What online retailers got wrong about algorithms and AI
Around the time the COVID-19 pandemic took hold in 2020, a cluster of e-commerce, direct-to-consumer fashion, personal care and prepared meal kit companies were hailed as cutting-edge retailers. reinventing the online shopping experience by crunching customer behavior data.
2018 industry trade journal RetailDive.com said Lake Katrina “Disruptor of the Yearfor her role as founder and CEO of stitch correction, a fashion site offering a subscription service of products curated by 3,900 part-time stylists. In an article published in the Harvard Business Review around the same time, Lake described his company as “a data science operation,” whose revenue “depends on excellent recommendations from its algorithm.”
Stitch Fix was one of the most visible examples of the rise of so-called subscription box retailers. The list includes beauty product retailers birch, which “curates” and ships subscribers a collection of products based on past purchases and algorithms that rank consumers based on age, location, and other data points. blue aprona ready-meal subscription service, was another notable entrant.
At the start of 2021, three years after the company’s IPO, Stitch Fix’s market capitalization stood at $10 billion.
Now, just eighteen months later, the stock has lost around 95% of its value and the company is expected to post its first annual decline in sales since its IPO in 2017.
In the same way, blue apron has turned into an even uglier investment trainwreck — five years after its stock debuted at $140 a share, it’s trading under $4.
Why were the disruptors disrupted?
It turns out the warning signs were clear in 2018. In an article on Quartz.comLuis Perez-Breva, a lecturer and researcher at MIT’s School of Engineering, warned that “many retailers have forgotten what really helps customers: in-store support by human workers.”
According to Perez-Breva, “in order to receive clean data for machine learning (artificial intelligence or AI), for example, many retailers send customers questionnaires that are easier for computers to process.”
But, he says, “customers are not AIs. Most never respond to questionnaires and many fill out what they remember. This leaves retailers with… bad data.
Also in 2018, consulting giant McKinsey & Co. surveyed over 5,000 US consumers on subscription services and found that “churn rates are high (nearly 40%)…and consumers are quickly canceling services that don’t deliver superior end-to-end experiences.”
The McKinsey report concluded that “consumers do not have an inherent love of subscriptions. On the contrary, the obligation to subscribe to a recurring service slows down demand and makes it more difficult to acquire customers. »
Meanwhile, several academics have written about the risks associated with collecting data on individual buyers. It can be helpful for a consumer to have a retailer know their shoe size and preferred color. But what happens when the data collected by AI and algorithms includes the purchase of birth control pills?
For a long-time participant and observer of the retail industry, an old maxim comes to mind: the more things change, the more they stay the same. AI is a powerful tool in managing logistics, inventory, and a host of other business management issues. In the case of anticipating consumer behavior, some of them are useful, but only if they are used correctly.
If retailers want to know what consumers want, they have a proven way to find out – by testing products and pricing before committing valuable capital. Instead of analyzing data based on past behavior or “curating” profiles of consumer subgroups based on machine learning, retailers can more accurately predict trends and future demand using insights collected online in real time with real buyers. And, if you’re going to apply an algorithm, you better prove it works time and time again.