Image via Made Lapuerta
FORECASTERS HAVE LONG ANALYSED FASHION TRENDS, TRIED TO FIGURE OUT WHAT’S NEXT AND PREDICT CUSTOMER DEMAND. BUT A NEW WAVE OF TECH-POWERED STARTUPS IS CRUNCHING BIG DATA NUMBERS - COULD THEY MAKE FASHION MORE SUSTAINABLE? SOFTWARE ENGINEER MADE LAPUERTA THINKS SO.
Harvard graduate Made Lapuerta is the founder of Data, But Make It Fashion, otherwise known as Dashion - an online-magazine-cum-data-driven science project. “Dashion was born in the summer of 2019, while I was working as an engineering intern at Google,” she says. “I had recently learned about using image recognition models to detect objects in images, and thought it might be interesting to see if I could detect fashion trends from runway photos.
“From there, it was a bit of a snowball effect. I realized that by detecting trends in images, you could quickly and accurately aggregate how much of a certain collection displayed a particular trend. Knowing which trends objectively stood out more than others, I felt, was an interesting way to look at a runway collection each season. I come from a computer science and coding background, so it was easier for me to understand fashion, and what was in style, when it was explained in an objective, quantifiable way.”
Dashion is not the only fashion tech platform embracing new approaches to using data more intuitively. The Yes has made a splash by using data to figure out what clothes appeal to customers. The app uses algorithms to streamline the personal shopping ecommerce experience. Retail giants like Amazon have long been experimenting with A.I. for similar purposes, while sports giants stared betting on mass customisation half a decade ago. Sourcing Journal’s 2019 Personalization Report found 40 percent of surveyed brands were manufacturing closer to market and 32.6 percent shortening development timelines through digitization to better enable it - which could mean less waste and more sustainability.
Also at the big end of town, Google is partnering with WWF Sweden on a new platform that will help designers make more responsible choices around fabrics. According to The Evening Standard, “The platform will analyse the materials and their sources, taking into account factors such as water scarcity and pollution, as well as estimating impacts on greenhouse gas emissions, drawing on Google Earth Engine data.”
Data can also help tell us about trends that we’re experiencing in the moment. Analyzing Google’s Year in Search Data showed surprising trends that emerged in 2020, not just cottagecore but also “e-girl” style and “Dark Academia.” Meanwhile, machine learning is creating collections from scratch - hello Deep Vogue. Will data scientists be the new designers?
We asked LaPuerta some probing questions:
WARDROBE CRISIS: Can data really tell us anything about fashion? Surely, fashion is emotional, creative… human?
MADE LAPUERTA: “While not every aspect of such a creative and artistic field can be quantified, data can certainly tell us a lot more about fashion than I previously anticipated. Personally, I sometimes use the data I collect from runway shows to determine what to wear. For instance, if certain trends stick out more than others, I might consider pairing them together or forming outfits based on the biggest trends of a particular collection I liked.
“Aggregated in real-time, data can also be used to predict what consumers are demanding and looking for. On our Instagram page, when we post timely updates on up-and-coming trends, they often comes from Google Search Data, which is a fantastically accurate indicator of what consumers want.”
How does the process work? How do you use the data to analyze runway shows and new collections? MADE LAPUERTA: “If there’s an image recognition model trained to detect a certain trend or set of trends in an outfit, we aggregate runway data by passing images through it. Otherwise (and as is most common), we run code in Python to tag trends in every outfit and aggregate what percentage of a collection showcases each trend. This takes about two hours per runway collection.”
Do you think data can be used to make fashion more sustainable? “Definitely. This is where I believe technology in fashion can make a significant impact. By better predicting what consumers want to buy, brands can stop producing goods that are going to end up in landfill, or go unsold. [While we rightly focus on fast fashion as a major over-producer] this is also important in the high-fashion [sector], where unsold goods are rarely put on sale.
“With the data we aggregate for Dashion, my hope is to encourage people to notice how some trends stick out more than others, and adopt a ‘quality over quantity’ mentality. For example, in the Fall 2020 collections, some of the macro-trends were turtleneck sweaters, high-waisted pants, and open-toed shoes. So, I can build a more sustainable closet by investing more into these fewer staples, instead of chasing and purchasing new, fast-fashion trends every week.”
Tell us about the process of putting together a book. Why did you decide to take your project offline and into a physical book? “While I understand that ‘data’ might not initially sound very glamorous or fashionable, I wanted to present this nerdy, coding work we do for Dashion in a way that would be accessible to many. I wanted our readers and audience to understand the importance and the impact of the data we collect. So, I decided to create a coffee-table-like book of all this season’s trends. It was a wonderful challenge to make numbers and data fashionable and aesthetically pleasing.”
What is the next big fashion innovation? “Hopefully, an increase in sustainability efforts due to leveraging algorithms to better predict consumer demand. Wink wink.”
Interview with Claire Kalikman.
