Why is iOS 14+ such a shift?

iOS 14 takes away privacy control from the app developers. A new framework, the AppTrackingTransparency (ATT) requires now users to give – or deny – access to their unique device identifier (IDFA) to be collected by advertisers.

What does it mean for apps?

App owners no longer have access to the user's IDFA by default, forcing mobile app marketers to rethink how to measure, attribute, and optimize their advertising campaigns.

Case A: User allows the app to track their activity - marketers have access to their IDFA.

Case B: User refuses the app to track their activity - marketers don't access their IDFA and thus cannot track this user over time or retarget him/her.

According to AppsFlyer, opt-in rates are between 30% and 40%. With most iOS users denying access to their user-level data in versions 14.5 and above, IDFA access is becoming more and more limited to marketers. Want to learn more about the different changes for mobile marketers between IOS 14 and IOS 15? Check it article out.

‍

Now, to measure the success of their campaigns, marketers have to rely on SKAN, the solution designed by Apple to measure attribution post iOS 14. However, it is quite limited compared to previous solutions.

Limitations of SKAdNetwork

The essence of SKAN is the anonymization of users. So, to prevent reverse-engineering for user-level matching, Apple introduced features such as thresholds, timers, and delays in postback to ensure no one can attain user-level granularity. This consequently restricts the measurement of campaign success and post-install user activity, both in the granularity and window time of events. That's why SKAN brings limitations compared to previous methods of attribution.

What kind of limitations are we talking about?

• Granularity
  SKAN marks the end of user-level data and user identifiers. The data is presented at the campaign level only and is limited to 100 campaigns per network, per app.
  ‍
• No web attribution
  The web is becoming an important touchpoint for app marketers, it's an attractive source for both acquisition and engagement with existing customers. Many journeys begin on Google Search, and, according to Vungle, about 10% of installs have a web touchpoint. However, SKAdNetwork doesn't support web attribution or any web-to-app solution that could attribute users on iOS devices who asked not to be tracked via ATT.
  ‍
• No re-engagement attribution support
  Deep linking or view-through attribution is not supported by SKAN, which, as well, does not consider anything but the act of downloading as attributable.
  ‍
• Conversion Value and postback delay
  Apple is now sending only one single postback to the app - between 24 hours and up to 30 days post-install, depending on the user's activity. This makes it almost impossible for advertisers to conduct real-time optimizations. In the case the user is not active within 24 hours after the install, the Conversion Value, 0, is sent. In the case the user is active during the 24 hours post-install, the Conversion Value will be calculated depending on how many events the user triggered (up to 6, during 30 days max). Notifications are sent 24 to 48 hours after the app was last opened or after the last event was triggered.

What's a Conversion Value anyway?

A Conversion Value is the sum of 6 values given to 6 defined actions a user might do in the app. It's the only piece of data that is available on post-install activity from SKAdNetwork - limited to 30 days post-install only.

Every action can be imagined like a switch that goes ON & OFF. The switch turns ON (binary digit 1) when the user completes the action - the switch stays OFF (binary digit 0) when the user has not completed the action. The value 1 or 0 of the in-app event is attributed to the source of the install.

By turning ON or OFF the 6 bits, there are 64 different combinations available - the lowest being 0 (all bits OFF) and the highest 63 (all bits ON). This single field of 64 values must contain every detail about the user’s journey as it's the only way for iOS advertisers to measure user LTV in SKAN campaigns. Now, by properly defining their bit assignment strategy and mapping out those 64 possible values, advertisers are still able to measure post-install revenue, activity, and retention.

Why predictive models are the future

Conversion Values are the only postback available, they don't have a timestamp and are mostly based on early signals in the funnel of the user journey. The ability to cohort users and predict their value is considerably decreasing - with these limitations predictive modeling is becoming a must-have for marketers.

Effective predictive models will unlock the power of those early signals, and predictive analytics will enable marketers to take user activity during their initial couple of days with the app, and correlate it with their long-term LTV.

For example, Tempr., which is a predictive and automation technology, translates the Conversion Value sent by SKAN into scenarios and predicts user activity based on data from ATT-consented users, organic traffic, Android users, and post iOS 14 data. Depending on its predictions, Tempr. suggests campaign optimizations to increase either ROAS or the number of events.

Are you navigating SKAN alone?

Are you facing challenges with SKAN? Don't worry, you're not alone! Join the first and only LinkedIn SKAdNetwork Community that gathers industry leaders, mobile app professionals, and marketing experts sharing tips, best practices, and useful articles around the topic. You'll get useful insights and can ask for help and advice at any time. See you there!

‍

Want to read more about SKAN and receive cool news? Sign up for our newsletter here.

‍