May 29, 2020
A new app, The Yes, claims to be the first to successfully merge computer vision, natural language processing and artificial intelligence (AI) to deliver truly personalized fashion recommendations.
“Unlike every other ecommerce site today, The Yes is adaptive, in session, which means as a shopper taps a Yes or No, the algorithm re-ranks products in real-time,” according to a press release.
The claim is gaining some merit because the app’s CEO is Julie Bornstein, was most recently COO at Stitch Fix. Previously, she was Sephora’s chief digital officer. Amit Aggarwal, the other co-founder, has experience at Google, Bing, Groupon and Bloomreach.
When first using the app, shoppers answer a few questions about their style, such as “What do you never want to see in your feed?” The app then produces a personal home feed with recommended brands and trends specifically for that user. Further “yes” or “no” clicks help the algorithm better understand the user’s taste.
At the same time, an extensive taxonomy covering each sales item fine-tunes recommendations. Ms. Bornstein told TechCrunch, “We leveraged both machine learning and computer vision to train models to understand how to absorb all pieces of data related to a product, as well as the image itself and how to read images. And it gave us a really strong understanding of 500 dimensions for every single item.”
The app launched with 145 brands, including Everlane, Gucci, Levi’s and Theory. Each brand’s full digital catalog is integrated into the algorithm.
The Yes, which takes a cut of each brand’s sales, uses each brand’s own imagery for its app listings. The platform does not hold any inventory. Brands ship the merchandise directly to the customer.
The app promises to fulfill the role department stores used to play in supporting discovery for consumers and customer acquisition for brands.
“You went to a department store because it offered something different from going to a brand’s own store,” Ms. Bornstein told Fast Company. “Our thesis is that the same is true in the digital world. There’s always going to be room for a multi-brand experience, as long as it’s a fun, seamless one.”
DISCUSSION QUESTIONS: What do you see as the technological keys to delivering truly personalized product recommendations in fashion or other categories? What do you expect the sales results of more personalized recommendations to look like for brands and retailers?