Wednesday, June 25, 2008

Retail Intelligence Guest Memo: Researchers Honing in on Ways to Use GPS Data to Pick Retail Store Sites, Better Understand Consumer Behavior

Gregory Skibiski, left, and Tony Jebara of Sense Networks, a company that uses location data to make recommendations for businesses and consumers. [Photo: Suzanne De Chille/The New York Times.]

By MICHAEL FITZGERALD
The New York Times
June 22, 2008

THAT hoariest of real estate truisms — location, location, location — may soon be a clarion call for all sorts of businesses.

We’re in the midst of a boom in devices that show where people are at any point in time. Global positioning systems are among the hottest consumer electronics devices ever, says Clint Wheelock, chief research officer at ABI Research, a technology market follower. And cellphones increasingly come with G.P.S. chips. All of these devices churn out data that says something about how people live.

Such data could redefine what we know about consumer behavior, giving businesses early insight into economic trends, better ways to determine sites for offices and retail stores, and more effective ways to advertise.

Just this month, the journal Nature published a paper that looked at cellphone data from 100,000 people in an unnamed European country over six months and found that most follow very predictable routines. Knowing those routines means that you can set probabilities for them, and track how they change.

“What we do is really not random, even though it may appear random,” says Albert-László Barabási, a physicist at Northeastern University who is one of the paper’s authors.

It’s hard to make sense of such data, but Sense Networks, a software analytics company in New York, earlier this month released Macrosense, a tool that aims to do just that. Macrosense applies complex statistical algorithms to sift through the growing heaps of data about location and to make predictions or recommendations on various questions — where a company should put its next store, for example. Gregory Skibiski, 34, the chief executive and a co-founder of Sense, says the company has been testing its software with a major retailer, a major financial services firm and a large hedge fund.

Tony Jebara, also 34, the chief scientist and another co-founder of Sense, said, “We can predict tourism, we can tell you how confident consumers are, we can tell retailers about, say, their competitors, who’s coming in from particular neighborhoods.”

Mr. Jebara, who is also an associate professor of computer science at Columbia University, says the key to drawing such conclusions starts with having very large sets of data that go back several years. Sense’s models were developed initially from sources like taxicab companies that let it look at location data over such a period. Sense also uses publicly available data, like weather information, and other nonpublic sources that it would not disclose. “We had three-quarters of a billion data points from just one city,” Mr. Skibiski says.

Mr. Jebara’s statistical models interpret those patterns and look at whether they correlate with things in the real world, like tourism levels or retail sales. The algorithms are complex. Even so, the model doesn’t work for everything Sense tries it on, often because more data is needed. But Mr. Jebara says that when it has the data, the model works well. Several hedge funds made an investment in Sense earlier this year.

The Macrosense tool lets companies engage in “reality mining,” a phrase coined by Sandy Pentland, an M.I.T. researcher who was also a co-founder of Sense and now advises it on privacy issues.

Sense is not the only company engaged in reality mining. Inrix, a Microsoft spin-off, uses traffic data to predict traffic patterns. Path Intelligence of Britain monitors traffic flow in shopping centers by tracking cellphones.

Reality mining raises instant questions about privacy, especially when cellphone data is involved. In the United States, it is illegal in many cases for cellphone companies to share customers’ location data without their consent.

Mr. Skibiski says that Sense is interested only in aggregate data and that it’s looking for broad patterns, not the specific behavior of individuals. But he recognizes the privacy issue. He says he believes that people should own their own data, control when it is disclosed and receive some remuneration for it. His original idea in 2002 was to pay people for their data, but a formula for doing so proved too complicated.

Instead, Sense decided to trade services for data. On the same day it released Macrosense, it announced a new software package called Citysense, which uses location data to show where people are going, say, for nightlife, and maps their activity. Consumers who have iPhones or BlackBerrys can sign up for the service, which does not ask for personal information. Over time, the software will learn their patterns and recommend places they might like to go, or show them where other people with similar patterns are going. If they want to purge their data, they can do so at any time. [Click here to view Blackberry graphic.]

There’s little doubt that products we use everyday, like our cellphones or cars, will increasingly allow for us to be tracked. And after years of hype, there also seems to be demand for services built around location. Gartner, a technology researcher and consulting firm, says that the market — which includes various navigation and search devices and subscriptions and services — will nearly triple in revenue this year, to $1.3 billion from $485 million in 2007, and will reach $8 billion in 2011.

Annette Zimmermann, a Gartner analyst, says Macrosense seems to have a novel offering, one with a potentially large market.

“So many companies are just sitting on data” that they can’t do much with, she says. That could make Macrosense a powerful tool.

Still, Sense’s model is not a sure thing.

“The reality is that location data is new, and we don’t have 10 years of history to work from,” says Ted Morgan, the chief executive and founder of Skyhook Wireless, which sells a service that lets people use WiFi network access points to get information about their location.

“But if their algorithms can do the things they say, we’d probably do a lot with them,” Mr. Morgan says.

Michael Fitzgerald writes about business, technology and culture. E-mail: mfitz@nytimes.com.

No comments: