A guide to incorporating SKAdNetwork into your current user acquisition reporting

A guide to incorporating SKAdNetwork into your current user acquisition reporting
January 20, 2021 Simon Whittick

Apple’s App Tracking Transparency (ATT) framework is expected to roll out early this year and we’ve been working closely with our customers and product team to help prepare for the SKAdNetwork era. In this post we share our thoughts on how reporting and BI systems should adjust to support the changes, as well as share upcoming product changes that make reporting stress free so you can focus on optimisation.

Recap of SKAdNetwork

To start, let’s quickly recap the key changes that will have an impact on reporting:

  • SKAdnetwork will become the standard attribution solution for iOS users. Limitations of SKAdnetwork include:
    • Not user- level, aggregated data only
    • Limited post install visibility, primarily just 24hrs of post-install tracking
    • Limited granularity, at a campaign, keyword or sub-publisher level
  • ATT opt-ins are expected to be low (~20% best case) to the point where they won’t deliver the full picture alone.
  • Fingerprinting is off the table as Apple has explicitly warned against it.

When it comes to reporting there are always two important elements: data sources & visualisation. Below we’ve covered the approach for both elements.

1. Data Sources

The data sources you traditionally ingest for reporting will remain valid. We typically see these consist of: 

  • Cost sources
  • Mobile Measurement Partner (MMP) attribution e.g. Adjust, Appsflyer
  • Internal revenue data
  • Predictive revenue data

However, in addition you now need to also include SKAdnetwork data. One of the most important parts of SKAdnetwork data will be your conversion value logic. We’ve interviewed dozens of customers and 90% will be using their MMP to handle the conversion value logic. It’s important to be mindful of conversion value setup and try to replicate events as closely as possible to your traditional user acquisition setup. Be aware that logic can vary by MMP so investigate this closely before setting the defaults.

From an Appsumer perspective we will either be able to ingest your SKAdnetwork data from your MMP with the correct events assigned or ingest directly from the ad networks and apply your own conversion value logic.

It’s also worth considering enhancing SKAdNetwork data to overcome some of its limitations with forms of media-mix modelling such as linear redistribution, probabilistic distribution or top-down incrementality. These solutions all help fix the disconnect between SKAdnetwork data and actual revenue, and in this post we’ve outlined these solutions and potential providers whom Appsumer can ingest data from to enrich reporting.

2. Visualisation

There are two types of data views we think are critical in the SKAdnetwork era. One to compare metrics by different sources (e.g. SKAdnetwork vs MMP) to examine deltas and one that combines metrics for top-level visibility you can optimise from. Let’s jump into those in a bit more detail.


With more sources for the same metrics than ever before it’s important to have a clear way to understand the deltas between different data sources. This will help you understand why different sources are presenting different numbers and which ones you want to trust at different levels of granularity:


Once you have compared metrics and decided which source you want to trust at different levels of granularity you can combine the metrics you select to get a top-level view. This gives you a clean report you can use to optimise against with trust and present to wider stakeholders. This is a feature Appsumer is launching to provide the best KPIs depending on granularity, for effortless decision making:

Concluding thoughts

There’s no doubt that data complexity is increasing and the data you use to optimise from will not be as accurate as the deterministic data you are used to. However, this is also an opportunity to take a completely new look at how you measure success moving beyond last click attribution to a more incremental outlook using probabilistic media-mix modelling.

Whilst we’re sharing this as advice for your own reporting this is the solution we’re building at Appsumer to support customers with SKAdnetwork reporting after much industry consultation. You can read more about that solution here and if you want to discuss how this might work for you, get in touch here