After weeks of lively emails on the pros and cons of profiling, this past week’s article on opt-in versus assumptive profiling (e.g., profiles based on observed behaviors such as clickstream data) was met with dead silence. Have we beaten a dead horse here? Next week, we’ll move on, but for now, a few final wrap-up thoughts.
The opt-in profiling crowd was far more vocal about the merits of its profiling system, but we still came up very short on actual case-study results. Here are the big issues for the opt-in approach as we see them:
- With everyone and his brother attempting to build opt-in profiling systems, will users take the time to fill out yet another form about their preferences?
- Can profiling companies actually provide meaningful enough incentives to get people to do this?
- Can they actually deliver enough value to profiler users, either in the form of targeted advertising, content or cash/discount incentives, and control the advertising messages sufficiently to keep people from opting out?
The answers to these questions will drive the long-term success of this approach.
The behaviorally based profiling crowd was pretty quiet. A few readers pointed to articles about how behavior-based profiling is working well in e-commerce applications, but we were more focused on the ad-targeting aspect of profiling. In response to our question of whether this type of profiling works, one astute reader commented:
“Yes, I think more often than not. It works due to relevancy but like any research data, if you scrutinize it too much, your data becomes unstable. Also, advertisers are falling in to the Internet trap: They have too many options and no idea of what they want to do exactly, and they usually choose the wrong ones, like a kid in candy store who can’t make its mind on what to have and ends up crying.”
“I’m not interested in getting into a debate about whether predictive targeting of Internet ads based on prior behavior is offensive or not. But I do want to point out that this sort of targeting is the basis for the direct-mail industry. The fact that the Internet can target anonymous (no name and address) users makes it potentially less intrusive than direct mail. As long as they keep away from tying cookies to name/address records, it seems to me that a strong case can be made: If the public has been able to live with direct mail, they can live with this. Isn’t improving the relevancy of the ads to the consumer a good thing?”
Finally, another reader questions the entire concept of profiling. He noted:
“Profiling by any means always has been and always will be a best guess. Whether done from surreptitiously collected cookie information or fib-infested user-supplied profiles, the intentions of the profiling system [it] has questionable economics. Why not skip the guesswork and use keyword and URL targeting in real time?”
Here’s our sense of the “state of the art” use of profiling in ad targeting: Opt-in profiling should be a clear winner if the issues raised above are effectively addressed. We suspect, however, that it’s a big “if.”
Behaviorally based ad targeting has tremendous potential, but both profiling companies and advertisers are in the very early stages of figuring out the best model for implementation and pricing. Thus, the success stories are not pouring in.
At last count, however, there were over 400 companies in the profiling space. We believe that some are already getting it pretty close to right and others will be hot on their tails. We’ll share further information with our readers as it becomes available.
Next week: the impact of “community” on media selling.