A recent Forrester study reveals that MSOs (cable carriers) and broadcasters have finally teamed up to deliver relevant advertising to viewers, thus overcoming the challenge inherent in ad-skipping from viewers with access to DVR viewing-boxes or those using digital recording devices that make it easy to move entertainment content from place to place (e.g. the Neuros OSD).
This new partnership will allow MSOs to analyze information about our viewing patterns from the cable box and would result in non-skippable ads being inserted into our viewing content based on our personal interest, even during DVR playback, hence the name "Personal TV".
In an April 2008 Forrester survey, 90% of advertisers polled indicated that they were interested in targeted Video On-Demand ads. This response has prompted me to look into psychographic modeling and targeted advertising technologies.
As companies look at new ways to scour the wealth of information we leave behind while we click through our daily online lives, each time we buy something we contribute to a database of purchasing decisions. Additionally, ad-exchange networks which position themselves between ad serving and content delivery are in a unique position to track our movement around the internet. These companies "mark" us when visiting an ad-exchange site and track us through other unrelated websites, thus accumulating an incremental portfolio of our activities and online patterns. This is a gold mine for advertisers who are now able to piece together the visiting patterns of a 30 year old visitor on 10 different web sites and deduct the psychographic profile (known as "persona") for that person: the visitor is a female; she is a single mother (she visited a support site for single mothers); she in an accountant (she has an online account with a Web 2.0 accounting software platform) and she is in the market for an new car (she visited a carmaker's website).
Now imagine if they can stream relevant ad content to this person's home via her MSO?
How else is psychographic analysis used?
A recent Business Week review of the book "The Numerati" reveals the power of analytics when it is applied to human behavior patterns. Scientists at IBM research have analyzed data about internal IBM's consultants. They looked at emails singling out communication patterns (who was sending emails to whom within the company); knowledge circles (who was being cc'd and who was being blind copied) and power circles (which managers were unaware of communications among their employees and knowledge circles been built by their direct reports).
Samer Takriti, a Senior Manager at IBM's TJ Watson Research Center, led this significant "deep computing" initiative which incorporated a team of specialists in data mining, computational biology, financial mathematics and statistics. His team constructed a mathematical model of 50,000 of IBM's consultants incorporating information about their skills, communication patterns and other demographics. The goal was to form an inference engine which could help future managers compile optimal project teams. Need to put together a team for a project in Malaysia? The algorithm would recommend a talent fluent in the local dialect, a system architect, developers, etc. Financial information would include, for instance, the hourly rate of each team member so that the manager can substitute lower cost consultants based on required skill set proficiencies and the project timeline. And that $1000/hr consultant that you decided should not be included on the project? He remains "on the bench" for longer periods of time because the model has attrition knowledge built into it so less experienced consultants are treated as "commodities" while the experienced consultants are given greater consideration to job satisfaction.
I find this application quite interesting in light of the Rummler-Brach method to improve organizational performance ("Improving Performance"). The authors hypothesize that in order to truly improve performance, a company must address operational efficiencies at three levels: organization, process and the individual employee.
The lowest level of their model, the individual employee, is where they claim many companies go wrong. Rather than throw unnecessary training at the employees and reprimand employees for inadequate job performance, the authors define the job of managing employees as managing the "Human Performance System". It is based on the premise that people are motivated and talented but are at time put in situations that are suboptimal to their skills. The questions that need to be addressed when managing the Human Performance System are: do the performers understand the outputs they are expected to produce? Do they have the proper support? Are they rewarded for achieving job goals? Do they have the necessary skills, knowledge and capacity to achieve the job goals?
Takriti's model may be a step in the direction of automating the management of the Human Performance System and can perhaps be an interesting incumbent to prevailing analytics aimed at building psychographic models of web visitors.