Advertising has always been a blend of art and science. Even in Q1 2020, as user acquisition advertising goes through massive change, performance is still about art and science.
Or more specifically, about algorithms and creative.
Every few months for the last few years, the machine learning algorithms at Facebook and Google have rolled out significant new updates. These updates have made the platforms’ algorithms increasingly more efficient at managing bids, budgets, placements, and audiences.
At some point last year, the algorithms tipped the scale. They became better than humans at managing many parts of user acquisition.
So, user acquisition managers can now automate more of their work, but they also have less control. Levers user acquisition managers had previously used for campaign optimization are gone.
This has fundamentally changed how UA advertising is done, and it’s time advertisers changed with it. Not in incremental, halting ways, but in step with the scale of these changes. Because “the rise of the machines” – the algorithms, that is – has sweeping ramifications of UA managers, the social ad industry, and third-party ad tech tools. It requires us to advertise very differently than we did even six months ago.
The most significant consequence of all of this, and the thing that will shape our industry more than anything else, is that creative is the most important advantage that social advertisers have left. Whoever builds the best creative and understands how to efficiently test and identify winners will dominate Facebook and Google advertising in 2020 and beyond.
Is that overstating the changes? Is it hyperbole to say “UA is dead” or “ad tech is dead,” or that user acquisition managers need to rethink how they approach media buying?
We don’t think so. And we think you’ll agree with those statements after you’ve finished reading our perspective.
But first, let’s do a quick recap of how we arrived at “user acquisition is dead”. It will help us understand where we are in Q1, 2020 and how rapidly things can change.
Back in ancient times, like early 2015, UA managers chased volume. We optimized for app installs and created hundreds, even thousands of campaigns and ad sets so we could test every possible variable in a campaign. We set up the intraday bid and budget ad tech solutions that rivaled high-frequency stock trading desks.
We were the quants.
Then the algorithms came of age. Google launched App Campaigns (at the time, under the name “Google Universal App Campaigns”). Overnight, all app advertising campaigns on the platform were automatically switched over the new campaign type, and the Google Ads’ machine learning algorithm took over the controls of budgets, placement, bids, and audience selection. They created an efficient social advertising black box.
If you were okay with automated campaigns, even if it meant you had far less control, this was all good news. As Google put it at the time.
“All you need to do is provide some text, a starting bid, and budget, and let us know the languages and locations for your ads. Our systems will test different combinations and show ads that are performing the best more often, with no extra work needed from you.”
Facebook took a different approach. Instead of modifying their algorithm in one sweep, they have incrementally automated certain levers, and they’ve usually made those changes optional.
But Facebook, like Google, has been moving toward algorithm-controlled media buying for a while. AEO (App Event Optimization), VO (Value optimization) and LTV (lifetime value bidding) were some of their first steps. (Google implemented their own Value bidding in August of last year.) Advertisers could specify those types of optimization goals and then let the algorithm figure out how to best get the most financially viable new users for the price set by the advertisers.
Then around February 19, 2018, Facebook introduced the idea of significant edits. Suddenly, those frequent intraday changes we had been making were no longer a good thing. Instead, they triggered what Facebook calls “the learning phase,” a campaign status where the algorithm had to re-adjust. Slowing the algorithm down by triggering the learning phase could suppress ROAS or more and reset their learning engine.
Throughout the latter part of 2018 and 2019, both Facebook and Google continued their incremental progression toward automated media buying. These changes were primarily focused on budget/bid management, ad placements, and audience selection.
Inch by inch, they’ve been moving us towards fully automated user acquisition media buying.
While they’ve been very successful in automating media buying, they have not cracked the code on creative automation.
Sure, both platforms took steps toward automating creative testing — Google with its ad variations, and Facebook with its Dynamic Creative. And while those new tools are helpful, both features were in essence just taking pieces of creative and re-assembling them into new variations using multivariate techniques.
At least for now – and probably for the next few years – humans are the primary drivers of creative ideation production and optimization.
