Programmatic media buying (PMB) has been touted as the future of media planning and buying, especially in the online digital segment. A September 2011 Forrester report ("The Future of Digital Media Buying," Joanna O'Connell and Michael Greene, Forrester Research Report, September 21, 2011) asserts that media professionals not engaging in programmatic buying will be in jeopardy of losing their jobs through obsolescence. I have recently spoken to over 25 digital media buyers from large agencies, all of whom told me that they were well aware of the benefits of and motivations for adopting PMB.
Yet, they haven't. Even after years of double-digit annual growth forecasts, PMB accounts for a meager 10 percent of total digital media buying. What allows this status quo to persist is the predictability and durability of the audience characteristics of digital media. An agency media planner, managing the launch campaign for an expensive running shoe for women, knows precisely which 5 web sites offer the best reach into wealthy female fitness lovers. Not only are these metrics unlikely to change in the near future, it is also improbable that new fitness sites will emerge to challenge the top 5. The confluence of these factors, known as the popularity persistence phenomenon, allows media buyers to stick to traditional negotiation-based manual buying. There is no compelling need for the buyer to go the PMB route to acquire inventory -- other than the theoretical promise of efficiency enhancement . Similarly, for the top publishers, there is no need to make their properties available through exchanges and real-time bidding systems -- they are secure in their prominence, feel no imminent threat from other publishers, and wish to keep control of their inventory. It is this duality -- the agencies and the publishers both secure in the status quo -- which I believe is the biggest impediment to large scale PMB adoption.
Imagine, if you will, an alternate media universe where these rules no longer apply. Here, high-ranking properties top the charts for very short periods: days and weeks, as opposed to months and years. Moreover, new properties emerge at high rates, and newcomers rocket to the top in no time at all -- often on their launch day. In such a universe, could an agency buy manually? No way!!. Here, our fictional shoe campaign planner could no longer find the 5 properties that provides the maximal reach into the target audience for the planning horizon, nor could she be sanguine that several other relevant top properties would not emerge in that time-frame. In such a volatile scenario, media buyers would be compelled to rely on a more reactive, spot buying scenario -- buying programmatically.
Publishers are greatly affected as well. No longer able to rely on creating properties with long-lived popularities and enduring reach, app publishers must adopt a strike-while-the-iron-is-hot approach -- ensuring inventory is available for buying during the relatively short periods when the property is hot -- achievable only through PMB systems.
Such media ecosystems are not imaginary -- they exist right now in the mobile app world. App popularities are notoriously volatile, and new apps break into top ranks at unprecedented rates. Here are some illustrative statistics from our scientists at Mobilewalla: in the ITunes Games category, the average stay of an app in the top 10 was just 6.52 hours. For Entertainment category the corresponding number was 4.4 hours. Over a period of 30 days, a total of 140 and 90 distinct apps made an appearance in the Top 10 for Games and Entertainment categories respectively. During the same time period, the top 5 fitness web sites remained unchanged. This pattern plays out in the entire app market, but especially in the audience-rich categories of Games, Entertainment and Lifestyle.
At Mobilewalla, we believe that apps represent a media ecosystem where programmatic buying is not only important, but also mandatory. However, the successful application for PMB in apps requires addressing a number of hard problems, mostly data-centric, that do not arise in the case of display environments. These relate to the issues of audience measurement and reach estimation, two key factors in making advertising decisions.
Audience measurement in apps is notoriously hard, as traditional panel-driven methods do not apply. Reach estimation is equally complicated owing to the transience of app popularity and the need for real-time tracking of this vast universe. Traditional digital media research companies as well as existing DMPs are not set up to perform the type of big-data collection and analysis that is required to compute and deliver these data signals essential to performing effective in-app advertising.
As a result, big-data ventures like Mobilewalla are inventing entirely new classes of techniques to perform real-time data collection, non panel-driven audience measurement and model-based reach estimations. The understanding and adoption of these signals have the potential to unlock the great promise of reaching engaged audiences via mobile apps.