By LU Yibei, ZHANG Xiaoyi

The food and beverage industry is increasingly reliant on data to develop recipes and manage their operations. Bubble-tea-making robots are no novelty. But much more happens behind the scenes.

A creamy, sugary, liquor-laced latte may not appeal to coffee connoisseurs. But when coffee chain Luckin launched such drinks two weeks ago, 5.4 million cups were said to have been sold in a single day. Second-day sales have not been a matter of much publicity.

After having been almost brought down by an accounting fraud three years ago, Luckin is opening more stores, selling more coffee, and coming up with more internet sensations than ever before. Its coconut latte, spawned countless copycats in 2021. The “cheese latte” and “mint latte,” both launched last year, and each sold millions of cups on the first day.

Luckin recently opened its 10,000th store in China and does it all through Big Data. It’s not enough anymore to simply know that customers prefer sweet and creamy drinks. Product developers try more than a hundred sugary, milky recipes – 113 in 2021, 140 in 2022 – and make quick tweaks based on sales.

Digitalization is everything. Starbucks announced a new innovation and technology center in August and says it will spend 1.5 billion yuan on "intelligence" in the next three years. Mixue, a bubble tea chain, earmarked 270 million yuan of its IPO money for digitalization. Chabaidao, which raised 1 billion yuan this summer, is building an information system with some of the funds.

ZHOU Weiming of Luckin’s R&D department says product development is highly data-driven. There are five functions under R&D: market analysis, menus, optimization, recipes and testing. The teams test ideas through data collected from sales and operations and performance is evaluated based on highly quantifiable metrics – sales, basically.

“Our secret is not one system or one method. It’s a process of different teams working together, only possible through digitized operations. And innovation gives us more data of higher qualify, which generates further innovations,” he says.

But are they buying bubble tea?

Luckin has all its coffee machines connected to an internal network that sends alerts when smalles trouble. Similar practices are quickly going from novelty to norm. Chains seeking expansion rely heavily on data for recipe development, quality control and cost management. Bust mostly on where to open new stores.

“We used to send people to wait around the street corner for a whole day,” said Zhou. “Their report can be subjective. Someone might say, I saw many people on that street. But are they buying bubble tea there? It’s more helpful to look at sales of bubble tea shops in the neighborhood.”

Where's my data?

For some, digitalization came rather late. When Naixue’s current CTO He Gang took over in 2020, he was horrified that the company had no centralized database.

Information on inventory, sales, and store operations lived in separate spreadsheets. The headquarters had no idea which franchisee bought from what supplier until customers complained about “funny tastes.” When the quality assurance team tried to investigate, it was a nightmare.

It took Naixue two years to build centralized data. Since last year, the chain has been trying to automate its kitchens and cut labor costs from 33 percent of revenue in 2022 to 26 percent in the first half of this year.

Storm in a data base

Other beverage chains are doing the same but approaches vary. Hay Tea, whose recipes call for large amounts of fruit chunks, is inventing machines that can peel, pit, and slice fruit. A manager says machines not only save on salaries and staff training, but they generate accurate data for sales forecasts and inventory management.

In the age of AI, chains are constantly dreaming up even more intricate data maneuvers. Some want to predict sales with weather data. Some are talking about smart and flexible shipping. A data-tested, liquor-infused latte is only the first draft of the story.