Scott Sternberg had a feeling about sweatpants.
With his first venture, the ready-to-wear line Band of Outsiders, the Los Angeles-based designer made his name twisting preppy staples. In 2016, when he began working on what would become Entireworld, he set out to build a world of basic items that were colour-rich and cleanly designed. Included on an early concept board was an image of the promotional poster for the 1983 French children’s film “La Petite Bande,” featuring an illustration of a pack of kids standing in a field of tall grass, wearing slightly shrunken sweatsuits in washed out hues. The first iteration of Entireworld’s own sweats, released in October 2018, retained the same youthful purity.
“Those were locked in,” Sternberg told me recently. “The fabric was ordered before we even launched the company.”
Now, the sweatsuit is the brand’s best-selling product, as well as its signature look — and not entirely by chance. Sternberg’s bet was that it would be a “hero” for Entireworld. But he could never have anticipated that the coronavirus pandemic would make sweatsuits the most coveted fashion item in the world.
Nor could he have known that an August New York Times magazine article detailing the unravelling of the fashion industry, cleverly titled “Sweatpants Forever,” in which he was painted as the protagonist, would bring so much attention to Entireworld that he would sell through nearly all of his available inventory within days.
Entireworld will be remembered as one of the fashion industry’s few pandemic success stories. It was a coup, especially for a brand that, having raised less than $6 million, wasn’t likely to achieve that sort of explosive growth through Instagram ads, paid media placements and other conventional forms of marketing.
But is it indeed possible for a brand to engineer Entireworld’s organic success?
The foundation of fashion is instinct. For decades, industry executives operated within the framework that being good at selling clothes required a sharp eye and an exceptional gut instinct. Designers set the trends, and it was up to store buyers to figure out which of those trends would hit it big with customers.
But in the 2010s, a cohort of entrepreneurs from outside of the industry hypothesised that collecting the right data, and crunching the numbers the right way, could take the guesswork out of design. They envisioned a more efficient — and profitable — industry.
The personal styling service Stitch Fix, for instance, hired a data engineer from Netflix to not only determine which products to send to customers, but also to help create the clothes themselves. Eric Colson, the company’s chief algorithms officer and Netflix’s former vice president of science and engineering, dubbed these data-driven designs “frankenstyles.”
“Frankenstein created his monster by combining different body parts from different people. In the same way, we borrowed a sleeve from one style, a silhouette from another and a colour and a pattern from another still,” he told me in 2016. “The data suggested that each of those things on their own were very successful with a particular audience. Combining them had signalled that they would be even better together.”
Over the years, the Stitch Fix’s data-driven design process has evolved. For instance, it now uses a tool on the site called Style Shuffle, where customers can give thumbs-up or thumbs-down to different designers, to hone the assortment.
Covid-19 rattled fashion’s faith in data. No predictive model or trend report could have anticipated a 30 percent contraction in sales, or that consumers preferences were about to make a 180-degree turn. In February, when the virus was still seen largely as a Chinese phenomenon by many Westerners, it appeared that high fashion was moving away from merch and casual wear, replacing designer trainers with heels — or loafers, at least — and logo sweatshirts with tailored blazers. Soon after, the return to formality was put on indefinite hold.
“Sometimes data says, ‘Do this,’ and what you should do is the opposite,” said Jarno Vanhatapio, founder and chief executive of NA-KD, a Sweden-based online fashion retailer that generated $237 million in sales in 2020.
NA-KD represents a promising merging of art and science. Vanhatapio, a serial entrepreneur who launched NA-KD in late 2015, continues to rely on data to run every part of his business. However, he also believes that leaning on it too heavily may shorten a business’ lifecycle. Early on, Vanhatapio created a business intelligence team within the company that operates independently from any other group. Their analysis informs each decision, but human instinct continues to play a role in certain scenarios.
Consider NA-KD’s approach to buying. More than half of the site’s budget is reserved for inventory bought in-season, which means that in order to practise “precision buying,” as Vanhatapio calls it, his team must use real-time data from customers and other sources in order to make fewer cold bets on inventory. They are often able to predict a selling curve for a product with this data, but it’s still not a perfect science.
“What to buy is always the hardest,” he said. “If we could 3-D print clothes, we’d go shallow on everything. But it’s not the reality.”
NA-KD releases an average of 50 new styles a day during peak periods, which are either bought in very small quantities (little risks) or very large quantities (sure bets). Vanhatapio has divided his product mix into three categories: Need to Have (classic items that almost every big retailer would sell, like black pumps); Expected to Have (NA-KD signatures, such as leather pants; and Unexpected to Have (novel styles that will keep the consumer from getting bored, like this season’s frilly sheer blouses). He makes the smallest inventory investments on unexpected items but feels they are just as necessary as margin-driving classics.
“Data can be very self-fulfilling after a while,” he said. “If you keep cutting back on the assortment, it becomes boring. You can’t only have data people, you need to combine this with talented fashion people. That’s where brand-building comes in.”
Even the most data-driven companies understand that human touch remains key to fashion success. Stitch Fix, which generated $1.7 billion in its most recent fiscal year, ending August 1, continues to employ human “stylists” that help further refine the product recommendations that its highly sophisticated artificial intelligence generates.
“Data is only oil when you remove the noise,” Ganesh Subramanian, founder and chief executive of retail intelligence firm Stylumia, told me. “What’s working in the market won’t necessarily work for your brand.”
Data has helped the fashion industry address its greatest inefficiencies. Brands and retailers can make better decisions about inventory — “there is no bad product if you made the right quantity,” Subramanian said — and more easily find customers. One element of Stylumia’s system works not unlike Stitch Fix’s “frankenstyle” model in that it can take two best-selling styles and create something new.
Sternberg, who has spent the last several months thinking about what comes after sweatpants, said that, as a startup with limited working capital, he still uses plenty of data. For instance, while neutral colours like navy, grey and black sell the best, he knows that Entireworld’s novelty shades are what bring people to the brand.
Sometimes, his inventory bets end up being smaller than they should have been, and he misses out on revenue. He only bought 250 units of a fashion-y wool sweater covered in a blown-out harlequin design, for instance. He assumed it would be a niche play: it sold out in a few days. But Sternberg would rather sell out than be overwhelmed with a glut of inventory. He has big ambitions for Entireworld — which generated under $7 million in 2020, according to market sources — but he doesn’t want to sacrifice growth for integrity.
“We’re walking this fine line between data and instinct, always,” he said. “Data does not create a space for new ideas.”
How Fashion Brands Can Predict the Future
How to Advertise in the Digital Age
Big Data Will Change All Aspects of the Fashion Industry