The "paradigm shifting" innovations that have moved online advertising forward have invariably involved a radical simplification and streamlining of what had been complex processes. The great example, of course, is Google, which, by focusing on simplifying the search experience for users, publishers and advertisers alike, made paid search advertising a fully coherent, compelling proposition. Display advertising is ready for a similar streamlining, one involving a synthesis of all the currently siloed targeting methodologies being advanced and used in isolation, Jim Barnett, CEO of the new automated targeting ad network Turn.com, argues below.
Behavioral Insider: You've been outspokenly critical about the way behavioral targeting has been discussed and deployed by most established ad networks. What's going to be different about Turn.com's use of targeting?
Jim Barnett: We start from the perspective that the way the discussion has been framed about behavioral targeting -- and frankly, targeting in general -- is largely nonsensical. The discussion's been stuck on the level of asking, is targeting by one kind of data better than another, is behavioral better than contextual, is demographic better than manual site selection, etc. To me it sounds like a futile approach. The answer is, they're all useful in some situations. Experience should have made it clear that these approaches, and many other types of targeting, are all valid and can be effective techniques. The pressing question is, how do you make them all available on a single platform, on an as-needed basis, when they're relevant.
Today most networks only fly one flag...
BI: Where specifically do you see conventional approaches being misguided or inappropriate?
Barnett: The core problem as I see it is that technology vendors -- and I think ad networks are equally guilty in this -- have been hung up on and preoccupied with the means (whether it's targeting methodology X and targeting methodology Y) rather than the end: what exactly does the advertiser hope to accomplish. For a network to be relevant in the future, it will need to blend all of these approaches, offering a multiplicity not only of targeting methods but pricing models as well.
Most advertisers ultimately don't care which particular targeting method is employed. What they really care about is accomplishing their goals at a reasonable price. The goal is to simplify everything by having a technology that analyzes all the complex variables that go into matching an advertiser's ROI goals, budget priorities and their message to particular placement options.
BI: What is the role or responsibility of the advertiser in all this?
Barnett: The responsibility of the advertiser is to set specific goals and know what they really want to accomplish from an ad. That's the front-end work where they really need to be spending their time. Right now, instead of focusing on marketing ends, the average advertising agency is spending inordinate amounts of time sifting through mountains of keyword lists or manually selecting sites.
BI: How does the new platform you're rolling out work?
Barnett: What we've done is develop algorithms that blend over 60 variables. The system we're developing blends site analysis, past performance, content category, user information, action type, brand strength among many other things, and sorts them out based on advertiser criteria for desired results to predict optimal placement.
Rather than serving an ad based on a particular standalone methodology, we'll use the entire continuum of targetable data based on real-time tracking being generated from each user on the network each time they visit a new page. The system basically searches the entire database to find the optimal ad to serve to that user.
BI: As I understand it, you already have over a thousand advertisers currently online and are doing a publisher's beta. What kind of goals do you have for the near to mid-term future, and where does targeting fit in the gameplan?
Barnett: In the ad network of the future -- and it's a future I think is coming faster than most people think -- networks won't be able to go to advertisers and say, 'well we can track by user search history, Web browsing and transactions and that will cost you X amount; or we can target sites based on their past performance and that'll cost you Y.' The new model will start with the question 'What is your goal and what are you willing to pay to accomplish that.' What we're aiming for is when a user comes to a site, the network will recognize the user based on profiles advertisers have provided about what type of users and what type of results they're looking for. Then the network optimization technology will make a decision on the fly about how to weight and target that user in terms of available options. The network will then select what particular ad makes the most sense for that particular user at that particular page at that particular time.