Skip to main content

Fashion’s AI Future: Will it Inspire or Inhibit Brands?

Retail may have entered the digital age more than two decades ago, but in 2020, technology is having an unprecedented impact on the way brands are doing business.

Advancements in artificial intelligence (AI) and machine learning aren’t just helping shoppers find what they want online or in stores—they’re changing the way retailers of all sizes curate their assortments, and the way brands design products.

Historical data can provide a useful roadmap for retailers. Understanding consumers’ demonstrated preferences—down to size, color and nuanced stylistic details—provides a compelling sketch for what a successful selling season might look like.

But some designers believe that removing the human element from the buying process could drain the color out of their collections. If stores are looking to the past to validate their future purchasing decisions, is there still room for innovation?

Fashion is about emotion

While the use of AI and algorithms has undoubtedly advanced operations across a breadth of industries, some designers are wary of its potential to restrict their artistry.

“Fashion is about emotion,” Cindy Traub, director of design for Rebels Footwear, said. “When you take people and product out of the equation, it’s not a good thing for creativity.”

Many department stores are removing flesh-and-blood buyers from their ranks, and making decisions about product assortment using sell-through data alone, Traub told Sourcing Journal.

“When you design into numbers instead of designing into a vision, it makes for a really flat market—and I think that’s what we have,” she said.

Related Stories

Retailers’ inhibitions are stifling brands and their creative teams, Traub argued. Designers know there’s little chance their most inventive designs will get picked up, and they’re being forced to re-work the styles that have driven sales in the past.

The result is a retail market that’s saturated with last season’s looks.

“Stores used to look to vendors to show them the way,” Traub said. Now, they’re looking to data. “Deals are made without product even in the room.”

Retail’s risk aversion is shutting out newcomers, making it difficult for burgeoning brands to find their footing in the mass market.

“It’s detrimental to the next generation of designers and people who want to be in fashion, because there’s such a barrier to entry now,” she said.

Esteemed partners like Nordstrom and Zappos are reticent to take chances on the new and untested, Traub said. That leaves brands on the brink of a mainstream breakthrough out in the cold.

Retailers’ need to see measurable success presents a paradox for brands. “You can’t meet the numbers without getting the opportunity,” Traub said.

Ultimately, she sees consumers driving some degree of decision-making back into the hands of experienced buyers. “Shoppers are not getting the breadth of choice that they want anymore,” she said. “Myself—I buy less. I can’t find anything that hasn’t been represented in my closet already.”

Traub believes her tepid enthusiasm is likely shared by today’s consumer. “If she’s already got it,” she said, “she’s not going to buy it again—and I don’t think that’s really being considered.”

Left brain meets right brain

While traditional fashion brands are being forced to adapt to data’s inescapable influence, others are harnessing it to their advantage.

“It’s kind of left brain meets right brain, engineering meets fashion, rational meets emotional,” said Aman Advani in describing his tech-driven performance wear company, Ministry of Supply.

The brand’s designs employ wearable technology that relies on AI and machine learning to determine a wearer’s needs.

“Instead of finding ways to make technology more wearable, we’re trying to make clothing more technical,” the co-founder and president said.

The Boston-based company also uses data to identify opportunities for innovation, allowing engineers and designers to create product with an eye toward consumer demand.

“The idea of market pull is important to us,” Advani said, referring to the need for solutions to specific problems, or fillers for gaps in the market. He has coined a method of analysis called “Quantified Empathy” to validate those opportunities.

Using a cross section of data from Ministry of Supply customers, people with similar profiles to customers, and people who have never purchased from the brand before, the company works to determine which products could potentially benefit and entice all three demographics.

“We consider this to be our secret sauce,” he said of the tactic, which he described as “focus groups on steroids.”

Sometimes, the initial idea for a product comes from a current customer or a Ministry of Supply employee. Then, it’s validated by the masses using Quantified Empathy.

“It drives us to understand what customer demand is before we spend a dollar, or an hour, on an idea,” Advani said.

In describing the process of creating the Mercury Intelligent Heated Jacket, the brand polled hundreds of respondents about their issues with outerwear, Advani said. “They would say, ‘I wear four layers, and I’m constantly putting things on or taking them off trying to find the perfect temperature, so we recognized that temperature modulation and regulation was a huge issue.”

Technology within the jacket gauges the wearer’s temperature against that of their surrounding environment, and also accounts for their rate of movement. Then, it determines the correct temperature for optimal comfort, and heats the jacket accordingly.

“You can either throw it on full power and blast heat, or you can utilize the machine learning,” Advani said. “Over time, the jacket gets better at predicting your overall heat profile the more you wear it.”

The brand’s use of consumer data and feedback to drive product development represents what Advani believes will be an enduring trend in the retail industry.

“Design used to be controlled by buyers who worked with designers to articulate their vision,” he said. “Now, that process has been democratized, and power has shifted to the consumer.”

While he sees that evolution as largely positive, Advani admitted that it won’t be without hiccups. When any trend emerges, he said, the industry naturally overcorrects in favor of it to appear ahead of schedule before eventually stabilizing.

“I think we’ll find some new normal, and it won’t be 100 percent data,” he said. “No matter how analytical we are about driving products that people actually want and need, we shouldn’t remove the opportunity for subjectivity and art.”

A crystal ball for the fashion industry

Brands may hold varied perspectives on data’s benefits and drawbacks, but trend-focused analytics tools stand to make an undeniable mark on the industry at large.

Stylumia, a four-year-old forecasting startup, provides brands, retailers and producers with data that helps identify trends, determine buy quantities and optimize retail distribution.

“The biggest problem facing fashion and lifestyle brands and retailers globally is demand prediction,” founder and CEO Ganesh Subramanian said.

Clients are stumped when faced with choosing designs, as well as deciding how many of each style to order.

Fast-fashion purveyors and luxury houses alike have long struggled to align demand with production, and have faced criticism for their penchant for overproducing. The practice results in margin-wrecking deep discounts and the landfilling of unwanted goods.

According to Subramanian, companies can develop smarter product planning strategies through predictive analytics and avoid those issues altogether.

“We predict winning trends for the future season, and demand for new and unseen product,” he said. “It’s a crystal ball for fashion industry.”

Stylumia helps brands map their target markets by collecting big data. The platform develops a “demand sense” of their relevant markets, and then provides clients with trend suggestions across genders, categories and other nuanced variables.

“This helps them choose ideas which have a high probability of winning,” Subramanian said.

A prediction model uses the brand’s own historical sales data as well as available market data.

The goal is to help brands curate an assortment with more top-selling products, and buy appropriate amounts of new product they’re seeking to try out for the first time.

“Any prediction is always from past data, irrespective of the domain,” Subramanian said. “Our human brain predicts our next step using all the past signals when we climb stairs.”

Stylumia identifies consumer patterns, he said, so that brands can develop smarter strategy around the risks they do decide to take. And he still believes there’s a place for people in the equation.

“We believe in a human and machine approach,” he said. “Machines learn from the data and domain expertise. Hence, the ‘gut’ instinct is an input into the modelling.”