Since the announcement of new attribution rules on iOS 14, we saw that Mobile Measurement Partners quickly divided into two camps. The first wanted to start building their technology from scratch based on the new attribution framework that Apple provided (SKAdNetwork). The second wanted to continue building on the usage of non-deterministic user-level attribution techniques.
We have spent the last two weeks debating internally, researching, and talking to our clients & partners, including Apple. As a result, we will focus on developing tools based on the new SKAdNetwork framework while bridging the gap using existing non-deterministic technologies for attribution and measurement. There will be a learning curve with SKAdNetwork and we want to make sure developers still have a reliable solution today while simultaneously investing in the future by understanding the new wants from customers under the new Apple technology paradigm. We would like to use this blog post to show advertisers what they should expect in iOS 14 coming September.
iOS 14 offers users more privacy and restarts performance measurement on mobile
The new iOS version requires a user's consent to attribute their install via IDFA. Given these new requirements, we believe methods that using IDFA for attribution on iOS will be ineffective or non-existent. We believe it’s our responsibility to prepare for this scenario even if we are wrong about the outcome.
As an alternative to using the IDFA for install attribution, Apple came out with a conversion framework: SKAdNetwork that can replace the IDFA install measurement on iOS.
Starting from September, SANs such as Google and Facebook will see 80% less attributed installs from all measurement providers
Ad networks are the foundation of the Adtech industry. To understand what those changes mean for advertisers, let’s look at the ad networks’ position when it comes to providing tools for measurement. From the attribution perspective, ad networks can be divided into two categories: Self Attributing Networks (SAN) and non-Self Attributing Networks.
Self Attributing Networks (SANs) - Facebook, Google, Snapchat
Integrations with Self Attributing Networks are 100% reliant on IDFA for measurement today. In general, we expect these SANs to be among the first adopters of SKAdNetwork given their strong ties to Apple.
SANs, regardless of the measurement provider supporting the integration, will attribute an estimated 80% fewer installs using the existing IDFA attribution. This is an industry-wide technology shift and applies to ALL measurement providers.
Once iOS 14 launches and a SAN integrates with the SKAdNetwork framework, advertisers using the SAN will play by a whole new set of industry rules. Here are a select few listed below (we will publish something more comprehensive in the coming weeks):
- Installs and conversions (events defined by the developer) are not tied to specific users and are only accessible to advertisers in the aggregate with dimensions on campaign, app, and source app.
- SAN mobile measurement partner programs on iOS will not work for any measurement provider as those integrations are solely reliant on IDFA.
- SKAdNetwork doesn’t provide the country dimension. You won’t be able to analyze such dimensions as country x ad channel or country x campaign.
- SKAdNetwork doesn’t provide a creative breakdown.
- Cohorted downstream metrics (like LTV, ROI, and ROAS) aren’t provided directly in SKAdNetwork
Non-Self Attributing Networks (non-SANs) - Ironsource, Applovin
On the other hand, not much will change in terms of workflow for Non-SAN networks in the next 6-8 months. Most ad networks in this category will automatically default to some form of non-deterministic attribution on iOS.
Non-deterministic attribution delivers similar granularities and insights as IDFA does today - you'll be able to see precise user-level attribution data, albeit with reduced accuracy. ROI, callbacks, multiple data dimensions, etc will all remain fully operational and functional.
Eventually, Apple will want all ad networks to work through SKAdNetwork and we will need to observe how the market responds to this.
A few ad networks will be SKAdNetwork certified right when iOS 14 hits, but most will not be. So we need to be prepared to serve a world where old attribution methods and SKAdNetwork coexist. This way if anything changes on either front, we are prepared.
Tenjin is building for the future with SKAdNetwork attribution
One of our priorities is to minimize the UA manager’s workflow changes due to any transition to the SKAdNetwork attribution framework. This could involve statistical, econometric, and probabilistic models used for many current non-deterministic scenarios. Since SKAdNetwork data falls into the bucket of data aggregation that will be joined with user-level metrics, Tenjin is very well positioned to tackle these problems. Tenjin is one of the first to build technology around joining aggregate datasets with user-level datasets and this type of problem is familiar. Tenjin customers will be able to collect this data for free as we don’t charge for any aggregated data.
We will continue to work with customers and Apple to support the use cases around SKAdNetwork as we believe that it’s going to be a part of the long term solution. We are actively looking at integrating SKAdNetwork conversions through various APIs with Apple and the ad networks.
As a mid-term solution, we continue to support existing non-deterministic measurement methods.
If you have questions regarding how these changes will affect you, please email us at email@example.com