Can data help predict the fashion future?
This is an Interesting article in the NY TImes on whether data can really help predict the next big fashion trends. Personally, I’m unsure that data or sophisticated algorithms could predict the emergence of a trend as ridiculous as this…
The fashion industry always wants to know about the next big thing, so Julia Fowler thinks it should take a closer look at the world’s financial markets.
Just as old-style traders have been all but replaced by analysts crunching vast amounts of data, Ms. Fowler said she believed that understanding fashion trends should be less about intuition and more about real or near-real time information.
Trained as a designer, Ms. Fowler established the fashion forecasting business Editd in 2009 with Geoff Watts, a programmer with a background in financial modeling. “Industries like the financial sector have used big data for many years,” she said. “The logical step for us was to apply a scientific approach to the apparel industry.”
Editd has 22 employees at its office in the Silicon Roundabout, an area of East London now known as a hub for tech innovation. Each work day Editd’s software gathers online information for a huge variety of garments and accessories and amasses 300,000 comments from social media ranging from what’s on store racks to indications about how long the passion for leopard print will last.
The information is transformed into data, compiled and repackaged into analyses that illustrate competitors’ product assortments, pricing, consumer mood and emerging trends for clients that include Asos, Gap and Target. (Editd’s fees begin at $2,500 a month for a small retailer in a single market, but rise sharply for larger clients who want more complex services.)
For example, one customer wanted to know about biker jackets, the kind of fitted, waist-length outerwear that brings to mind James Dean, or Marlon Brando in “The Wild One.”
A report detailing sales at British and U.S. premium retailers over a three-month period shows Selfridges stocked the greatest variety (41 kinds), while McQ, Alexander McQueen’s secondary line, was the most popular brand. It also indicates that just 3.1 percent of the jackets were marked down from their original price, and it took an average of just 28 days for a style to sell out — indications that the style has staying power.
As recently as 20 years ago, forecasting came in the form of hefty books issued twice a year that detailed trendy colors, prints and silhouettes. When things moved online in the late 1990s, the books were replaced by sites like WGSN (Worth Global Style Network) and Stylesight.
These subscription-based sites still dominate the sector today, offering vast quantities of information assembled by teams of forecasters: from how car shapes, architecture and nanotechnology will influence fashion in 2020 to the most popular width for stripes on next season’s Breton top.
Not surprisingly, some traditional forecasters are suspicious of a data-driven approach.
“Right now data is the buzzword,” said Isham Sardouk, chief creative officer at Stylesight in New York and once a design director at Victoria’s Secret. “But for me, data is not everything. It’s just a portion of the information that’s out there. I think that intuition is underrated, and when people think of a trend forecaster they imagine a crazy guy in a room experiencing visions of salmon pink. But it’s a group decision. I have a team of 200 industry experts feeding information from all around the world.”
Looking at things as a calculated risk is part of the job, said Sarah Rutson, fashion director of Lane Crawford, which is expanding beyond its Hong Kong and Beijing stores to open in Shanghai in September. “A lot of the time it is about genuine gut instinct,” she said. “Maybe a trend didn’t work before, but this time you know it is right for now. There might not be data to tell you what to do, but you just instinctively know it will be strong and it’s absolutely worth the perceived risk to get behind it.”
Fashion is a multi-billion-dollar global industry that, at its heart, is based on creative egoists. Could that fundamental respect for the spark of genius be behind the reservations about analytics?
Topshop’s head of design, Emma Farrow, does not think so. “We’re a business so of course we look at data,” she said, “we look at the weather, we look at what’s selling, and we try out trends in select stores. But the danger is when data becomes prescriptive, because that fails to acknowledge what influences trends.
“It is no longer about buying a garment,” she continued, “it’s about buying into an aesthetic that could be linked to music or a celebrity or the Punk exhibition at the Met. I think that’s what the modern customer gets excited by.”
Fashion’s broadening geography and the growing number of seasons also have increased the variables. While the Big Four fashion weeks are starting points for trends, their influence is far from absolute, and certainly not global.
Ms. Rutson of Lane Crawford said trends and celebrities in South Korea and Japan have a huge impact on Chinese consumers at Lane Crawford. And while “lace and leopard print sell regardless of global fashion trends, tartans, brocades and thick boiled wools are a problem because the Asian customer is sensitive to these fabrics” — even if they are all over the European catwalks.
In a world where two collections a year have evolved into eight or 10, and where retailers can get products from concept to store in as little as three weeks, Ms. Fowler said sales planning must come down to comprehensive analysis and timing.
Pressure from clients actually is leading some traditional forecasters to make changes.
Catriona Macnab, WGSN’s chief creative officer, said her company was beginning to invest in a more analytical approach, as clients increasingly say they want data to back up intuition.
This year WGSN analyzed every outfit from the autumn 2013 catwalk shows in New York, London, Milan and Paris to produce a numerical index for trends on colors, patterns and products.
It found, for example, that there were 73 percent more outfits with camouflage prints in those shows than there had been in the autumn 2012 shows, and 43 percent fewer outfits using sheer fabrics than there had been the year before. The result, WGSN says, gives fashion houses the data to back up decisions for last-minute tweaks in their own autumn 2013 collections to match the runway trends.
Ms. Macnab said client response to the initial effort had been so positive that WGSN will expand the analysis program after the spring/summer 2014 shows in September. (Neither WGSN nor Stylesight would disclose their subscription prices, though they are known to use sliding fees based on a client’s size and service expectations.)
Still, Sandra Halliday, WGSN’s editor in chief, echoed the warnings. “You have to be careful with data,” she said. “It can’t be the end of the story because the way tastes develop is too nuanced. If you rely too heavily on a route, as we’ve seen the financial industry do, dangerous things can happen.”
And, she added, fashion thrives on the unexpected.
“In the late ’90s everyone was talking about the death of denim and how it had become the domain of middle-aged men,” she said. “Then Tom Ford brought out this collection of embellished denim for Gucci’s spring 1999 show. The numbers didn’t add up but it instantly changed the mood and impacted trends for seasons. It’s that element of fashion that is tricky to quantify.”
Paula Reed, fashion director at Harvey Nichols, agreed. “I still think that a magpie instinct and a sixth sense are still the best tools in the unending quest for the next thing,” she said.
Noting that Diana Vreeland had once declared that great fashion was still about giving people what they didn’t know they wanted, Ms. Reed added, “And I’m not sure I’d want to find an algorithm for that.”