Beating SKAdNetwork With Predictive Analytics
If you’ve heard of SKAdNetwork, you probably wish you never had.
SKAdNetwork is the framework Apple created to restrict detailed information on app installation and usage that developers used to receive. With iOS 14.5’s release in 2018, SKAdNetwork became the “privacy-friendly” method Apple uses for the attribution of ad campaigns. Apple now requires that ad networks register with them. Furthermore, developers have to maintain compatibility with those registered networks and SKAdNetwork.
To be sure, major platforms and device makers are constantly modifying their technologies. Adapting to those changes is part of the deal.
But SKAdNetwork brought about a significant change that affected app developers’ and marketers’ processes. SKAdNetwork shares mobile install attribution data with advertisers, but not on the individual user or device level. Instead, only aggregated data is shared. In these circumstances, it’s possible to assess how app install campaigns and various channels perform. However, details on individual users’ identities are missing.
In the meantime, users can consent to share the Identifier for Advertisers (IDFA) information. Recent data reveals that 3 in 10 users of gaming apps have agreed to use IDFA. That’s the highest opt-in rate of any app vertical. Consent is even less common among users of health and fitness apps (11% of users), fintech apps (11%), and entertainment apps (14%).
The loss of nearly all granular tracking information dealt a severe blow to mobile developers and advertisers for whom that detail was instrumental in targeting efficient campaigns. Additionally, to further complicate things, SKAdNetwork provides only 24 hours of data from app usage and introduced randomized delays into data collection.
Dealing With SKAdNetwork’s Limitations Through Predictive Analytics
One of the ways marketers are dealing with SKAdNetwork constraints is by optimizing their use of the conversion value. The conversion value is a 6-bit value sent to ad networks with every attribution of an app installation. Within those 6 bits, there must be “encoded” as much information as possible to provide a strong signal about whether and how the user’s journey will proceed.
There are many possible ways to optimize that conversion value. However, one of the best is sending the available user and app usage information into a predictive model. For example, data might include source and attribution details and data on the completion of key in-app events, such as the level achieved.
Then, the predictive model rapidly and accurately calculates each user’s potential to become a payer, to be a high-value user, to churn, or to complete other actions of interest to you. Finally, predictions for these different outcomes can be passed to your MMP or ad networks through the encoded conversion value for use in quickly optimizing campaigns. Overall, introducing predictive values to the conversion schema helps resolve many of the biggest challenges presented by SKAdNetwork.
An Example of Using Predictive Analytics Within SKAdNetwork Constraints
Let’s say you want to use the conversion value to predict whether a user will become a payer or a high-value customer. A predictive analytics approach — as is available with Pecan’s accessible, automated platform — builds models and provides app developers with predictive outputs. Then, developers of an advertised app can pass these predictive values as one or more of the 6 bits in their conversion schema.
Given that predictive outputs can take into account a wealth of historical in-app data, these values are far more actionable and provide greater insight into the future value of a user-acquisition campaign. Essentially, you’re getting information about what will happen on day 30, but on day 2. Imagine having the ability to not just describe on day 2 what happened on day 1 — but also to predict on day 2 what the 30-day value of an ad campaign will be. In this situation, you can proactively craft your campaign messaging, channels, and more to maximize its benefits. Results can include dramatic improvements in user LTV, churn reduction, lower CPI, and other benefits to critical KPIs.