78.9% of ad inventory is SKAdNetwork-compatible. Subscribe for weekly updates on SKAN inventory & non-IDFA traffic.

To Allow or Not Allow Tracking. That is the Question.

March 9, 2021
INCRMNTAL
To Allow or Not Allow Tracking. That is the Question.

The clearest guide to a world of confusion.

The coming privacy update announced by Apple during WWDC, June 2020, included a major update requiring users to provide their consent for apps to access and be permitted to share the device IDFA.

Apps are only allowed to ask users once, they cannot limit app functionality if the user does not allow tracking, and are not allowed to incentivize users to opt in.

Apple released an updated attribution framework (SKAdnetwork 2.2) which provides attribution data on campaign+publisher level, but only in aggregate form, limiting advertisers ability to match a single user to a click or impression.

Opt-in rates are estimated to be around 50% – but the implications of “50%” will have more of an impact than what you may believe. Let me explain.

The 1st Factoring Problem

In this example, we have 10 users, 3 publishers and 1 advertiser. From first look, it seems that since all 3 publishers and the advertiser have an opt-in rate of 60% – those 60% will share IDFAs amongst one another, allowing advertisers to track and attribute conversions using IDFA matching.

But this isn’t how the opt-in mechanism works. 

Both publisher and advertiser must request permission for tracking, making the factored opt-in rate, lower:

The Matching Problem

For attribution to happen using identifiers, users need to engage with ads on the apps they allowed tracking. Making the factored opt-in get another factoring based on the app where the users saw and clicked an ad for the advertiser…

In the example below, the publishers where the users clicked an ad (and installed the app) is marked in pink.

To provide you with some clarity:

  • Your attribution provider (MMP) will have tracked 3 installs out of the 10 actual installs that happened. 
  • SKAdnetwork will report all 10 installs in aggregate form to the networks.
  • AdNetworks will receive a total of 13 postacks – the 10 from SKAdnetwork and another 3 from your MMP.
  • None of the parties involved have a way to deduplicate the postbacks.

To add another caveat…Apple Search Ads gets attributed by a different API and not through SKAdnetwork – i.e. another set of attribution with no ability to deduplicate.

Confused yet?

The Allow Tracking vs. Do Not Track Problem

As consumers – we love “2 for 1”, but as marketers trying to get accurate reporting for optimization, we rely on the data to be accurate. The matching between opt-in vs. not opt-in users creates a big problem for conversion tracking. 

Check out the diagram below:

In this scenario:

  • Advertiser received 1 install.
  • SKAdnetwork will track 1 install, and attribute this to publisher B.
  • MMP will track 1 install for publisher A, and if the MMP is also handling SKAdnetwork postbacks – the MMP will show both the install they tracked, as well as the install tracked by SKAdnetwork.
  • The MMP has no way to deduplicate the 2nd postback or know which one was the last touch point before the user converted.

And remember, SKAdnetwork won’t send a postback for a single install, as posting a single install defeats the purpose of a privacy first attribution solution.

Attribution and tracking in a post-IDFA environment will fundamentally change marketing on iOS, requiring us to adjust how tracking, attribution and measurement are done. With iOS 14.5 just around the corner and a steep learning curve ahead, now is the time to get started.