Are you ready to have an algorithm do your job? Ready or not, it’s happened. Late in 2017, Google moved all-new app install campaigns over to Google UAC (Universal App Campaigns). About a month later, they turned off any Search, Display and YouTube app promo campaigns that had been running. If you wanted to do mobile app install campaigns on Google, you were going to do them via Google UAC. And that meant you were going to let Google’s algorithm manage many of the functions in your campaigns.
Facebook didn’t take long to follow suit. Their move happened in February 2018, when they rolled out what we call their “best practices” update. Facebook’s changes weren’t as forced as Google’s, but they were just as consequential. The shift has meant, basically, that we’ve handed over most of the control of some of our primary advertising tasks to the algorithms. This does mean we’re freed up to focus on things like creative and audience expansion. But we’ve also given up a lot of control.
Fortunately, giving the algorithms this much control – and having algorithms sophisticated enough to do this work in the first place – actually has a lot of upsides. Now that so much of the work of a user acquisition manager has been moved over to the algorithms:
It’s possible for less-experienced advertisers to get results comparable to their more advanced peers. This means more advertisers can profitably use the platforms.
Many (if not most) third-party advertising tools have become less necessary. Ad tech tools had been a significant competitive advantage before, but now both Facebook and Google UAC basically offer comparable tools for free.
Most of the companies we work with used to spend 10% of their budgets on Google. Now they spend 30-50% Google. A little competition is always a good thing.
Prior to February 2018, Facebook allowed advertisers to run a lot of ads. The ads could have overlapping audiences, and there were no penalties for making frequent bid changes, even if they were made every few hours. Ads could be paused and budgets could be modified all the time.
Our internal AdRules tool allowed us and the companies we work with to edit bids, budgets and pause rules with the speed and precision of a high-frequency trading desk. We would get early indications of how an ad or ad set was performing, and then modify settings to either maximize that ad or ad set’s exposure or to kill the ad/ad set if it was underperforming. We can still do all that and more with AdRules, but how we use that tool has shifted.
So basically, prior to February 2018, optimization was done through thousands of actions that were controlled by the advertiser or a third-party ad tech tool.
Then everything changed. All those micro-management changes started to incur penalties, and it became clear that Facebook would reward advertisers for running their campaigns according to the best practices outlined in Facebook’s “Blueprint Certification.” Those best practices include:
“When marketers advertise with Facebook, they want to build campaigns that ultimately drive efficient return on their ad spend. Yet managing and optimizing their campaigns previously required using proxy metrics such as clicks, impressions, and installs to gauge whether or not a campaign had driven meaningful business impact. Today we’re introducing value optimization so that marketers can focus their campaigns on anticipated purchase value.
Value optimization works by using the purchase values sent from the Facebook pixel to estimate how much a person may spend with your business over a seven-day period. The ad’s bid is then automatically adjusted based on this estimation, allowing campaigns to deliver ads to people likely to spend more with your business at a low cost.”
This makes sense; it’s revenue we want – not necessarily just app installs. So if the algorithm can crunch the data well enough to show us not just who’s most likely to install an app, but who’s most likely to purchase from it, that’s a big help.
This also illustrates how the new algorithm-driven customer targeting allows us to vastly expand our audiences. If the algorithm is smart enough to handpick people, there’s no need for narrowly-defined audiences. Cast the net wide, and give Facebook or Google plenty of people to choose from. They’ll find the right ones.