Does the fashion industry still need trend forecasters?

For years trend forecasters held an unrivaled position in the fashion industry. Brands from the luxury heritage labels to global mass market giants looked to them for direction on the road ahead. Trend consultancy agencies such as Peclers Paris or Trend Union produced lavish tomes of seasonal inspiration, each page containing beautiful compositions of color palettes, fabric, concepts, often punctuated with physical print swatches, paint chips, threads and buttons, interspersed with photos of the stylish citizens of Tokyo, London, Berlin shot by the agency's roving reporters whose job was to travel the world on a healthy budget spotting trends. These books sold to brands for several thousand dollars.

But the arrival of AI-generative technology has given every designer the ability to conjure up any combination of styles, characters, and concepts on their computer screen with a few prompts while social media informs us in real time of underground movements, emerging subcultures and street styles happening anywhere in the world. This immediacy has led to a reported shortening of trend cycles, so by the time many have discovered something on social media it’s a thing of the past. Fashion thrives on looking ahead, on newness. Forecasting is its superpower. But how can mere forward-gazing humans compete with TikTok or the sheer amount of information that AI can process instantaneously? Is the job of the trend forecaster on the chopping block? To answer this FashionUnited spoke to Avihay Feld, CEO and co-founder of Browzwear, who has been a pioneer in the development of digital technologies that enable the fashion industry to thrive in an increasingly digitized world since the late 1990s, and Tessa Mansfield, Chief Creative Officer at trends and insights experts Stylus.

Feld believes that any brand still attempting to predict a trend 15 or 16 weeks ahead of time is entirely out of touch. “There used to be predictions, a prophecy that made itself truth and became a trend because many of the brands hooked into trend forecasting and were receiving the same sets of information,” he says. “These are going to be the colors, silhouettes, prints, and each brand handpicked some of it, so it was no surprise to find similar items in stores.”

It was a sort of self-fulfilling prophecy. But was it really what consumers wanted? Feld argues that with 60-70 percent of manufactured garments going unsold at the end of season, the answer has to be no. That was demonstration of a failed system. Even when those garments went on sale, a huge percentage still ended up in landfill. Says Feld, “You go to market with an order that includes millions of pieces per style and no one knows if it’s going to work." Yet the industry clung to this system. At least until technology prised loose its hold.

One answer, according to Feld, is the model used by digital native brands that “manufacture to trend.” This process involves scouring social media, particularly TikTok, and analyzing everything that users are sharing and posting, taking into consideration location, demographic, and from the data, distilling trends. “Image processing and deep learning has got to the point that it can actually understand not just silhouettes and colors but the textures of a garment down to the details,” he says. When the designers have honed in on the data, created their own version, the brand produces one physical sample to determine if the style works, after which it manufactures small quantities super quick, often using existing materials, thereby ensuring less garments are destined for landfill. “In less that 4 weeks the item is on the app, they can hit a trend in the middle of the season. If it doesn’t work, they drop it," says Feld. "The risk is small because they did few pieces. The ability to manufacture to trend differentiates them from regular fashion brands.”

Has AI placed the trend forecaster's job on the chopping block?

Mansfield doesn’t disagree that a combination of AI tools scraping social media could get quick results on trending fashion items. “While there's a place for AI generated social media-led product development, it's actually only about the ‘here and now’ and therefore limited to very immediate response fast fashion,” she says. These short-term trend seekers have the potential to cannibalize each other, fighting as they are over the same crumbs. Commonly known as “Social scraping” the process of automatically extracting data from TikTok, Instagram, Youtube, connotes the idea of removing surface particles or layers. And according to Mansfield it is exactly this inability to go deep that gives trend forecasting experts the edge over AI. ”Many nascent youth-led trends hitting that landscape aren't easily tracked through social scraping,” she says. “You might source the most mainstream ideas, but are unlikely to source the many more underground aesthetics and nuanced drivers.”

Technology has democratized every element of the industry, often eliminating the gatekeepers or those who claim to hold the key of knowledge. But the differing views of Feld and Mansfield illustrate that trend forecasting is not a one-size-fits-all solution but depends on specifics such as brand identity, market position, and customer expectation. And rather than occupying a diminished role, Mansfield says trend forecasters have become an invaluable support to industries seeking to adapt and successfully navigate and build strategies to compete in these times of enormous change.

The digital native brand has the ability to respond immediately to what people are wearing right now on the street. Feld estimates that such brands update their app with anywhere from 700 to 1000 pieces a day and shoppers come just to see what’s new. It’s a contemporary form of lunch-hour browsing or after-work window shopping. “It’s the fear of missing out on a daily basis,” he says. “But it’s s a completely different user experience from traditional fashion brands’ website where you get perhaps 6 updates a year.”

Fast fashion fixes versus trends with shelf life

Mansfield rejects the idea that all consumers are chasing fast fashion trends and even the idea that trend cycles are shortened. “The biggest trends are becoming ever more evolutionary, macro and longer in trajectory. At Stylus, our clients are looking to understand how to address rapidly shifting consumer demands (across work, active lifestyles, digital and IRL identity), and critically, how to respond to broader contexts, like the cost of living, the inclusivity imperative and sustainability crisis.”

Feld agrees that some brands set trends while others capitalize on them. And the digital natives that he refers to, the Sheins and Amazons of the world, have not built their reputation on sustainability. Often in the headlines for producing cheaply made garments that don’t factor in living wages or labor rights for the garment maker, they can arguably also promote a disposable culture around clothing due to the ease of their returns process. The landfill that was avoided at sampling stage might just show up later.

“It’s addictive,” he concedes. “But I’m not here to effect crowds of people not to eat sugar, I don’t know how to do that. As long as people buy it will continue.” He insists his message is not that trend forecasting is dead, only that it must evolve: “Setting a trend does not need 15 weeks. You might not be able to do it in 4 weeks, but the truth lies in between.”

Mansfield’s clients, she argues, some of the biggest global brands, have different expectations and the guidance she offers goes beyond product creation and even innovation to a view of the bigger picture and trends with shelf life. She sees the role of the forecaster as connecting the dots between emerging trends while looking at the full breadth of consumer lifestyles, and revealing opportunities for cross-industry collaboration and inspiration. 

“The forward-thinking brands and agencies we work with know that trends don’t exist in isolation," she says. "Most fashion brands still need to work 1.5/2.5 years out (especially in categories like active) and this timeframe can’t be well informed by scraping social for inspiration. Inspiration that can literally appear and disappear in a matter of weeks only tells you what's already happening.”