A lot of work has already been done for showing targeted ads to visitors based on their stored profile information. For instance, Facebook users can currently be targeted not just on simple demographics attributes such as location, age, sex, and marital status, but also their major at college, which groups they are members of, and which movies or actors they’re fans of.
However, a lot of research on the Web is done by anonymous users – people who are not registered with a system. And more and more people are balking at registering themselves and sharing their data just for researching product information.
At last month's Innovations in Management Conference, Glen Urban of the Massachusetts Institute of Technology presented a study demonstrating how certain aspects of a visitor’s profile can be built just by monitoring his click stream. Thus, a personalized advertisement can be shown to an anonymous (“guest”) visitor.
In a study done by Urban’s team, the team first surveyed 600 people – a short representative sample of all visitors who volunteered to participate in the study – on CNET, a popular electronic gadgets website. The survey asked them a series of questions to determine their cognitive style in two different cognitive dimensions: (1) deliberative/impulsive and (2) intuitive/rational.
Based on their answers to the 20–30 or so questions for each of these two dimensions, they were divided in four classes of cognitive styles: deliberative-intuitive, deliberative-rational, impulsive-intuitive, and impulsive-rational.
Then, 15 high-traffic pages were selected, and click-stream information was gathered for this reference sample of 600 people as they clicked on links on those monitored pages.
After a few weeks, extensive click-stream profiles were built from this reference set of people. These profiles could then be used to determine the cognitive style of any visitor to these 15 pages based on his or her click stream.
During the rest of the study, the deduced cognitive style was used to show the all website visitors who visited one of those 15 monitored pages a variation of an advertisement for a new HTC phone (selected from a set of four variations of the same advertisement) based on the cognitive style deduced from his or her click stream. For instance, in the ad for deliberative types, the call-to-action button said, “Learn More,” while for the impulsive types, the button said, “Get It Now.” In the same vein, for rational types, features of the product being promoted were listed in detail, while for viewers classified as intuitive, all the ad said was “Online deals! Free shipping!”
Once one of the four similar-but-unique advertisements corresponding to the consumer’s classification is shown, that ad is repeated throughout the visitor’s stay at the website.
The results were very interesting. Compared to those who were shown a generic ad, there was a whopping 87.1% more clicks from visitors who viewed the ads personalized to their cognitive style.
What does this mean to us who engage in social and mobile commerce? The future of targeted advertisement goes way beyond information provided by the people, i.e., what people say about themselves; it will be based on profiles that are built on the fly, based on what pages they visit and what they “Like.”