There has been much buzz lately about the impact of behaviorally-targeted advertising on consumer’s privacy online. Noting that consumer’s online interests are now tracked and profiled by thousands of pixels and cookies, Senator John Kerry has proposed legislation to force websites to allow consumers to opt-out of tracking. This comes on-top of the FTC’s own “Do Not Track” program and the self-regulation efforts of the Internet Advertising Bureau and the Better Business Bureau.
Most of these efforts are motivated by a desire to retain the potential benefits of targeted advertising while averting any chance of an Orwellian scenario where the mindset of every citizen can be predicted from online data collected without their explicit consent by websites and traded on exchanges like BlueKai and eXelate.
However, I’m not sure if anyone who is presently involved in regulating or self-regulating online advertising has ever tried to answer a question which I just set out to answer: how good is the actual data quality within advertising exchanges?
Data quality in behavioral advertising has significant implications because it may very well be that the most harm to consumers could come from advertisers “knowing” a lot of false facts about consumers and targeting advertising against them. For example, if BlueKai thinks that I’m a Male, 35-44 who lives in a home valued at $100-200K and drives a Domestic SUV, then I’ll see a number of advertisements for Lowe’s and Ford that are almost completely irrelevant to my, actual, Manhattan lifestyle. Meanwhile, a company whose products I actually care about, such as American Express, may use this information to determine NOT to provide me with their best possible terms and promotions. In this scenario, a customer fails to benefit from receiving properly targeted advertising while being harmed by the possibility that information about them will be mis-used.
Thanks to Evidon’s “Online Profile Manager” I was able to catch a glimpse into the information kept about me by a number of major online advertising brokers including BlueKai. The information provided captures my membership in a number of “audience interest groups” which advertisers can target banner ads against. For example, using the data below Capital One might see that I’m a Consultant (i.e. frequent traveler) who is interested in Credit Cards and Airline Reward Miles, so they would send me ads for their Ventures reward card:
Here’s what I learned about data quality on BlueKai:
- BlueKai tracks 285 “facts” about me and my interests… but most of the topics are innocuous: BlueKai tracks my interests in shopping and travel, propensity to purchase a variety of products: from life insurance to cold medication to tissue paper and demographics at the state, age and income bracket level.
- BlueKai is correct about my interests just 35% of the time
- The erroneous information provides a pretty lousy profile of me: they think I watch a TV show called “One Tree Hill” which I’ve never heard of. Apparently it’s popular with teens. Yet, they believe me to be a Male age 35-44 who lives in a home valued at less than $100K (is that possible in New York?). While they recognize that I live in New York City area, they also believe that I live in a “Average to Low Cost of Living” area. I’m simultaneously a member of both the “High Net Worth” and “Lowest Wealth Decile” groups and I am included in the interest group who shops at “The Gap” and “TGI Friday” but can’t recall visiting either chain in the last 13 years.
- Profiles miss most of the important traits which actually define me: I was raised in Ohio but support the Red Sox; I love learning, revere Enlightenment values and studied history at Cornell and business at Chicago; I speak four languages; I don’t use Facebook much but I share interests like dining, world issues and technology innovation with a few close friends. I recently entered my thirties and will be getting married soon so am beginning to take pride in “growing older” while also worrying about the responsibilities that entails. If I could receive a behaviorally targeted ad for something, I’d like it to involve learning to drive race cars or buying something nice for my parents…
In conclusion, this little experiment has been enough to show me that the data available about me on BlueKai hardly instills any Orwellian fear. The central problem is that I generally wouldn’t recognize myself from the profile presented on BlueKai. This means that if I want properly targeted advertising, I’m not getting it.. but if I want to stay off the radar, I may be better off staying “opted in” to BlueKai’s scrambled data — letting them think they are reaching a a potential Chevy truck buyer who also shops for M.A.C. eye liner — than “opting out” of it.
