In June 2017, Facebook launched the value optimization campaign strategy to allow advertisers to target high-value audience segments without needing to constantly create and update custom and lookalike audiences. The value optimization campaign strategy differs from the app event optimization campaign strategy, which was launched roughly one year earlier, in that it optimizes for magnitude of revenue (“value”) versus merely the incidence of some event: with VO, an advertiser seeks to reach users that are expected to not just make purchases but to make some combination of purchases that produce the greatest extent of overall value. For this reason, VO campaigns can’t be cost constrained in the same way as MAI and AEO campaigns; they can merely be constrained in terms of ROAS with the min-ROAS setting.
In this article, I provided a list of key innovations that Facebook has brought to its ad platform over the past five years. In taking note of the chronology of these product features (which could also include Dynamic Language Optimization and min-ROAS bidding), it’s easy to spot a trend: Facebook has consistently shifted targeting, creative variant production, and campaign and ad set optimization away from the purview of the advertiser and onto its own tool suite. With the launch of min-ROAS bidding for VO campaigns, Facebook has reached something of a logical limit to the way in which it can present three individual, separate levers (campaign management, creative management, and targeting) to advertisers to utilize in operating their campaigns: at some point, future product release are going to force those levers to converge. The question is, how?
It’s likely that Facebook will take its product cues from Google in this respect and attempt to make its platform look more like UAC, Google’s Universal App Campaign system. To understand why this is the case, it’s first important to understand what exactly UAC is.
A brief history of Google UAC
Google launched UAC in 2015 as a means of helping advertisers to automatically target audiences across Google’s portfolio of placements. Anecdotally, it wasn’t very popular: most advertisers preferred to manage campaign budget allocation themselves, principally because of substantial, fundamental differences in quality between YouTube, Google Display Network, and Search inventory. In October 2017, Google announced that it would fold its AdWords inventory on mobile into UAC, meaning that advertisers would cede almost all control over creative design, placement optimization, and audience targeting to Google. Predictably, something of an advertiser revolt transpired, although ultimately mobile advertisers acquiesced to the change: what choice did they have when Facebook and Google dominate digital advertising?
Google explained the benefit of UAC as a shift away from “one-dimensional proxies” to “multi-dimensional signals” in terms of audience targeting. What Google meant by this is that Google has the ability — because of its superior machine learning technology and its massive proprietary data set — to combine vast numbers of features together in evaluating potential traffic in a way that is more specific and thus efficient than what almost any advertiser can do. When an advertiser buys traffic on the basis of a quality proxy derived from some small set of dimensions, such as “US iPhone users” being “worth” $5, they are using broad averages that inherently overstate the value of much of the traffic they’ll ultimately receive (more background on this idea in this article and this article). Google’s point was that it has more data, and it is better at utilizing that data, than almost every mobile advertiser — so why not let it handle all aspects of campaign management?
What advertisers bristled at with UAC was its lack of transparency, particularly around creative and placement performance. UAC is an asset-centric system, meaning that advertisers don’t provide it with complete ad creatives but rather “components” that might serve as standalone ads or might get combined with other components into synthetic creatives. The benefit of operating the platform this way is that an advertiser only needs to produce the atomic units of ad creatives, and Google can use those to create every possible creative variant needed across all of its placements.
The downside of this, combined with opacity around placement performance, is that an advertiser can’t really know what worked in their campaigns, or why: they can merely feed Google’s machine as many components as possible and trust that the machine is benevolent. Some advertisers try to game this dynamic by using high bids in conjunction with limited component types to try to force Google to only serve to specific placements, but at some point the tide always turns: Google operates the dominant advertising platform in the world, so it’s perhaps best to use it as intended.
Another issue with Google UAC is that the placements identified above all vary quite drastically in terms of visual fidelity / information density. A well-targeted, well-crafted YouTube ad should convey more information about a product than a text ad; it is logical that YouTube inventory should, on average, be more expensive than eg. search ad inventory. But because Google operates across a very long lookback window in calculating campaign performance metrics (“its algorithm moves slowly”) and because it creates so many different variants of ads via component combination, updates to campaigns tend to not create visible performance changes quickly unless those changes are dramatic (ie. they revert the campaign back to the learning phase).
Because of this, advertisers may have to make very large changes to bids or budgets in order to break out of a type of advertising purgatory: poor performance because the advertiser’s assets have been relegated to the lowest-value inventory. This situation is not only frustrating to an advertiser, but “bidding up” a campaign to force experimentation with higher-quality inventory is also risky: if the better inventory doesn’t actually deliver better performance, the advertiser just conducted a very expensive experiment.
Given the thread that runs through Facebook’s contemporary ad product releases — automated creative variation, automated value targeting, automated budget allocation — it is hard to believe that these systems won’t be combined into a UAC-like black box at some point. The UAC approach is simply so much more lucrative for an ad platform: it controls where and how budget is allocated and thus could, potentially (and cynically) provide advertisers with just enough performance to justify continued spend, but no more.
The VO campaign strategy combined with min-ROAS bid strategy is an excellent example of this dynamic at work. Because VO campaigns are “greedy,” meaning they need to experiment on as large of an audience as possible, constraining VO campaigns with very specific audience definitions will either handicap the advertiser in the auction or result in quickly-deteriorating performance as the audience gets saturated. Every advertiser knows who the obviously high-value users are; the point of VO is to allow Facebook to find new ones for an advertiser. So with minROAS, because the advertiser can’t set a cap on cost, Facebook simply delivers traffic that meets the minimum threshold for ad spend recoup, without any guidance as to who was targeted, or why. Facebook owns that insight — maybe it could have targeted users better, or maybe not, but the advertiser will never know.
It’s obvious why the black box, totally automated ad platform paradigm is beneficial to platform owners: it virtually guarantees maximized profits as the platform strives to deliver minimally to a goal, and it also helps small advertisers get onboarded very seamlessly. And it seems almost inevitable that Facebook would evolve its platform in this direction — because where else can it go? The “Facebook UAC” is just a combination of dynamic creative, automatic placements, dynamic budget optimization, and AEO with a cost cap or VO with a min-ROAS setting. It’s fairly easy to make the leap from a campaign with these settings explicitly selected to merely “a campaign.”