Online shoppers often fill up their carts and bail at the last minute when the bill comes. Usually it’s because they can’t remember their log-in, don’t have their wallets handy or can’t bear to fill out a credit-card form. A Swedish startup called Klarna has discovered a way to get people to complete more purchases over the Internet: Don’t ask for money up front.
Klarna uses a sort of honor system, requiring customers to enter only their name, shipping address and e-mail account, where they’ll receive a message with instructions on how to pay. There’s also a printed invoice inside the box when their purchase arrives. Customers have two weeks to settle up by entering their credit card or bank account number or by setting up an installment plan.
Offering to ship people whatever they want and trusting that they’ll pay later may sound like a dumb idea, but Klarna claims to have a way to outsmart most anyone trying to cheat them: big data.
More from the Buried in Big Data special report:
- The Big Data Dump: How Info-Hoarding Can Overwhelm Startups, Spy Agencies
- Big Data Is Really About Small Things
- 10 Surprising Ways Your Daily Life Is Feeding the Big Data Beast
- Saving the World? How Big Data Is Tackling Everything From Cancer to Slavery
Buy now, pay later is just one of the many ways big data is transforming retail, both online and off. Stores that know how to analyze the billions of bread crumbs browsers and buyers unknowingly leave behind are finding themselves at a major advantage over shops that rely on intuition. Big data enables companies to create comprehensive customer profiles, precise product recommendations and one-click checkout.
Today, people are finding the products they want quicker than ever, and in many cases the items are finding them, thanks to targeted ads. By placing buying habits under the microscope, companies are perfecting the science of impulse buying as sales at online businesses worldwide totaled $1.25 trillion last year, up from $1.06 trillion in 2012, according to researcher EMarketer. Fewer than 5 percent of e-commerce companies use big data or predictive analytic software, according to a July 2013 report from research firm Gartner. But the biggest companies, including Amazon.com, China's Alibaba and Japan's Rakuten, are putting their muscle behind it.
Brick-and-mortar retailers are racing to catch up with the advances in big data that have made online buying so irresistible. Even Wal-Mart Stores, a corporate giant with more than 11,000 locations worldwide, couldn't compete with the insights that Amazon can glean from the trillions of data points generated as people browse its site. So the Bentonville, Arkansas, company formed @WalmartLabs in 2011 after acquiring Kosmix, which developed a search engine geared to products. In the first year the technology was used on Walmart.com, the number of shoppers completing a purchase after searching for a product jumped 20 percent, because they were able to quickly find what they were looking for, says Ron Benson, the vice president of engineering at @WalmartLabs.
The technology group has also helped bridge Wal-Mart's in-store and e-commerce data, Benson says. This combination could give large retailers with Web presences a leg up over purely digital shops, says Gene Alvarez, an e-commerce analyst at Gartner. Being able to see what people are looking at and buying offline as well as online is a big advantage, he says. "The masses in e-commerce cannot do it right now.”
Physical retailers are also using big data, storing massive amounts of information on severs and using software to search for trends, to drive more people into their stores. Target creates customer profiles containing a person’s buying history, age, marital status, estimated salary and Web history, according to the New York Times Magazine. The company has learned that when women buy certain combinations of lotions and vitamins, there’s a higher likelihood they’re pregnant, and good targets for ads, the report said. This upset more than a few people.
Whether the targeted ads are online or off, the challenge for retailers is to avoid seeming creepy. Retailers have discovered that customized circulars mixing targeted ads with items the customer probably won’t want — a coupon for diapers alongside a lawnmower — are more effective, because the recipient won’t assume she’s being spied on. We'll buy that.
“If we come across as a stalker, we break trust,” says Gartner's Alvarez.
Putting the right product in front of online customers only gets retailers so far. The rise of Klarna in Europe demonstrates there’s much to be gained by improving the checkout process.
Invoicing isn’t a new idea, but it’s traditionally been reserved for large purchases or low-income households living beyond their means. Klarna has taken invoices mainstream in the 15 European countries that use it. Klarna accounts for as much as 60 percent of purchases at the 45,000 online stores that accept it as a payment method. The startup had $200 million in revenue last year, up from $140 million the year before, and is profitable, says Klarna CEO Sebastian Siemiatkowski. Klarna is used on Spotify, Vistaprint and the Nordics’ largest e-commerce sites.
While the company declines to discuss its fraud-detection system in detail, it uses an algorithm with more than 200 variables measuring risk. They include previous purchases, the time of day the customer buys goods, the frequency of purchases and even how shoppers type their names. The company has 80 data scientists who sift through academic research, public data and the 200,000 transactions Klarna's system processes each day for insights into how to predict fraud. By running its background check on each buyer, Klarna can find strange situations — for example, a grandma buying a computer game at 3 a.m. — and take extra precautions, such as asking the customer to put in more information or call a support representative.
"Data analysis is at the core of what we do," Siemiatkowski says.
Still, stockpiling information isn’t effective if you don’t have enough of it or the means to properly sift through it. With limited data, insights can be incomplete. For example, Wal-Mart noticed that families, especially those in lower-income areas, were buying smaller packs of groceries and household goods during economic downturns, says Karenann Terrell, the company’s chief information officer. As a result, it watched the global markets and planned its inventory accordingly. But when it factored in more data sources, including information from its online store, Wal-Mart realized that demand for bigger items jumped back up just as much during inclement weather as customers stocked up for a long winter.
"It's not as simple as them switching from larger to smaller," Terrell says. “Big data is helping us be a better company."
For Ozon.ru, the Amazon of Russia, weather serves as a particularly important factor on sales. The country is known for its heavy snowfall and brutally cold winters. After analyzing 16 years of shopping data, it found something surprising, says Maelle Gavet, the CEO of Ozon Holding. In addition to the usual winter items — gloves, coats, scarves — book sales would spike as people searched for ways to keep themselves from going stir crazy. Now, when snowflakes begin to fall, Ozon ratchets up its recommendations of books targeting each user. The next step for Ozon, which operates its own delivery trucks because Russia doesn’t have UPS or FedEx, is to use big data to reduce fuel costs and the number of trips drivers must make, Gavet says.
Amazon and Alibaba are renowned for their ability to suggest products based on who you are, what you look at and what you've bought. Where just about every site falls short, however, is recommending what to buy for your mom, sibling or significant other. The data are floating out there unused, says Gavet. For example, Ozon can infer when a user is buying a gift and who it’s for, as when a shopper buys jewelry just before Mother’s Day and has it shipped to another address. Currently, Ozon ignores the information, but it could one day be used to make gift-giving a less painful process, Gavet says.
Wal-Mart has experimented with using information from social networks to provide gift-giving advice, Benson says. It was ineffective, so development has been tabled. While useful as a complement to dozens of other data sources, sentiment analysis and other chatter from Facebook and Twitter are “not a silver bullet,” Terrell says.
For now, the focus is on getting people to spend more on themselves. That doesn't sound so bad. Does it?