I've registered the fact that I am a fan of Judas Priest's "Defenders of the Faith" album in my Facebook profile. The album came out in January of 1984, and I saw them live at the Cow Palace in San Francisco later that year.
That is a single data point about me that has been declared to the public, using a media vehicle that collects and keeps data points.
Based on just that singular datum, a person -- or decision engine technology -- can deduce that I:
a. Am a Judas Priest fan
b. Like heavy metal
c. Like things from the 1980s
Through a combination of correlating technologies, based on just the one fact that I like Judas Priest's albums, one could also reasonably conclude:
d. My age
e. That I had a tract-house suburban upbringing
f. That I'm a white guy
If my fandom of "Defenders of the Faith" is registered on Facebook, but that "fact" can be a constituent of as many as six behavioral profiles, then who owns the original data point and its derivative profiles? Does Facebook own the original data point and all its extrapolations, being ground zero for the original record? Does the agency that developed the definitions for the derivative profiles own it? Or does the analytics outfit that concluded correlations based on the original data point and packaged them up for sale to supplement media efforts own it?
Does each advertiser own their own version of the profile that is defined when their ad is served to me and data is collected based on my actions?
Or do I own it?
Data sourcing and ownership is going to be -- if it isn't already -- the issue facing online advertising in the coming year.
The recent changes GroupM has made to all of its insertion order terms and conditions is only the first sign of things to come (see Tom Hespos' column, The curious problem with WPP's GroupM).
The even more recent hubbub over Facebook's changes to its terms of use -- and the subsequent suspension of those changes until further review -- also brings to the fore the impending struggle over "whose data is it, anyway?"
The focus on data ownership among publishers, networks, agencies, advertisers, and data aggregators is motivated first and foremost by who will reap the rewards in an evolving data economy. To do that, two things must happen: the value of data -- each piece of it -- must be determined, as well as who owns each datum.
The fact that I'm a Judas Priest fan is of value to those targeting Judas Priest fans, and to those who are targeting white boys who grew up in the suburbs. And that value is determined not just by the targetable characteristics, but the product or service that seeks the advantage of being in association with it.
Hespos lays out perfectly what's wrong with the notion of there being a single point to which data ownership can be attributed, and why it is impossible to read a single point of data in a way that would limit it to only one reading. In the Judas Priest example, one data point can be read in at least six different ways. But is it the data that is owned, or its interpretation?
Some kind of "reasonable use" clause is going have to be introduced in concert with reasonable efforts to preserve confidentiality. Confidentiality ultimately is that of the individual, not the collector of that data.
Example: I click on a heartburn ad on iVillage. The ad belongs to the advertiser, but the space it runs in belongs to iVillage. The click and ensuing data, while I'm still on that site, belongs to both. The data generated when I go from iVillage to the heartburn medicine maker belongs to the heartburn medicine maker. The advertiser for indigestion treatment that is running on the same page might know that I clicked on the heartburn ad owns that datum. The fact that it is me clicking belongs to, well, me.
Protecting the "me" is going to be where the whole data quandary needs to resolve itself, and it is here where -- if the industry doesn't find a way -- the government will step in. And I fear that the government, or some other regulatory body, is going to have to step in. Aside from the Motion Picture Association of America, few industries do a good -- or even adequate -- job of regulating themselves if they have to.
If the issue of data ownership isn't settled properly, those who know far less about how behavioral targeting works are going to be the ones to shape how this business evolves.
Of course, the industry may answer another question that is important to the practice of behavioral targeting in such a way that may naturally limit its scope.
Agencies, their clients, publishers, and data aggregators are all featuring more and more their ability to render human behavior into machine-readable form so as to minimize the uncertainty of determining human action taken in response to marketing messages. It is marketing that finds itself far more interested in the application of better math than in better psychology. But so much of human action is instigated by the secret motives of the human spirit, which is not easily subject to algorithms or applied to click-streams.
Does more and more data lead to better results for advertisers? There is evidence that data combined in some ways leads to better results, but just how far can it all go?
We are better, after all, at collecting data than we are at doing anything with it.
Media strategies editor Jim Meskauskas is vice president and director of online media for ICON International, Inc., an Omnicom Company.
