
Last week, I had a chat with Maria Mitsostergiou, one of the Drum’s Top 50 Most Creative & Innovative Women working in the Search Industry. Having managed app marketing campaigns for a string of global brands including Shop Direct, The Economist, New York Times, LeoVegas and myLotto24, Maria brings a wealth of Search Engine and App Marketing experience to the table. I was lucky enough to pick her brains on a subject that has troubled many app marketers – Google’s Universal App Campaigns.
It’s been 3 years since the official launch of Google’s Universal App Campaigns (UAC) and over a year since AdWords mobile app install campaigns’ sunsetting, or so they were formerly known. How have your feelings towards UAC evolved since the launch?
As app marketers, it took some time to adjust to Google’s use of machine learning technology, which eventually pushed us to give up (some) control of campaign management. No matter how hesitant we were in the beginning, we learned how to trust this hands-off management requirement once we started enjoying great performance, in-line with our business goals. Since then, Google announced the evolution of UAC – new aims to measure post-install actions via the launch of in-app actions and in-app action value. This was truly revolutionary as it enables each business to choose the goal that matters the most by applying the right bidding option.
With this in mind, how can we ensure that our ads are shown in as many ad placements as possible across Google’s stack? I.e. how do we catch those app users in “discovery mode” on Google Search results, the Google Play Store, YouTube, Gmail and on the millions of websites and apps in the Google Display Network?
Well, we need to make the most out of the UAC elements we have control over.
Let’s examine those elements more closely:
- Optimization Goal: specify the audience we want to target and the optimization goal we are after.
- Bidding Option & Budget Cap: choose the right bidding option depending on the optimization goal we selected and consequently set the appropriate budget cap per app campaign (if multiple).
- Creative Assets & Reporting: upload creative assets for the audience we are targeting.
- Type & Number of Events: decide the in-app actions most valuable to our business and set-up our primary (and secondary, if applicable) goals.
We have quite a lot to play with and if adding into the mix whether we are referring to an iOS or Android app, things become even more complicated.
What 3 things do you think that app marketers can do to get the most out of UAC?
Considering all the factors, it’s essential to determine the set that majorly influence app marketers’ decisions while setting up and running UAC.
Experience has shown that
- Originally setting a budget cap per campaign’s optimization goal and bidding option is vital, followed by frequency and degree of changes.
Once target CPI and/or target CPA are defined, I recommend to follow the well-known minimum budget cap best practices from the get-go:
- Budget Cap (Target CPI): 50X target CPI
- Budget Cap (Target CPA): 10X target CPA
Those budget caps are by no means set in stone, as regular reviews must be performed subject to campaigns’ performance and maturity.
This naturally leads us to think about the next point:
2. Time and degree of budget/bid changes are critical.
UAC takes time to deliver results. An average of at least 4 weeks would be required to gather learnings before starting to apply changes to budgets, bids and/or assets. The more historic data Conversion Optimizer has to play with, the more effectively the algorithm would find new users who could help us meet our app campaign goal. If it is prominent that conversion volume is or isn’t there during the launch period, changes that do not exceed 20% upwards or downwards would be safe to be made. Even so though, it should be expected that Conversion Optimizer will need time to adjust to those changes and performance fluctuations might occur for approximately 7 days (or less).
3. Full utilization of ad assets and regular performance review.
The main purpose of our ads is to be shown in as many eligible ad placements across Google’s premium inventory as possible. Therefore, I recommend to upload as many ad assets as possible. This implies that regardless of text ads being the required ad asset, image and video assets should be added as well. It isn’t a coincidence that Google launched the UAC education program earlier this year, half of which is focusing on ad assets for app campaigns. Happily, for all app marketers, we are now able to evaluate ad asset performance per campaign and test different variations. This has been a critical component to further improve our overall app performance and it is a step forward in terms of UAC reporting allowing us a certain degree of control.
Can you share some results from your current and previous experience of working with UAC?
Luckily, I have experienced working with UAC at its infancy, when I had the opportunity to encounter its great potential via app install-focused UK campaigns for Shop Direct, while heading up Somo’s Search offering. The results really do speak for themselves. We saw:
- 189% increase in installs
- 53% reduction in the cost per install compared to previous standard campaigns
- 4.5X increase in conversion rate compared to previous standard campaigns
As the UAC offering continued evolving, I came across new challenges:
- Multiple campaign set-up and management
The main challenge here was twofold:
- Determining the target CPI we were willing to pay and set a target CPA that translated into a higher target CPI compared to an app campaign focusing solely on target CPI. This was a testing exercise, which was heavily driven by business goals and actual performance by campaign.
- Defining success versus proxy actions. Recent experience has revealed that the guided minimum of 10 people in the app campaign completing a success action daily is truly the absolute minimum. Depending on the vertical each app falls under, this minimum could vary but it certainly fluctuates between 20-40. In cases this proved to be a challenge, the right choice of a proxy action leading to a success action was made. Typically, this was referring to registration (proxy) vs. sale/purchase (success), where choice of the appropriate action was based on the company’s size and app engagement.
- Android vs. iOS offering
Tested a bidding model of 1.5X higher for iOS campaigns compared to Android, which has paid off in most cases and would be recommended as a good starting.
Google promised UAC would simplify the process of ad creation and optimization. What’s your experience and that of other app marketers?
As previously mentioned, Google has recently introduced a UAC education program, which gives great insights around ad creation and effective optimization via ad reporting on Google Ads UI.
Given that text ads are the minimum requirement, I would say that what really stood out in terms of performance was appropriate creation and testing of image ads following the recommended dimensions listed on the last column:
In the process of evaluating performance per ad type (normally fully utilized text/image and video ads), it has been a game-changer – the ability to run detailed ad report, segmented by Google network. What I initially found quite confusing was the labelling of ad performance grouping (low/good/best), which required a lot of testing for understanding its value.
Sounds like you’ve driven some great results using Google’s UAC. Any final comments to app marketers who are still on the fence with this campaign type?
UAC’s machine learning deployment is a great way of effectively scaling mobile app marketing efforts. Focusing on the UAC elements we have control over and by fully utilizing what Google allows us to test, we can gather meaningful insights and steadily grow our user acquisition base. What we need to ensure is that our UAC optimization goal(s) are supported by relevant set of ads and our UAC bidding option(s) are directly linked to the in-app actions most valuable to our business.