• by Omar Tawakol , September 20, 2004

In behavioral targeting we talk a lot about how a publishers' greatest asset is their audience. Obviously, the audience is what advertisers are buying, but the audience can also help you with some of the toughest decisions you face. In fact, your audience as a group may be so good at solving problems, predicting the future, and fostering innovation that you cannot ignore its ability to help you build your business.

Examples of this "collective intelligence" are all around us. For example, on "Who Wants to be a Millionaire?" when the audience was polled it generally did pretty well; answering questions with 91 percent accuracy while the experts only achieved 65 percent accuracy. Other examples of the power of audience include Google's page rank and Amazon's recommendations. Could you imagine a set of merchandisers trying to compete with the intelligence of Amazon customers to generate recommendations?

James Surowiecki's "The Wisdom of Crowds" does a good job of explaining this dynamic and what distinguishes a smart group from a mob. According to Surowiecki, a smart group requires the three key characteristics: diversity, independence, and decentralization. With these, each member of the group (if given the autonomy) can make a decision based on their own private information and biases. For example, eBay allows millions of bidders with vastly different desires to bid on a single product and the outcome is collected in a final price. This process works very well. So if we believe that groups or audiences have a collective intelligence, how can we apply this to online advertising?

A major issue for advertisers and agencies is figuring out which members of an audience will respond well to their message. Since advertisers can't afford to show their ad to everyone (and since consumers don't want to see every advertiser's ads) a decision has to be made. In the classical, top-down approach, an agency generates a creative for a target audience and then develops a strategy and tactics for reaching that audience.

Today, agencies test which creative and which message works best, so in a sense they are already asking the audience for their collective opinion. Why not go one step further and let the audience determine who gets to hear the message? The model behind keyword advertising already achieves some of this. Instead of assuming we know exactly what keyword works, the CPC model allows us to test and see which keywords generate success. This is really a fancy polling mechanism where the audience tells the advertisers which keywords match which ads. Clearly this is preferable to trying to find an omniscient employee to determine the right keywords.

Behavioral targeting extends this thinking beyond the CPC world for branding purposes. In essence we can let the audience who likes a message (or those we know will like the message) recommend who else would be interested in the message. This approach turns the whole process into a collaboration rather than a target practice!

Another classic problem publisher's face is inventory allocation or yield management. Given a set of advertisers and an audience, how do I price my inventory to make the most money? The current, place-based approach to pricing inventory severely undervalues a publisher's audience. It also does a very poor job reading the demand signals that advertisers provide.

Let's consider a case where an advertiser wants to reach C-level decision makers. By relying on the audience's collective intelligence, we can identify C-level decision makers and then allow advertisers to bid on access to that audience - and only to them. In addition, we would allow advertisers to bid on access to personal technology lovers, wealthy investors, influencers, and other high-value groups. The beauty of this is that C-level technology decision makers are also wealthy investors. That means that demand for one audience can create scarcity in another audience.

We are really only starting to scratch the surface in our understanding of how publishers can leverage the collective intelligence of its audience to improve advertising issues ranging from categorization, to content linking to pricing, and so on. What is clear is that, while publishers are doing a great job of attracting and packaging their audiences, the next step is to employ those same audiences to improve their businesses and deliver greater value to advertisers.

Omar Tawakol is senior vice president of marketing for Revenue Science, a behavioral targeting firm. He can be reached at feedback@revenuescience.com.