Wednesday, December 29, 2010

Do we Understand OBA and Tracking?

With all the discussion in the social and political spheres about online behavioral advertisements (OBA), there's an awful lot of heat underneath the two points of view about tracking.  On one side, people are firmly asserting that most folks don't want targeted ads, while on the other side the assertion is that targeted ads are much much better (and less annoying) and people actually prefer them.

I think there's a bit of a gap in which assumptions are employed.  It's not clear how much the average consumer knows about targeted ads, and without knowing that, we can't assume anything about what they want. Furthermore, if we ask questions that assume unknowns, the results are useless -- asking for approval of targeted ads means two different things if the respondent does or does not understand that the ad network has to profile them to accomplish this.

Collecting Data.
First of all, it's important to acknowledge that these ad networks are already accumulating a whole lot of information for a variety of purposes: click-through rates, ad placement effectiveness, ROI, etc.  They just don't directly tell consumers about this.  In fact, many networks have been tracking individuals from ad impression all the way through purchase (conversion tracking) as yet another metric of successful advertising -- and for those of you who buy ads, you understand how powerful this knowledge is.  These already existing uses of data are incredibly helpful for advertisers (and the networks too, for ad placement), but it's not clear if consumers realize this is happening.

Online behavioral advertising is probably just a new use for already collected data.  The data's already warehoused, and it would take an awful lot of arm-twisting to reduce what ad networks collect -- because it's used for more than just OBA.

So I think the set of questions we ask consumers should be changed from "do you want targeted ads?" to the more precise pair of queries: "is it acceptable for ad networks to build a profile on you?" and "may ad networks use this profile to customize ads you see?"  We don't yet understand if people want this data collected, moreover want it used for OBA.  This distinction is necessary, because I think there are three groups of people here:
  1. people who don't want a profile built or their browsing habits recorded at all.
  2. people who accept conversion tracking but want to see the same ads as everyone else.
  3. people who accept tracking and want the profile created to be used for displaying targeted ads.
My hunch is that group 2 is smaller than 1 and 3, but I haven't seen a survey measure precisely this.

Measuring Desire.
Gallup has taken a step to gather preliminary information about consumer preference; unfortunately, it only seems to address the vague question "is it okay if ad networks show you custom ads?"  I'd really like to see a series of questions that measure habit-tracking approval as well as approval for use of tracking in OBA.  Once we know this, we can figure out if people need a tool to opt-out of data collection, or if a tool to control data use is enough.

My personal point of view (not that of my employer or anyone else) is that OBA is not an invasion of privacy -- the stealth collection of data about my preferences is the invasion.  Perhaps it's worth the convenience like cell phones or ez-pass toll devices, or perhaps it's not.  Whether or not this invasion of privacy is acceptable should be ultimately up to you.  As with many things, you can trade bits of privacy for something new.  Would you like to "buy" better ads with your browsing history?

3 comments:

asadotzler@gmail.com said...

I think your final paragraph is the critical point. The big online privacy problem is not customized ads. It's the data that is collected that allows for things like customized ads.

This is why the "opt out" of behavioral advertising proposals from the advertising industry isn't sufficient. It's not that I don't want targeted ads (I actually prefer them over generic ads) but that I don't want these companies to have ridiculously detailed profiles of me and how I use the Internet.

Just seeing the targeted ads might be creepy to some. But it's creepy because of what it means, not what it is.

Anonymous said...

While I generally agree with your conclusion, I think the set-up needs work. I don't hear very many people asserting that most folks don't want targeted ads. I do hear a lot about the opacity of the entire process from the consumer perspective, and about the rich ecosystem around personal data collection, storage, aggregation, and use (and abuse).

There are many scenarios, each with its own set of issues: first-party data collection on users who have logged in to a service, third-party collection of data for the current browser session, length of time that personal data is stored, secondary uses of data and data sharing, how uniquely a data set identifies a person or device (including de-anonymization), how broadly personal data gets aggregated across time, web sites, browsers, devices, etc.

There is incredible complexity in the spectrum of personal data use for OBA and related applications, which makes it extremely difficult to generalize about whether, how, and to what extent, various aspects of OBA are desirable to users.

Anonymous said...

I think you're dead on here, Sid, but I'd add that the second assumption (targeted ads are better) also could stand scrutiny. I say this because I'm pretty promiscuous with my data, and yet the evidence suggests to me that (outside of when meg whitman was clearly flooding california computer users with advertising) most advertising does a spectacularly lousy job of targeting me. At best, like with Whitman, I get geo-targeted ads, but otherwise I've never seen anything that effectively targets anything about me; nothing pegged to any demos I'm a part of.

Perhaps this is because they aren't collecting data very extensively yet, but I have to wonder if it isn't that they're just not very competent with the data they do have. (Data point there: Angry Birds can only be played effectively on Android phones less than a year old, and yet at least 1/2 of the ads on Angry Birds are for... new phones. Most of them not Android.)

[One exception: when I was "engaged" on Facebook... that got me tons of targeted ads. None effective, mind you, but at least visibly targeted to something about me other than "english speaker."]