Commentary
New Ad Model: Minimum Motivational Frequency
- by Dave Morgan , Featured Contributor, March 11, 2010
The next five to 10 years will bring enormous changes to the television media and advertising world. Not the least of those changes will be the development and use of new models to optimize the purchase of TV ad spots.
The Internet has shown marketers the power of delivering their messages to people, not just pages, and measuring the actual result of each and every impression. The same model is coming to television, where the availability of massive amounts of anonymous set-top-box viewing data is now enabling TV networks, advertisers and their agencies to begin to separate people from the programs they watch, and start adjusting media and marketing to viewers in ways that were not possible before. The concept of reach and frequency will never be the same.
When you focus on the people watching ads, not just the programs where the ads are presented, lots of things change. There is no better place to see this than watching how people respond to on-air program promotions, since anonymous set-top-box data permits you to measure the actual impact of the results with near-absolute certainly, and very quickly.
advertisement
advertisement
Using my company's access to anonymous viewing data from more than 15 million U.S. set-top boxes, we try to understand how people view television programming, and how to best inform them about shows they might enjoy.
One of our most important findings -- though not surprising -- is that different people view television differently, and they have different thresholds of responsiveness to program promotions. These thresholds are very predictable across segments of people according to their historical preferences for genres of programs, or loyalty to certain networks, or loyalties to watching television on certain days and at certain times.
Some segments of viewers only need to see one program promotion within 24 hours of the airing of a program to dramatically increase the number of those viewers that will actually watch the show. Some other segments of viewers might need to view three or four on-air program promotions within several days of the show to increase their likelihood to view. And yet other segments won't increase their likelihood to view a show no matter how many promos they see.
Different segments of people require different "minimum motivational frequency" to cause them to change their behavior, or to reinforce historical behaviors. In television of old, there wasn't very much that you could do to either discover or exploit the concept of differential minimum motivational frequency. Our research has shown that legacy modeling methods like demographics are lousy predictors of people's motivational thresholds. As you can imagine, delivering on-air program promotions optimized to the minimum motivational frequency of the recipients will result in significantly more effective program marketing, and dramatically less waste. Further, if it works in program promotion, it certainly should work for most marketing categories, from consumer packaged goods to consumer electronics.
What do you think? Will concepts like minimum motivational frequency eventually supplant more generalized media models like reach and frequency? Why shouldn't all ads, across all media, be optimized that way over time?
I think that metrics like these are going to change the fundamental way in which TV time is purchased. Potentially it will impact the fees that agencies charge their clients. We may well be entering a world where agencies who can sort through all of this will be worth a premium rather than the commodoties they are currently considered. I linked to your article with more comments in my blog post at http://ow.ly/1hkAR
What do you think if there was the substitution of programs/advertisers to ideologies/theocracies? How many times would a view move to engage in questionable behavior? Who is going to control it? The deliberate "meaness" of some on the Supreme Court should give pause. Unintended consequences can have momental consequences.
Bruce ... great blog post. I am hopeful that we'll see that kind of change.
Similar to browser sessions on the internet, if back end systems like yours are able to dynamically figure out the profile of the user controlling the remote (housewife, the son..) which would then enable "ad servers" or triggers sent to PVR to play the appropriate pre-populated ads (based on trending) stored locally would shift the game to device/person based rather than channel based - aka behavioral targeting (your forte) vs contextual targeting.
Actually, while the TV infrastructure may some day be able to support Behavioral Targeting, I am not sure that either public policy or the economics will ever fully justify it. Delivering ads to large, anonymous segments of users based on their probability to be interested in those ads will probably deliver returns close to those of profile targeting, but without the costs to build out fully addressable targeting technology systems and databases and without the public policy costs of targeting viewers in ways that they may not feel comfortable.
Dave, good article and I agree Bruce's blog post is also good. But what really grabbed my attention with a "YES!" was your follow-up comment: "Delivering ads to large, anonymous segments of users based on their probability to be interested in those ads will probably deliver returns close to those of profile targeting, but without the costs to build out fully addressable targeting technology systems and databases" You can get away with saying this, but most people can't, lest they get branded as dinosaur-era heretics, but nevertheless, I believe you're absolutely right.
I have long pondered whether, when addressable TV finally gets here and brands pay a premium to use it in order to minimize waste, the direct response people will be buying the leftover, non-addressed, "waste" inventory at a discount, and laughing all the way to the bank as they watch it deliver a premium ROI.