The founder of online lingerie store True&Co thinks she has women all figured out, and thanks to the reams of data the site has collected, she just might be right. Today, the retailer—which until now only sold a handful of bras of its own creation, alongside more mainstream brands—announces a new bra sizing system and lingerie line based on the analysis of hundreds of thousands of women’s body types and preferences.
Great algorithms seem to be the key to e-commerce these days: Netflix is famous for its ability to guess what movies you’ll like, and everyone wants a quiz to magically present them with ideal products, perfectly curated to suit their needs. True&Co, which uses a home try-on model (à la Warby Parker) to sell bras and other lingerie, has followed this creed since its creation in 2012. Bra-buyers answer questions about their current sizing and problems with fit in an ever-evolving quiz, which the site uses to make possible lingerie picks. Shoppers receive a box of five bras and try them on, keeping and paying for the ones they like. But the key is that they’re asked for feedback on all five.
“It’s a virtuous cycle of data,” True&Co founder and CEO Michelle Lam told Quartz. “Our customers write us essays on why a bra didn’t work.” And by analyzing the feedback of 500,000 users, Lam says the site has a unique grasp on what makes a good bra.
“We knew going into this that a great fit was more than just a band and cup size,” Lam says. “And this isn’t a solution designed by two women. It’s been designed by half a million.” In contrast, she says, most existing bra companies work based on a single fit model. “If you’re buying a bra from an American company,” Lam said, “You’re getting one based on an Asian model’s breast shape.”
Lam expected to see a hundred or so basic body types. Instead, True&Co has classified more than 6,000 distinct types already. She’s used these to create TrueSpectrum, a “shape identity process” that will help women identify bras that fit their body type. And with it comes “uniform,” a line of basics tailored to fit particular groups.
“We’re all about the combination of art and data,” Lam says. “It informs the design, but never governs it.” The site’s quizzing found, for example, that 45% of women complained of painful underwires, but “were reluctant to go with a bralette [a bra with no wires] that felt too unsupported.” Uniform’s bralette has been tailored to give a bit more support than most would expect.
Data also showed Lam that women preferred dark colors, which outsold lighter bras 3:1. “But we didn’t want black,” she says. “That’s so boring.” So designer Nikki Dekker used “elevated neutrals” like charcoal for the new line. When it came to women’s underpants, “We also got very precise about cheek coverage,” Lam says. “Our customer is a little bit sexy, but she wants her hipster to cover enough that she feels comfortable.”
And it seems that True&Co customers have enjoyed being part of the data collection experiment: Customers are more likely to complete the quizzes when they’re longer, Lam says. But that’s probably because the company has drawn a crowd that enjoys feeling like they’re being matched to an ideal bra. “We’re like an OkCupid [the dating site] for your body,” Lam says.
Companies like True&Co and Pinrose, a recent startup that uses quizzes and data analysis to suggest and create perfumes, are among many firms using large amounts of data to make irritating processes slightly more enjoyable. Neither store will actually cater to every potential customer on earth—many women will still have trouble finding bras that fit them on True&Co, and Pinrose can’t make scents that please everyone. But if e-commerce sites are to use algorithms as more than a gimmick, this is what it will look like.