There Is an Algorithm for Everything, Even Bras

THE two and a half miserable hours that Michelle Lam spent in a fitting room, trying on bras, one fine summer day in 2011 would turn out to be, in her words, a “life-changing experience.” After trying on 20 bras to find one that fit, and not particularly well at that, she left the store feeling naked and intruded upon.

A screen shot from True&Co’s bra-fitting quiz. A customer starts by entering the size and manufacturer of her current favorite bra. The company then uses an algorithm to try to find a better fit.

“It occurred to me in that fitting room, as I was waiting for that saleswoman to bring me bras: Wow, this is the worst shopping experience on earth,” she said. (My wife concurs.) From her frustration that day emerged an idea for a business called True&Co.

The history of e-commerce is marked by start-ups devising ways to sell products that were once thought of as unsuitable for sale online. Shoes were not supposed to be something customers would buy online, but then Zappos showed it could be done. The same thing was said about eyeglasses, until Warby Parker came along. But bras, which are among the most personal items someone can buy, represent the Everest of online retail challenges.

Ms. Lam’s company opened True&Co last year along with two co-founders, Dan Dolgin and Aarthi Ramamurthy. The company, based in San Francisco, is certainly not the first to sell lingerie online. Older sites include the Web arm of Victoria’s Secret and HerRoom.com, which was founded in 1998, near the dawn of the Age of E-Commerce.

Professional bra fitters have also moved online. Linda Becker, whose family owns two bra stores in New York, says she sells twice as many bras online today at LindaTheBraLady.com as she does in her stores. Some of her online customers have previously visited one of her shops and been fitted in person. But new customers take their own measurements and work with customer service representatives on the phone. She says only 10 percent of online orders are returned.

But some customers turn out to be extremely hard to fit and it’s hard to tell why, Ms. Becker says. “That kind of customer will be impossible to fit online because the problem is unseen. There’s no way of figuring it out over the phone.”

True&Co’s innovation is to put a batch of bras into customers’ hands so they can choose what fits best. New customers take a quiz â€" modeled on the ones in Cosmopolitan magazine that Ms. Lam fondly remembers filling out in high school â€" to collect the information needed to fit the bra properly. They are then invited to pick three bras in different styles.

True&Co uses an algorithm to pick two additional bras to send out, based on what can be discerned from the customer’s choices. So the customer ends up with five bras to try on at home, with no obligation to buy. Most of the company’s bras are priced from $45 to $62.

The 15-question quiz asks for the customer’s band and cup size and the manufacturer of her current “best fitting (and beloved) bra,” and works from there to determine how the fit of that favorite bra could be improved. Other quiz questions include: “Do your cups runneth over” citing things like cleavage or underarms â€" or “No spills, all good.” The question “What is your shape” is followed by these choices: Well-Rounded, Bottom Happy, Taking Sides and Bottom & Sides.

“We have an algorithm that defines 2,000 body types,” Ms. Lam said. True&Co does not make customized bras for each of those 2,000 body types, however, so much of the taxonomy’s precision is lost when it must be translated into the far fewer combinations of band and cup measurements used by bra makers.

True&Co has drawn the attention of some skeptics. Last month, a blogger at Open Source Fashion, Sindhya Valloppillil, dismissed the company’s bra-fitting algorithm as “ridiculous,” arguing that a bra must be “touched and tried on.” She mocked the credulity of True&Co’s venture capital investors in a post titled “V.C.’s Think My Boobs Need an Algorithm.”

True&Co actually makes no patently ridiculous claims about the algorithm, which involves matching a woman’s body type to a particular bra based partly on consistent variations among manufacturers for a given size and style. One manufacturer’s 32C may work better for breasts of a certain shape, for example, even if a woman is used to buying a 34B.

Customers buy an average of two bras from each batch of five. The company says women end up buying more of the bras chosen by the algorithm than the ones they select themselves.

But as with shoes and eyeglasses, so too with bras: it’s love at first touch and try, even in the digital age.

Randall Stross is an author based in Silicon Valley and a professor of business at San Jose State University. E-mail: stross@nytimes.com.

A version of this article appeared in print on February 24, 2013, on page BU3 of the New York edition with the headline: There Is an Algorithm For Everything, Even Bras.