Using Big Data to help shoppers choose

A few start-ups are using new technology to make online shopping a highly personalised experience, says Molly Wood.

Katrina Lake is staring at the ceiling, trying to remember the last time she bought clothes in a store. Lake is impeccably dressed, in crisp navy pants, tan sandals and a semi-sheer white cotton top. Simple bangles and an elegant gold necklace top off her look. She obviously shops, but in recent years, all her clothes shopping has happened online. She thinks others are ready to shop online just as much - they just need a helping hand.

That is why she started Stitch Fix, a women’s clothing retailer that sends its customers boxes of clothes that are picked by a combination of personal stylists and big data. It is one of a handful of startups making aggressive bets that highly personalised online shopping, in which sites choose items for you based on your preferences and their algorithms, can deliver a better e-commerce experience.

Trunk Club, for example, is similar to Stitch Fix, but is for men and is more expensive. Club W tailors wine to its monthly subscribers, who fill out a “palate profile” when they sign up. And Birchbox, a beauty store, delivers monthly installments of sample-size goods tailored to your needs for $10 (Rs 604) for women or $20 (Rs 1,208) for men.

“At a mall or online, the choices are overwhelming,” Lake said. “Then if you buy something and you want to return it, it takes a long time. People just expect a more personalized experience now.”

The sort of shopping offered by Stitch Fix is still a tiny slice of the overall e-commerce pie, which the research company eMarketer predicts will increase by 20 percent in 2014, to $1.5 trillion globally. 

With this kind of rapid growth, many tech entrepreneurs see room for opportunity, and some traditional retailers are picking up on the trend, too. Late last month, Nordstrom said it had agreed to buy Trunk Club.

With these new personalised shopping sites, the magic comes from data. The sites learn about you and then compare your information with profiles of thousands of other people like you to predict what you might want to buy.

People using Birchbox tell the company their skin colour, hair type, age and favourite types of products, for example. The company compares that information with other profiles, sees what those people are most likely to buy, then fills a box with beauty samples that match those preferences. Full-size versions of the products can be bought on the site.

Stitch Fix hired Eric Colson, its chief analytics officer, from Netflix, where he was responsible for TV and movie recommendations based on what customers previously watched. A computer program used by Stitch Fix helps filter a huge catalogue of clothing and accessories.

The program then matches items to a separate, and also huge, collection of information about people, including their sizes, shapes, colour preferences, style habits and current trends.

 “I think there’s way more data science at work here than people may realise,” said Bill Gurley, a general partner at the venture firm Benchmark, an investor in Stitch Fix, who is on the company’s board.

 “There’s a 15-page profile, there are over 66 characteristics tracked and there’s a predictive heat score for every single item against every single user,” meaning a way to determine the likelihood that a customer will keep an item.

Ultimately, though, a human stylist chooses items from the computer-edited list and packages them into a nice little box with a description of each one and a personal note explaining what you have received and how you might wear it.

Lake said her concept for personalised online shopping came out of a desire to combine the upside of in-store help with the convenience of e-commerce. “In traditional retail, they have the benefit in that they have real people that in theory could offer a very personalised experience,” she said. “So, how do you bring the experience that people are having online, and how do you personalise and get to know someone well?”

That sort of real-person interaction might still happen at high-end physical stores like Nordstrom. However, physical shopping is increasingly a transaction of convenience, often involving something researched online before being bought.

 “Over 90 percent of commerce still takes place in the brick-and-mortar store,” said Bill Martin, founder of ShopperTrak, which uses Nielsen-like tracking devices to monitor shoppers’ visits to stores and malls. “But 60 to 70 percent of purchases are now researched online.”

Retailers big and small are trying to personalise shopping in physical stores, too, say retail analytics companies like CQuotient. These efforts include loyalty programs that let sales employees easily pull up account histories of their customers’ purchases, and even in-store beacons that communicate with a shopper’s cellphone and can send relevant offers and coupons if the shopper uses the store’s app.

Retailers like Urban Outfitters, Mango and Zara appeal to their mobile-friendly young shoppers with apps that offer benefits like exclusive access, rewards and even, in the case of Urban Outfitters, a streaming radio station that evokes its hip shopping experience.But none of those options, as yet, direct shoppers to personalized selections. That is something Birchbox is hoping to achieve, in addition to its online option, with its recently opened first physical store in New York’s SoHo district.

Inside the store, current customers can tell store employees their email addresses and be directed to items that are most relevant to them. 

New customers can fill out a version of the Birchbox questionnaire to find the products they are most likely to appreciate. They can try out products at their leisure, to avoid a situation in which the personal touch feels more like a hard sell.

“There is still this desire to work with somebody that you really trust,” said Katia Beauchamp, a founder and chief executive of Birchbox. “You probably trust the consumers and the overwhelming number of consumers who are like you, and who are then buying something.”

Will a virtual personal shopper really fill the void? The act of shopping - going to the mall, browsing the aisles - is still a powerful cultural and social draw. And it’s still easier, sometimes, to discover new items hanging on a mannequin or even tucked away on the sale rack.

Also, the cost of multiple personalized subscription services for beauty products, wine or clothing adds up quickly. But personalised selection could make e-commerce accessible for people who might find pages and pages of online shopping both overwhelming and impersonal - and who find the mall the same way. In this case, though, it’s the web and not the stores that provide the personal touch.

“We realised that shopping online, if you know exactly what you want - it’s fast, it’s so simple, it’s cheaper,” Beauchamp said. “But it’s not about the enjoyment, the hobby, the sport.”

“We use curation and personalisation,” she said, “as a way to make the Internet have some of those fun and satisfying elements of shopping.”