As machine learning algorithms improved with the bid, budgets and placements, all the “features” that had been rolled out started to gel into a new way enabling near fully automation media buying.
In mid-2019 Facebook rolled out their Power5 recommendations, and then a few months later enhanced them further with their Structure for Scale best practices. That was the signal to advertisers that these features were no longer a series of experiments: We had crossed over into a new kind of automated advertising and we all needed to prepare. The roll-out of Campaign Budget Optimization got more attention than the earlier updates but was just more evidence that the algorithms were preparing to automate. They could manage certain parts of UA advertising better than people could and did so without bias, rest or complaints.
Now that Facebook has rolled out CBO, we’ve moved one step closer to total automation. Humans no longer pick bidding strategies. And really, the most powerful lever UA managers have now in Q1 is which optimization event they target (purchase, registration, level achieved, etc.). Being able to pick which event you optimize for in the sales funnel has a powerful impact on campaign performance, and is an elegant application of Facebook’s Structure for Scale best practices. We expect Google Ads to make a similar update in the first half of 2020.
So automation isn’t just coming – it’s here. If you’re doing UA advertising the way you used to do it even six months ago we know you could be optimized better.
Facebook may have implemented a lot of changes in the last few years (as has Google), but they gave us a valuable new playbook last year in their Structure for Scale best practices.
Basically, Structure for scale is designed to optimize campaign setup and management in a post-AI advertising world. It’s a set of best practices advertisers should adopt if they want the Facebook advertising algorithm to operate at peak performance. Many of the recommendations for Facebook listed here can also be applied to Google App Campaigns. Structure for Scale is an evolution of the Power 5 recommendations Facebook released earlier in 2019 and honestly, shows how quickly things are changing.
Those best practices (at least for now) include:
Here are just a few examples of the performance gains some advertisers have seen after adopting the Structure for Scale best practices:
So the benefits of Structure for Scale are clear. Here’s how to apply them:
Basically, Facebook wants you to KonMari your campaigns and ad sets. To go from this:
Why? Because it gives the algorithm more room to learn and to more efficiently optimize.
To work well, the Facebook advertising algorithm needs enough data to crunch. This is core to how machine learning works, and it’s a concept you must grok if you want to advertise successfully now and in the future and believe in the process.
Here’s why this is such a big change, and why it forces UA managers to give up so much control: Before we had an algorithm to manage campaigns, it made sense for human ad managers to create many (sometimes hundreds) of campaigns to control all sorts of variables – bids, targeting, audiences, placements. This lets us test settings, isolate audiences, and manage many other levers of campaign management.
One of the ways to improve the algorithm’s performance is to simplify account structure by minimizing how many campaigns and ad sets you have.
Another way to improve performance is to use Campaign Budget Optimization to free up how efficiently your ad budget is spent. And yet another way to give the algorithm the flexibility it needs is to let it pick ad placements so it can test your ads across the 14 different placements available.
This means, of course, you’re giving up control. Some advertisers don’t like that. But here’s how Facebook views this “control” versus “algorithmic optimization” quandary:
Which would you want: More conversions or more control? Personally, if I could be assured the value of those conversions is as good as what I was getting before, I’d go with more conversions.
But what happens if you can’t cede control? If you insist on running your campaigns the old way?
Then your ads will stay in what’s called “the Learning Phase” or “Ad Set Calibration” longer because the algorithm can’t crunch data in the way it was built to do.
Staying in the Learning Phase for any longer than necessary is not a good thing. To get out of it, each ad set needs to generate approximately 50 unique conversions per week. To stay out of it, advertisers need to not make any “significant edits.”
In a way, the learning phase and the significant edits that trigger it end up being almost a “human-management tax” on your campaigns. Facebook isn’t overtly punishing us if humans intervene in the algorithm’s work, but because the algorithm is now running the show, anything that impedes its work does tend to impact and slow down learning and performance.
