Privacy and Data Mining
An off-handed quotation last week from a vice president at Sabre Group
Holdings sent shivers around the Internet. Sabre runs one of the world's
largest travel reservation systems, used by 50 airlines and 30,000 travel
agencies around the world. And according to an article published in PC Week
Online, Sabre was about to start selling the names and destinations of
individual travelers --- presumably so that the information could be used
for marketing purposes.
"We're sitting on a wealth of information," Steve Clampett, senior vice
president of Sabre Technology Solutions, told the online news service.
"Think about how much companies would pay for [the names of] people who have
reservations to go to specific places at specific dates and times."
How much indeed! For a country that's deluged with junk mail, dinner-time
marketing calls, and increasingly personalized advertisements on the
Internet, it's not hard to imagine that Sabre's information would be a
treasure trove for marketers. You might make a reservation to fly to Greece
for a week in August, and then discover that you're being bombarded by
telephone calls and mail-o-grams from tour operators in Athens, each one
trying to get you to sign up with their service. Then, when you finally
return from your vacation, you discover that your home has been burglarized
by a group of thugs who also had access to the database.
Protests against Sabre's plans followed from consumer groups, privacy
activists, and even from several attorneys with the Federal Trade
Commission. Most people said the same thing: making individual flight
records available to third parties would pose a tremendous risk to some air
travelers, and would potentially expose all travelers to unwanted marketing
efforts.
Then, just two days later, the Sabre Group issued a statement saying
essentially that PC Week had gotten the story wrong. "We do not sell
passenger names or other private information to third parties without the
consent of the passenger, and have no intention of doing so in the future,"
the company said.
What Sabre actually plans to start selling, spokesperson Jennifer Hudson
told me, is "aggregate anonymous information. We're not talking about
personal information or anything that would identify the individual."
What many people don't realize is that aggregate information that's
carefully processed is potentially far more valuable to businesses than
individual travel plans, thanks to a variety of sophisticated database
analysis techniques that have been developed in recent years. That's because
the most you can do with a person's individual travel plans is try to make
another sale. But in the right hands, aggregate information can be used
dissect a business's day-to-day operations, pinpoint problems, and improve
overall service.
Consider the Sabre. For many business travelers, Sabre lists every flight
taken, the kind of car that was rented, and the hotel where the traveler
stayed. The database also knows how much was paid for the ticket, the car,
and the room. It even knows who is paying the bill.
A simplistic way to use this information would be for a hotel to buy the
names of all travelers who were planning on staying at competing
establishments, and then send them coupons to stay elsewhere for less money.
It's this form of marketing that privacy activists are worried about.
This kind of marketing is also problematic from a business standpoint,
because it ignores one of the most valuable things in the Sabre system: the
historical record of how individual preferences and plans have changed over
time.
Here's another way that the Sabre database could be used. If a hotel
discovered that its occupancy rate was down, it could ask Sabre to find out
where all of its customers had gone. Using a new generation of online
analysis tools, Sabre could do a database search for all of the people who
stayed at the hotel last year and then see where those people traveled this
year. If they went to other hotels in the same city then competitive adverti
sing and special deals might make sense. But if a hotel in Boston found that
its customers were going to other cities, then it might make more sense to
invest in advertising that boosts Boston as a destination, rather than
advocating any particular hotel within the Hub. A big database search might
reveal that different advertising campaigns should be used in different
cities around the US.
With a little more processing, a database like Sabre's can even be used to
predict the future. Lots of businesses try to project future sales by
comparing orders in the current year with orders in previous years. But
using the cross-sectional information that's inside Sabre's data warehouse,
it's possible to take many more factors into account, and thus build a
statistical model that's far more accurate.
The techniques that I'm describing here fall under the broad category of
"data mining." It's a new science made possible by ultra-fast computers that
can handle massive amounts of information. Two leaders in the field are
Burlington-based Thinking Machines (www.think.com), which has found the
market for data mining software far more lucrative than the massively
parallel supercomputers that the firm once built; and Virginia-based
MicroStrategy (www.strategy.com), which sells a suite of data mining
products and recently completed its initial public offering.
Data mining is an incredibly powerful technology that could benefit
businesses and consumers alike. Run the systems on medical records, and
these systems can tell you which drugs work better on which cross-sections
of the population. Run it on credit-card records, and you can learn which
restaurants in a town have the most repeat business.
Data mining understandably worries privacy advocates. That's because it
requires huge databases of personal information in order to operate
properly. Today there is no guarantee that a data mining firm will use this
information to benefit consumers and society as a whole, rather than to
harass us with targeted advertising campaigns. Experience has shown that
legislation may well be the only way for us to secure such a guarantee in
the modern age.
================================================================
This message (C) Simson L. Garfinkel. .
-- Rohit Khare -- UC Irvine -- 4K Associates -- +1-(626) 806-7574 http://www.ics.uci.edu/~rohit -- http://xent.ics.uci.edu/~FoRK