This is an English adaptation of a FoodBud historical article originally published on June 24, 2022.
Nayuki's Tea had quietly begun installing its self-developed automated tea-making machines in stores by June 2022. According to FoodBud, the equipment could complete a tea drink in as little as under 10 seconds in the fastest case and lift production capacity by about 40%.
The machine had already reached its second-generation version. FoodBud described it as the first self-developed automated tea-making system in China's new-style tea drink sector to be deployed at scale by a brand. At the time, Nayuki had put the machines into use in about 120 stores in Guangzhou and Shenzhen, with a plan to roll them out fully in the third quarter.
Repeated Covid-related disruptions in China created major operating volatility for tea drink stores: stores opened and closed repeatedly, employees could be locked down at home, and long interruptions affected both sales and staff skill levels.
For Nayuki, automated tea machines were intended to address two pressure points:
In store, the machine connected tea barrels above and juice bottles below, with an electronic screen and QR-code scanner in the middle. After a customer ordered, the system generated a paper ticket with a QR code, while the screen showed options such as category, sweetness, and ice level. Staff attached the QR ticket to the cup, scanned it at the machine, selected the relevant buttons, and the machine dispensed liquid bases such as tea and juice through outlets below the screen. The process took around 10-plus seconds.
Ingredient ratios for different SKUs were stored in the cloud, allowing production through one-button operation. Although upfront equipment spending could weigh on short-term performance, Nayuki expected efficiency gains from reducing senior-staff training needs and cutting at least three months of new-employee training time.
FoodBud reported that the hardware cost of an automated unit was not high, at just over RMB10,000. The difficult part was software. Nayuki had a 200-person digitalization team, and its technology-team spending in 2021 exceeded RMB100 million.
The goal was to encode store SOPs into the system: once a customer placed an order, the order would be transmitted directly to the store and trigger production of the corresponding product.
Once automation was deployed broadly, stores would need fewer highly experienced staff, reducing training pressure. During pandemic disruptions, Nayuki could also use an automated scheduling system to reduce store staffing.
FoodBud described the intended staffing model as dynamic: a high-sales store might use 10 employees, while a low-sales store might use three. By tying labor hours to sales and dynamically adjusting full-time and part-time staffing, Nayuki expected the number of full-time employees to be cut by at least half. It also aimed to make raw-material costs more dynamic, using inputs according to actual sales and linking ordering plans to sales conditions.
Nayuki hoped to keep single-store labor costs below 20% in 2022 and companywide labor costs at around 25%.
After releasing its annual report, Nayuki had also communicated that, beyond automation, it planned to shift rent, labor, and procurement costs from rigid costs to dynamic costs.
For rent, Nayuki's previous store-rent structure combined minimum guaranteed rent with a revenue-based percentage. Because pandemic conditions were dynamic, Nayuki wanted to negotiate rent more as a pure percentage of revenue, such as 10% of sales, so that store profitability could be better controlled even when operations were affected.
FoodBud compared the direction with Luckin Coffee's store operations, which it had previously summarized.
Luckin's store workers operated under highly standardized rules: a cup of coffee had to be completed within two minutes; hands had to be washed before making coffee; each handwash had to last more than 20 seconds; towels had to be changed every half hour; disinfectant had to be changed every three hours; and used utensils had to soak in disinfectant for five minutes before being rinsed clean.
The so-called barista role was largely about memorizing the operating manual and knowing how many times to press machine buttons for different recipes. For example, a vanilla latte at half sugar required two button presses, while full sugar required four, followed by placing the cup under the machine. Luckin had many recipes, and FoodBud noted that some people left before memorizing them. After memorizing recipes, staff had to take an exam through the Luckin University app: two coffee names were randomly assigned, and the employee had to make both within five minutes and upload a video before being assigned to a store.
Luckin also had an intelligent order-allocation system. When customers ordered online, the order was not necessarily sent to the nearest store; if that store had too many orders at the time, the system could automatically assign it to the second-nearest store.
Luckin employees sometimes jokingly called themselves coffee button operators. The core model was to follow the operating manual and press machine buttons according to recipe, similar to assembly-line work. Luckin standardized complex processes, improved store output efficiency, and minimized labor and training costs.
By comparison, FoodBud noted that stores operated by Heytea and Nayuki had more complex production processes.
At Heytea, store work was divided into three lines: K, B, and S. The K line handled fruit cutting, tea brewing, and cheese foam preparation; the B line mixed tea drinks; and the S line handled cup finishing, cheese foam, and ice cream. Some store employees spent their shifts peeling grapes, picking strawberries, or cutting mangoes, peaches, waxberries, and other fruits used in drinks.
FoodBud's broader point was that as chains scale, the next development path is to improve efficiency with machines while also innovating on freshness-preservation and other supply-chain technologies. Consumers generally do not care whether grapes or other fruits are peeled by people or machines; the core issue is the quality of the fruit product.
Every market, after a period of extensive growth, gradually moves toward more refined operations.
Note: rollout timing and labor-cost targets were forward-looking statements from 2022 and should be read as historical guidance.