The darker side of data analytics, cyber 'profiling'

Travel site Orbitz has been presenting high-priced hotel offers to visitors who use Macintosh computers. Has it gone too far with the way its technology prioritizes search results?
Written by Heather Clancy, Contributor on

Should someone be presented with higher-priced product or service options as the result of an Internet information search just because they use a certain sort of personal computer?

That's the question raised by the news that Web travel site Orbitz Worldwide has been experimenting with showing different hotel offers to visitors, depending on whether they use a Windows-based personal computer or one running the Macintosh operating system.

According to an article in The Wall Street Journal, Orbitz has found that Macintosh users historically book hotels that are $20 to $30 per night more expensive than visitors coming to the site from a Windows PC.

That trend information, surfaced by the site's powerful analytics technology, spawned the experiment that has prompted the debate. Apparently, Macintosh users are about 40 percent more likely to opt for a pricier room. People aren't being shown different prices for the same room. Rather, Macintosh users are being presented with more expensive room offers than PC users, according to the WSJ.

Orbitz Chief Technology Officer Roger Liew told the WSJ: "We had the intuition, and we were able to confirm it based on the data."

I have to say that this article really gave me pause, even though I am a huge proponent of the idea that smarter analytics will be central to success for every business in the offline and online world. To me, this strategy seems akin to "profiling" -- making assumptions about someone's habits just because he or she happens to be associated with a certain demographic.

Maybe this struck a particular chord because I am a Macintosh user and have been one for a very, very long time. I can also assure you that price is a very big deal for me when I book travel of any sort: as a freelance journalist who pays her own healthcare premiums every month, I can assure you that getting a good deal is very much front of mind. Even though I use a two-and-a-half year-old MacBook Pro.

Reading this story will make me think twice about where I decide to book my next flight or hotel.

Don't get me wrong, I remain a proponent of business intelligence and data analytics. I absolutely think the companies that will lead in the future are those who closely follow the habits of those who visit their Web sites with personal computers, smartphones and tablets -- and record as much data related to those visits as possible.

I just happen to be troubled by the habit of using operating system as a criteria for how deals or search results are presented.

According to the WSJ, some other factors used to search and prioritize searches on Orbitz include:

  • Deals that are being offered by hotels and other partners
  • The referring site where a search originated (for example, if you started out on the Kayak site)
  • The visitor's booking history
  • The visitor's geographic location

The Orbitz experiment should definitely get marketers thinking carefully about how to prioritize the different sorts of data points the collect in their pursuit of more customized marketing strategies -- and the implications of prioritizing factors that make assumptions about someone based on their demographic and not on their demonstrated habits of the past.

There is a very thin line that companies must learn to walk between personalization and a far more negative phenomenon, profiling.

[via The Wall Street Journal]

This post was originally published on Smartplanet.com

Editorial standards