January 12, 2018
Calling it now: automation will be the top buzzword of 2018.
In all fairness, automation has been and will continue to be a major conversation in nearly every industry. Digital marketing seems to be at the very forefront of this trend. Account managers are increasingly adopting machine learning based strategies for bidding, targeting, and creative delivery. While adoption rate increases, some platforms are starting to remove the manual option altogether. A recent example of robots taking over for their fleshy friends came when AdWords retired App Search and Display App Campaigns in favor of Universal App Campaigns (UAC).
Now that UAC adoption is mandatory for promoting an app on AdWords, it’s important to compare Universal App Campaigns vs Mobile App Install Campaigns. In order to achieve this, we’ll discuss the fundamental differences, walk through a brief launch guide, and conclude with a performance review.
What is a Universal App Campaign?
A Universal App Campaign, or UAC, is a campaign type that integrates machine learning into every phase of app promotion. Ads are eligible to run across the Search Network, the Display Network, and Youtube. In the past, Search App Install Campaigns relied on keyword targeting and offered an array of bidding options. Likewise, Display App Campaigns provided layered targeting techniques. UAC, on the other hand, execute nearly all of the heavy lifting with the advertiser responsible solely for setup, budget maintenance, and target CPA changes.
Setting Up a Universal Install Campaign
Setup for UAC is impressively simple. To begin, navigate to the add campaign button within the interface and click “Universal app campaign”
Once in the guided setup, you’ll need to name your campaign and select the proper mobile app. After this, move forward with ad creation. Here you can provide up to four text ideas, an ad video (for Youtube), and up to 20 images. Ultimately machine learning will dictate what combination of these pieces are shown in response to your audience’s behavior. During this process, make sure to provide as many image sizes as possible to maximize your number of eligible placements. If you opt not to include a Youtube asset you will still remain eligible for some auctions. However, we’ve seen negligible Youtube volume in accounts that made this strategic decision.
Once images and text ideas are added, all that remains are high-level decisions. Pick the proper location and language for your ad. Ensure you have the right optimization objective from the two options:
- Get new users to your app
- Get new users who are likely to complete in-app actions that you select
Currently, the only bid strategy offered is target CPA. Once you set your target goal and a daily budget you’ve officially navigated through the entire launch process.
Best Practices For Universal Install Campaigns
Best practices for Universal App Campaigns are fairly limited. To be honest, there just aren’t that many levers at an advertiser’s disposal here. Such a lack of targeting options can create issues regarding brand safety considerations. For this, make sure to throw obvious “no-go” sites in your “Account Level Placement Exclusions List” found under the shared library. Beyond that, topic and category exclusions are available on the backend — but you must have the help of an AdWords account representative. Same goes for placement reports which are inaccessible to the individual advertiser without Google support.
Beyond brand safety, success with UAC requires a combination of diligent attention and patience in the machine learning process. Like most automation, performance will likely improve over time with increasing data history. That said, we prefer to launch new efforts with both a conservative budget (roughly 65-75% of actual target) and a conservative target CPA (roughly 20% below goal) in order to avoid inherent inefficiencies in the beginning. When it comes to ongoing optimization, keep in mind that major changes to the budget or target CPA will throw off the algorithm. If possible, avoid making changes to either of these metrics over 20%. Lastly, continually monitor performance. Chances are you’ll eventually see spend limited by budget. When this is the case, lower your target CPA by 10%. Despite AdWords doing the heavy lifting, routine optimizations are still required to find the perfect medium between spend and bid.
Case Study: Analyzing Performance
For one client I work with, the retirement of Search App Campaigns was a rude awakening. Our strategic approach required a large degree of manual processes which can only be replicated in Search App Campaigns. In particular, their extensive brand safety approach made UAC a major risk. Furthermore, Search App Campaigns had outperformed UAC during a previous test. Despite these reservations, a mandatory launch proceeded which was closely followed by impressive results. Over the last two months, UAC outperformed historical Search App Campaigns by 22.60%.
Furthermore, CPI started at an aggregate of $ 0.60 followed by a drop to $ 0.40 in December, and $ 0.30 in January. As the campaign builds history, delivery has become increasingly effective. We see this trend across every single geo we are currently running in.
Beyond CPI, the conversion rate is an extremely important metric to keep an eye on for App Install advertisers. While the first month showed a major drop-off in this metric compared to our Search App campaigns, each month since has provided incredible growth to the point that UAC is now performing more efficient.
One trend to keep an eye on is the % of delivery by network. As previously mentioned, ads can show across search, display, and Youtube. My hunch is that during our previous UAC vs Search App Test, the Search App campaigns were monopolizing most of the search inventory and thus leading to artificially poor performance for UAC. As the graph below shows, an increase in % of delivery within the Search Network directly correlates with our improving CPI and CVR.
Fortunately for our team, UAC has been an incredible success. Diligent budget and bid management have maximized performance, while machine learning has consistently improved
The switch to UAC is not optional. Fortunately for advertisers, Google has done a quality job automating the majority of the process. Once launched, an account manager should be confident that performance will slowly improve in both volume and efficiency. Despite much of the control being removed from the individual, persistent budget and bid management are required to achieve optimal performance. It remains unclear what the future of UAC holds. Will Google open up more control to the advertiser in areas such as brand safety? What steps (if any) can be taken once the algorithm establishes a performance ceiling? I think it’s safe to say this a polished product, yet one that is likely to see updates down the road. If you have any thoughts or questions feel free to connect with me on Twitter @Will_Larcom.
Cover photo courtesy of Jeremy Thompson