate Point Expansion

The development of unmanned retail in China began in 2016. After Amazon Amazon Go (unmanned retail store) emerged, the shopping method of taking and eating immediately attracted the attention of the capital market, and major companies and retail giants entered the market one after another.

However, due to the continuous increase in operation and maintenance costs, the lukewarm sales volume, and the complex supply chain construction and regional management expenses, for a time, discussions about the failure and reshuffle of unmanned retail consumption were rampant.

With the reshuffle in recent years, the development of its business format has gradually entered a relatively mature stage, and each company generally turns to in-depth efforts in soft power such as AI algorithms.

Zhou Qi, an investor who has long been concerned about the retail field, told Yingke that the cost of a single unmanned vending machine commonly seen in the market is about 20,000 yuan. "Every link from each track inside, the motor behind, the delivery channel, and the order screen outside is money. The equipment also needs to be very large, and it is very difficult to move the venue or recycle it for the second time."

In contrast, the intelligent cabinet selected by Feng e Zu Shi has reduced a lot in terms of equipment cost. It is understood by Yingke that the current cost of unmanned retail intelligent cabinets in the market has dropped to several thousand yuan, which has more market advantages than the traditional vending machines.

"It has the same lifespan of about 7 years as the expensive equipment, but the most essential difference is that an intelligent solution can be applied to reduce operating costs." Zhu Tao said. With the self-service pickup method, users open the door and pick up the products and then place the order, which is not only more efficient, but also saves a lot of costs by reducing the mechanical structure, and the volume is also much smaller.

Feng e Zu Shi Intelligent Cabinet (Image Source/Official)

In terms of hardware integration, its vending machine adopts the method of freezer plus perception camera. The compressors, sensors, temperature controllers, etc. used in it have a large degree of repetition with traditional refrigerators and freezers. Some supply chain resources of the traditional manufacturing industry can provide it with a strong low-cost and large-scale mass production capacity, effectively reducing the overall cost of the unmanned vending machine.

Zhu Tao told Yingke that in the current field of intelligent cabinet retail, "Whether it is equipment, manpower or transportation capacity, it has more advantages to penetrate into every sub-scenario of retail. Ultimately, it reflects that we can be closer to consumers than others."

Up to now, Feng e Zu Shi has more than 140,000 points nationwide, and the service network covers more than 70 core cities across the country. It is the largest unmanned retail service provider in China, and it is expected that the total revenue in 2024 can reach 2 billion yuan.

In addition to reducing equipment costs, the direct impact brought by the AI algorithm effectively reduces the expense of manual maintenance.

Take Japan, where vending machines are highly popular, as an example. According to the data of the Japan Manufacturers Association, as of 2022, the number of vending machines in Japan has exceeded 6 million, with an average of one device for every 23 people; each replenishment staff maintains an average of more than ten devices per day. Their daily work is to go through all the devices regularly and replenish the goods in time when the devices are out of stock.

However, the per capita of Feng e Zu Shi equipment can be responsible for the operation of more than 100 vending cabinets per month. According to Zhu Tao, the backstage of the Feng e Zu Shi storeowner has introduced a series of algorithms, which can provide a customized route according to the equipment and out-of-stock situation of different storeowners. The storeowner only needs to pick up the goods according to the source of goods prompted by the backstage every day and replenish the goods according to the optimal route planning.

Yingke learned that among the current management and storeowner personnel of Feng e Zu Shi, many have a background as SF Express couriers and have rich experience in regional logistics operation, warehouse supply chain management, etc. With the gradual increase in the scale of points, the scale of its front-line development, logistics, sales and operation and maintenance personnel will further expand.

In the past few years, the unmanned retail market has been highly competitive, and many manufacturers and brands have been reshuffled in the extremely fierce competition. Of course, there are also enterprises such as Youbao, Nongfu Spring, and Genki Forest, who have scrambled to seize the market opportunity through point expansion in order to gain an advantageous position.

However, it needs to be clear that the competition for points is not only the competition for territory, but the key is whether the core needs of users can be met. With the diversification and personalization of user needs, a new battle line has been drawn, and unmanned retail is also constantly exploring the possibility of more scene expansions based on the integration of AI.