“Wisdom of Crowds”, a best-selling book, hypothesizes that a diverse group of individuals is likely to be more accurate in decision making than a single individual or expert. It’s a fascinating book that borrows many themes from market theory, saying that ‘crowds’ are more accurate and faster than experts; and crowds passively coordinate and cooperate for maximum utility and maintain control while being decentralized. Pedestrians in New York walking on a crowded sidewalk, for example, rarely seem to bump into one another on the path to their destination. Author James Surowiecki provides four criteria for a ‘wise crowd’:
- diverse and independent opinions
- local and decentralized knowledge
- ability to aggregate individual judgment into collective decision
Now think of the internet as a “wise crowd”. Clearly, with hundreds of millions of people across the globe, it is quite diverse. Second, as users browse the web, the desire to jump from one site to another is an independent decision. Third, there is no central authority telling people where to visit or how to get there. Finally, aggregating individual judgments form collective decisions that take place in many ways.
Let’s look at search. The more a web site search result is clicked on by users, the more likely the site is going to be listed at the top of page one for a given set of search results. Collectively, that makes the top search listings the most important according to the crowd. This is a relatively passive way users express opinions. Facebook’s Graph Search relies on “likes” which is more of an active way of opining. In either case, they are both collective decisions across a diverse group of users determining what is relevant.
Now consider other ways people find sites besides doing a search. Facebook, Twitter and LinkedIn are leading sources of traffic for many web sites. Clearly, the wisdom of crowds is very much at play in these decentralized social environments.
Another major source of traffic for many sites is other sites – content links between sites. It’s such a large source of traffic that an search/ link optimization sector of the online marketing industry is devoted to maximizing traffic from content links, or what practitioners call backlinks. In “web site analytics” terminology, all of the above mentioned sources of traffic are called “referrers” (See Adobe Site Catalyst and Google Analytics).
The wisdom of crowds is apparent with search and social referrers, but it’s also at play with backlinks, just in a more complicated fashion. With backlinks, there is no outright voting such as a Facebook “like”. There are no explicit keywords associated with a page or site, like you will find with search. And there are significantly more potential backlinks to optimize (14 billion pages indexed by Google) than there are potential social network users (6 billion people in the world). Combined, these factors make optimizing backlinks/referrers equally, if not more challenging, than search or social optimization.
What if you wanted to target an ad based on where the user came from: the referrer? The mechanics of targeting referrers is simple: the referrer is available with most page views. One can simply read the referrer for a given ad call and analyze (however, most ad exchange impressions do not have a referrer). With search, that’s a simple task, since you can scrape the keyword from the “q=” in the referrer. With social you have access to a treasure chest of social graph profile data.
Now try to target a referrer. Do you target a keyword like search? Or a user profile like social? Or something else entirely? It’s not that clear nor is it easy. Sequential Media has solved this puzzle.
Sequential Media analyzes the referring backlinks for a given site to understand where a user is coming from (source site) and where they are going (destination site), codifying that relationship into a series of keyword concepts. Advertisers can then target these referring backlinks in real time based on the keywords we have identified. (How we actually distill all of these relationships down into a keyword is explained in our patent.) There is no cookie being set. There is no user being targeted or tracked. We’ve essentially analyzed what the “crowd” has passively come to understand as the relationship between two web pages. Advertisers can then target that site to site relationship and the ”audience” of people who travel along that path.