App Marketers Compelled To Embrace Automation – And Give Up Control

The robots are taking over user acquisition on Facebook and Google, and that’s (mostly) a good thing.

Whenever one of the large platforms automates an aspect of UA, it means more time for focusing on strategy and creative rather than drudgery, said Jess Hasenplaugh, head of UA at Berlin-based mobile game studio Wooga. “I’m happy when the stuff I get tired of doing manually gets automated,” she said.

Wooga spends the bulk of its budget on Facebook and Google AdWords, around 35% and 40%, respectively.

But set-it-and-forget-it buttons come with a loss of control.

Earlier this year, casual gaming studio FreshPlanet put all its budget against a test of app event optimization, a Facebook product that lets app marketers target install ads to the people most likely to take specific actions. Facebook claims the tool generates the best possible return, but FreshPlanet said it saw results when it manually controlled the bidding.

Manual control is not where the industry is headed, though.

For the last few years, both Facebook and Google have been layering more automated functionality into their UA offerings to streamline budget allocation, automate ad placement and adjust bids on the fly.

Automatic placements are par for the course across Facebook and Google, and both have AI tools that optimize ad delivery toward the users most likely to spend money in an app after they download.

Facebook rolled out campaign budget optimization at the beginning of the year, and, in 2017, Google started requiring that all paid app-install ads run through Universal App Campaigns, which uses machine learning to automatically decide when and where to run app campaigns for the best return.

Advertisers have a few levers they can pull to exert control over their bids.

Marketers can devise little hacks to get around the automation. It’s not possible to manually exclude YouTube placements through UAC, for example, but if marketers don’t want to run video ads on YouTube, they can simply make sure not to upload any horizontal video as part of their campaign, Hasenplaugh said.

App marketers can also target by language and decide what metrics to buy against through UAC – either cost per action or target return on ad spend – or set minimum spend against ROAS for ads targeting potential whale spenders on Facebook.

But now that automation is the name of the game, the primary lever at the marketer’s disposal is the creative, said Brian Bowman, CEO of Consumer Acquisition, a Facebook and Instagram marketing partner that mainly handles UA campaigns.

“As the duopoly’s algorithms get better, creative is the next logical thing that can provide the best and biggest bump for an advertiser’s performance,” Bowman said.

Over the last year or so, Consumer Acquisition has been rejiggering its platform to focus more on creative services and transitioning its creative marketplace from static ads to video. It’s a necessary move for the company and others like it.

Consumer Acquisition, like many of its competitors, got really good over the years at gaming the Facebook ad auction. The strategy hinged on running an enormous volume of ads – in some cases thousands of ads simultaneously – against hypersegmented audiences that often overlapped and on changing up the bids multiple times per day depending on performance.

In that way, the company was able to build a complex system to game the Facebook algorithm – “it was not unlike a stock market trading platform,” Bowman said – and it worked really well for clients … until it didn’t.

In mid-February, Bowman and his team noticed that their campaigns were suddenly tanking. Facebook had tweaked its algorithm to reward broader targeting, fewer ad sets and fewer intraday campaign changes. Advertisers that followed those new best practices started to see way better results and others were penalized.

“I think there’s going to be a significant fallout in the coming months for companies that focus on ad tech solutions for the duopoly,” Bowman said. “Native platforms are always going to do optimization the best, but that makes creative king right now – and the good news is, winning creative can’t be generated through machine learning or AI – at least not yet.”

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!

1 Comment

  1. Defnitely agree that in principle automation is a good thing because it has the potential to free up time from repetitive work to spend it on more value adding work.

    However, there is a difference between letting a platform take care of automation vs taking away control and transparency.

    I believe the basic ground layer any ad platform should provide is transparency around the choices automated systems make. Because only with that transparency advertisers can a) learn about what works and what doesnt (and inform more macro choices) and b) be confident that the automation system is actually acting in their best interest. The latter point especially becomes important when the incentives of the vendor and the client are not fully aligned.

    After solving for transparency, control should form the other basic layer of an ad platform. Prima facie, there is a potential trade-off between control and automation. But I think if you take a closer look the perspective changes. Its quite possible to have a system fully automated while still giving users control. And even more so, this system can be superior in delivering results for the customer.

    There will always be pieces information which algorithm(s) don’t have at their disposal and where a human brain can literally connect the dots and make choices that are better than the algorithm choices. This requires the person in charge of these choices to be very aware of how the algorithm works, to assure the levers pulled are in harmony with the algorithm rather than “against it”.

    I believe this is where the real root cause of this problem sits: the algorithms at play are not well exposed and transparent (often considered to be “secret sauce”), so even an informed person making choices on top of algorithmic decisions can not play along well with the algorithm.

    Transparency, control and automation can live in harmony with each other, if the product is designed to solve for this. I just doubt that the teams at google and facebook really care about it enough to design their products that way. Or to put it in other words: they optimize for their own goals, which are not always in line with the goals the clients optimize for. A common example I have often heard is google UAC diverting spend to youtube even though its eCPIs were terrible. For google, this makes sense because they want to optimize yield across all their inventory, including those inventories that don’t perform that well for direct response.

    Now, what I am really curious to see after the release of iOS 14, is if automation will still be the name of the game. Because the feedback loop between campaigns and down funnel KPIs will be less tangible and direct within Apple’s SKAdnetwork. Also, within SKAdnetwork Apple is the authority which decides on attribution and thereby destroys the inherent advantage walled gardens have built up with their self-attributing systems.

    How many of the post-view conversions Google calls “engaged clicks” (effectively treating a view as a click in its self-attributing systems, a relatively big share of app installs that come from youtube) will Apple still accept as clicks? Self-attributing networks (at least on iOS) will have operate in world where grading their own homework isn’t enough anymore to pass the exam.