This is the overarching principle in UA advertising right now: To remove limitations on the algorithm in order to gain conversions. Part of that means setting campaigns up in a way that allows the algorithm to do its work, and part of it means minimizing how much tinkering we do with the system once it’s running.
While UA managers have lost a lot of control, they do still have a few key levers to guide their campaigns with.
Facebook recommends four ways to do this:
Facebook has 14 ad placements across its family of apps right now. Trying to manage them is complex, to say the least. So it’s really better to let the algorithm manage placements, especially when you add in other performance factors like ad sizes.
Still don’t want to give up control? Consider this: Facebook says shifting to automatic placements reduces the cost per conversion by 71%. But if you just really must have control over where certain ads appear, use “Asset customization.” It will let you choose which images or videos people see in your ads depending on where your ads appear.
Optimizing performance across placements requires different creative formats, of course. Which is why Google has its own best practice recommendations for creative assets, both for the types of formats you use and for the number of variations you use for each format, as shown below:
Google also recommends creating video ads in multiple lengths and aspect ratios. This gives the algorithm lots of different assets to use as it tests where your ads will perform best.
Even if you minimize control on every other aspect of campaign management, without freeing up your budgets, the algorithm is still constrained. So:
The days of human-managed bidding are over. It’s vastly better to let machines manage the complexities of changing bids. But choosing a bid strategy is still best left to humans. Here’s what we get to pick from:
If at all possible, bid according to lifetime value. It makes no sense to bid like all your prospects are created equal. They aren’t.
For niche or higher-value products with fewer purchases, try moving your conversion event up towards the beginning of the funnel. This usually gives the algorithm more data points to work with because for every step of the funnel toward the purchase, there are fewer conversion events.
For example: Because we can have the algorithm to optimize for specific events (like in-app purchases, registrations, or specifically, $20+ worth of in-app purchases), Facebook will deliver a far more precise audience.
When using AEO and VO, we also allow the algorithm to automate placement, bid management, and the creation of audiences. This is exactly where that afore-mentioned leap of faith takes place: UA managers should now let the algorithm manage these key areas of their campaigns.
See our article about Structure for Scale best practices for more details on automated media buying.
So that’s how to manage campaigns now that the algorithms have taken over. Or, maybe to put it more accurately, that’s how to let the algorithms manage your campaigns in Q1 2020.
We expect the best practices to change, possibly even in this quarter, as Google and Facebook continue to roll out more updates and features that further establish how to do automated UA advertising.
So now you know how to work with the algorithms for optimal campaign performance. But the evolution of Google and Facebook’s ad platforms have had one other massive consequence: They’ve made third-party adtech tools unnecessary.
We’ve talked before about how “adtech is dead.” It’s dead because everyone now has access to world-class advertising tools. They’re built right into our Google and Facebook advertising dashboards. Expensive third-party adtech tools are no longer as much of an advantage.
This means that any competitive advantage you’ve been getting out of your single-network adtech is gone… or, at least, its advantage is diminishing rapidly, even daily, as
1) the advertising platforms get better and better and
2) more and more advertisers figure out how to use those platforms’ new, very powerful features.
Optimizing these platforms has one other major consequence: Because Facebook and Google’s ad platforms are increasingly automated, they’ve made it possible for almost anyone to advertise profitably. Increasingly you don’t need to be an advertising whiz to get good ROAS.
This means there may be more competition as many of the smaller advertisers (or even very small advertisers, like local businesses) compete on a more level playing field. Automation may end up being the best move Google and Facebook have ever made to grow their advertising user base.
So user acquisition has been being slowly taken over by algorithms for quite a while. It’s well past time for user acquisition managers to embrace that. Part of that means managing campaigns differently, as we’ve outlined above. And part of that means pivoting into roles and skills where humans can still outperform machines.
Those areas are, namely:
Yes – part of it. We’ve reached the tipping point where machines can do some things better than a human can. So if the bulk of your time has been spent running reports and poring over spreadsheets to find small pockets of opportunity, you need to expand your skills. The machines can simply do this faster and better than people can, and by several orders of magnitude.
