The Algorithmic Fashion Companion: Making Machine Learning Human (And Fashionable)
Zalando’s “Algorithmic Fashion Companion” utilizes AI to create a unique boutique custom-tailored for each one of Zalando’s millions of customers, while aiming to be a very human and organic experience. How do we do this?
While wrestling with algorithmic fashion recommendations that weren’t quite hitting the mark, we decided to conduct a Turing-like test to understand where our algorithms failed and how they compared to the fashion expert advice they were trained on. Through both this Turing-like test and live A/B tests, we’ve created principles for designing with data that can help guide anyone to create very human and organic experiences at a large scale. It’s tough to balance individual personal taste, trends, and general “rules” at this scale, but we’re well on our way to creating a unique boutique for each of our millions of customers. Most importantly, our learnings apply to other industries as well, such as education and healthcare. It’s a powerful time to be a designer, and we need more people to create human-centered AI for a better future.
My life mission is to create things, mainly digital or musical, that help people express themselves better. One way I’m currently doing this is as a Senior Product Designer at Zalando, Europe’s popular fashion e-commerce platform. Within the Personalization department, I aim to help Zalando’s millions of customers learn about how they want to express themselves through their daily fashion. One of my biggest problems to crack is how an “Algorithmic Fashion Companion” can utilize AI to create a unique boutique tailored for each one of Zalando’s millions of customers…and yet still be a very human and emotional experience.