The duopoly’s shift into event-based bidding models for mobile marketing campaigns has caused much consternation amongst mobile developers around the loss of both control and transparency afforded to them in their advertising bidding strategies. As the duopolies (and other mobile advertising broker networks) transform from mere gatekeepers of inventory into algorithmic delivery services — absorbing all of the targeting and bidding responsibilities that used to be the sole domain of mobile marketers — the changing mobile marketing landscape engenders uncertainty around the efficacy of these algorithms for identifying appropriate app audiences as well as the role mobile marketers should be playing in a world where algorithms do the heavy lifting of campaign optimization.
Mobile marketers must accept a few tough truths about the changing marketing terrain. First, algorithms are objectively better (more efficient, faster, etc.) at targeting and making bid adjustments than humans. This is unassailably true; campaign optimization is precisely the kind of work that humans should want to cede to algorithms. Second, this is the new status quo: algorithmic bidding is not going away. Given these two facets of mobile marketing’s new reality, how should the mobile user acquisition function evolve to accommodate it?
In order to arrive there, it’s helpful to put the current state of mobile marketing into historical context so as to understand how these changes fit a broader marketing framework. If the 2012-2015 era of mobile was dominated by the “LTV > CPI” paradigm, then how do the components of that way of thinking break down into atomic units that can be reconstituted to form the contemporary approach to mobile advertising? With algorithmic campaign management, the marketer is deprived of the targeting levers that powered the LTV > CPI model, which dictated where and to whom ads were shown:
In the above model, targeting parameters are static and measured for performance via the LTV / CPI spread produced by users from some demographic. If the cost per install of users from that demographic exceeded their lifetime value at some bid level, then the bid was changed or the campaign was shut down.
In the current model, the traffic providers handle targeting and optimize that against the events that are transmitted back from users in those cohorts. So while the developer can’t define targeting parameters, it can create many types of ad creatives and measure the performance of those against specific events to guide the delivery algorithm toward the most effective traffic:
It’s important to note that, even with algorithmic bidding, targeting doesn’t go away: it just moves up and down the funnel into places that it can pre-empt or assist the delivery algorithms. Creative is the new targeting domain; ad creative now determines which demographic groups are most likely to click, install, pay, etc. via relevancy. In this new model, the marketer has essentially shifted her focus to the inputs (ad creatives) and the signals (events) that produce the desired level of return on ad spend (LTV and CPI are deprecated in this model because the marketer has shifted her attention away from costs) for some level of budget.
And thus the marketer is now both a creative director and a data analyst, using the black boxes of UAC, Facebook’s AEO and VO models, and the retention-oriented models of broker networks as fulcrums to produce optimal return on ad spend (note here that budget management becomes much more important in this paradigm as very expensive traffic could very well produce the highest relative ROAS but be out of reach for budget-constrained advertisers). The mechanics of the new operating model are better, but the process is merely different; just as the proliferation of video in 2013-2016 changed the reality of mobile marketing, the rise of algorithmic bidding necessitates a change in approach from mobile marketing teams.