Back in 2009, I read an article that compared the price of Kim Kardashian's tweet with the purchase price of commercial airtime, and banner ad placement. What struck me as being so interesting was the fact that Kim Kardashian was being paid $10,000 per tweet for Carl's Jr. While most articles focused on this astounding price point, I was instantly more interested in the fact that the channel method of ad delivery was twitter. As an advertising medium, twitter is unique because it is(was at the time) assumed to be personal. On the surface connecting the dots between the number of followers a celebrity like Kim Kardashian has and the value of the media impressions targeting her fan base is both easy to understand and a good idea. Below that surface is the value of the individual and the complexity of the metrics that make that valuation possible. As I thought more about this, I realized that the future of advertising will predictably lead to in depth micro-analysis of market behavior and will also lead to unpredictable changes to our society. Here I will informally explore my developing theories about how this will be done and what may result.
For me, the most interesting thing about Kim Kardashian's tweet price is not the value placed on her tweets, nor the number of followers she has, but that she is valuable because of an assumption about her leadership. As we start to value the numbers of followers a celebrity has, we will start to value the details that tracking public data from social networks like twitter possess. This data is potentially more valuable than the media impressions garnered from one tweet. For instance I heard a story that Jay Z had set London as a stop on his tour and approached Spotify for information about his tour locations and based on the geo-data provided by Spotify users, they were able to inform Jay Z that his fan base was stronger in Manchester. He changed his tour and the stop sold out. This is one specific example about the value of social network data farming. Currently most of the data that we farm or at least use for leverage in an article is crude. We name the number of youtube video plays, retweets, hashtags and likes when citing the popularity of a given media subject. What I propose is that explaining social behavior from youtube video hits is like trying to explain our society based on the amount of morning commute traffic on a given freeway. While the ways that we collect data are becoming pretty sophisticated, the ways that we analyze the social aspects of these metrics is in it's infancy.
The first term I coined is called Social Flocking. I chose this terminology because a part of predicting mass behavioral trends is observing how a collective behaves. In the way that a flock of birds or a school of fish organically shifts it's trajectories, a sub-culture, demographic, or network will flock towards or away from behavioral trends. Where trending refers to the increasing popularity of a behavior, social flocking would refer to the cohesiveness of given collectives of people. Where we might observe that the use of the "Draw Something" app is trending, an understanding of social flocking might tell us that users over 50 or app users that listen to Justin Bieber and tweet about the show Glee usually indicate that the Trend is on it's way out. Observing not just the number of hits but predicting the changes in the trends social gravity are far more valuable.
Which lead's me to another important aspect of mass-behavioral observation. Specific sub-cultural demographics or even individuals are leaders while others are followers. We will eventually care whether or not users are early on a trend or late on a trend and the frequency with which they lead a trend. We will care because we not only want to know how many people a tweet from Kim Kardashian reaches but we will want to know which people react by following the behavior and which people do not. Perhaps Justin Bieber's "Trend-Leadership Capacity" is greater than Kim Kardashian's so advertisers should value his tweets more because they ensure more leverage on top of the volume of media impressions. This is where things start to get complicated. Measuring an individual or flock's trend-leadership would require really difficult and vast amounts of information to process.
I don't have specifics yet, seeing as how I am calling attention to the problem more than suggesting solutions, but my initial proposal is that we start charting the metrics on social flock cohesiveness and trend-leadership utilizing hyper-dimensional data models. First of all, any hope of making sense of micro-scale mass cultural behavior models will require new ways of comparative analysis. x,y line charts, pie charts, and bar graphs will not cut it. This is because one would need to observe too broad a scope of behavioral statistics. Connecting the fact that a trending joke like "That Shit Cray" over twitter to the value of purchasing a specific celebrity as the spokesperson for a shoe would require cataloguing bar charts and statistical data on all tweets as well as demographics and history of shoe purchases. This is pretty much the current standard model. But knowing that a sub-culture of Americans under 25 buy nike basketball shoes and that there is a connection between individuals who have tweeted "Cray" so that the Shoe company can specifically target tweeters that followed the "Cray" trend early on by tweeting the word with in the first month of the trend is a suggestion for the future. Recognizing these more specific demographic targets with specific target outcomes will both save money and reduce collateral damage so to speak in the way that laser guided tomahawk missiles are an advanced technology that surpasses and makes firing scud missiles obsolete. I think hyper-dimensional data models will allow the ability to simultaneously compare large amounts of data. We'll probably have to start using terms traditionally used for explaining quantum physics such as gravity, magnetism, super position, or quanta to explain the complex evolution of trends and other types of abstract data. Between the development of augmented reality and touch-tablet technologies we will be able to sift through hyper-dimensional data models in ways that we were never able to do in the past. Otherwise the suggestions about targeting specific demo's based on the language they tweet or the music they play on spotify would be just as much a matter of science fiction as manipulating a hypersphere representation of social flock cohesivity metrics with your hands using augmented reality glasses.
Now what happens when we start analyzing the value of social flocks and individuals as having higher trend-leadership capacity for a given product or brand? Advertisers will undergo a shift from buying demographics to buying trends. Advertisers will not only care about individuals like Kim Kardashian but they will care about all individuals. The first thing I started to wonder about this phenomenon was whether or not we will start to become our own individual forms of currency. Company's will give coupon discounts to person A because person A is a trend setter for person B and person C. While this thought is pretty gigantic I think it goes even farther by effecting our individual relationships. What if society knew that our values went up based on who's posts we liked and who's tweets we retweet. Our lives are already changing for the worse because they have become public but on top of our relatioships and behaviors becoming public they become a matter of business. What happens when people become friends with other people because of the number of followers they have and we talk to each other because of how much attention it potentially gets us? Oh wait, we already do? It happens for celebrities and bloggers and public figures now. It's only a matter of time before it is the way of life for everyone.
#nowtrending
-MM
**UPDATE**
I found this video on Text Mining. So since I read the Kim Kardashian article, companies have popped up that analyze text information. My suggestions about what we can determine is way ahead of this. Text Mining is only one part of behavioral trend prediction.
No comments:
Post a Comment