While the algorithms may be great at crunching data, they still can’t do creative. Creative strategy, competitive analysis, user motivation or player personas analysis, and intelligent creative testing – that’s all beyond their ken.
So start building your skills in those areas. Here are two of the best ways to do that:
1. Learn how the machines do their optimization work, so you can manage them appropriately.
Some experts have compared this to a pilot flying a plane. The pilot has this huge dashboard of data inputs they monitor, even though the plane automates a lot of its own systems. But there’s still a keen need for a human to be there, making sure the machine takes appropriate actions.
The human is there to overcome the primary weakness of the machines: The algorithms only do what they’ve been coded to do based on patterns they’ve seen in the past. If you give an algorithm a dataset unlike anything it’s ever encountered before, it chokes.
2. Prove your value to your employers or your clients in new ways.
You won’t be making bid and budget edits or sifting through thousands of audiences and ad sets all day anymore.
Don’t mourn this. You’ve got better things to do – like developing better creative.
So let go of the geeky intra-day campaign management and go have some fun with creative! In this new algorithm-driven environment, creative ideation and testing is the best way to deliver value to your company and your clients.
Just don’t take an “us versus them” view of algorithms and humans. The good news is that when the algorithms’ campaign management capabilities are partnered with a human creative and creative strategy, the combination is very powerful.
We’ve finally got two bits of intelligence (human and machine) each doing what they do best. Balancing and leveraging these two ways of seeing the world is the key to effective user acquisition advertising right now.
And it doesn’t just apply to you. As a UA manager, it’s part of your job to get your whole team aligned with this new reality. So here’s what we see for UA teams in 2020.
This shift towards automation is obviously going to take some work away from user acquisition teams.
So yes: Expect some layoffs and reduction in user acquisition team sizes in 2020. Even Uber laid off a third of its marketing department this summer. We suspect this is driven at least in part by new automation capabilities.
Unfortunately, the automation driving those layoffs won’t just affect a small group of well-funded advertisers who have access to expensive adtech. This could affect the entire user acquisition community, including ad agencies, marketing consultants, and in-house marketers all over the world. In fact, automation could force the internationalization of UA to talent in less expensive markets.
It simply takes less work and expertise to manage digital advertising for small to medium accounts than it used to – that means there’s less work to do.
Again, you’ll be relatively safe if you pivot to creative strategy, creative competitive analysis, and creative testing and optimization. Or if you become a specialist in how the algorithms work and in the new features the platforms are offering (like Google’s Pre-Launch tool).
But everyone’s going to have to level up, and that includes employers. As business owners consider their own UA team now and for 2020, there may be an opportunity to reduce some overhead or hire less experienced media buyers who are less expensive.
So leaner UA teams are probably the future… and we mean “the future” as in Q2, 2020. But it’s also possible we’ll see UA teams merge with growth and data science teams.
For a long time, we’ve built a bit of an organizational wall up between UA teams and growth teams. The functions of these two groups were fairly separate: UA teams got the customers, while growth or product teams tested the app for better engagement, monetization, and lifetime value.
But is that separation necessary – or even beneficial? As advertising platforms simplify so advertisers provide a user profile, LTV model and then send the algorithm out to get that type of customer, might it make sense to overlap or merge the staff and the functions of UA and product teams?
Time will tell. But given the pace of this industry, we won’t have to wait long.
So that’s where we see user acquisition right now: Increasingly run by algorithms, without the old competitive advantages of adtech, and requiring a pivot to creative as the top priority.
Over the last two years, as we’ve been watching the machines get better and better at optimizing the quantitative side of advertising, creative has surfaced as an answer to many of the problems UA managers face.
So we don’t have the old advantage of adtech, but if you pivot your focus to creative, you can still squeeze out the competition. And if you don’t want to lose your job to an algorithm, then pivot to a skillset the algorithms can’t compete against.
Creative is the best advantage a UA manager has in terms of campaign performance.
Creative is the best shot we’ve got for success… but not just any creative will do.