No More iOS14.5 Confusion: Shedding Light on LAT, SKAdNetwork & Incrementality
IOS14.5 is expected to roll out imminently. If you haven't prepared your ad campaigns and budgets for all the changes coming out of it, the time is now.
Change in how we run advertising on iOS is inevitable. The disruptions on user privacy, attribution frameworks, and optimization methods available are changing the way mobile advertising is run for all parties involved.
Apple's upcoming update kicking off the real-world experience will give us more clarity on the path forward. It’ll most likely validate how well the industry has been preparing during the recent months by testing LAT traffic, implementing SKAdNetwork (SKAN), testing opt-in prompts, etc.
If it all sounds a bit confusing still - don't despair. This article will give you an overview of the essential concepts, cover the most urgent questions and how you can be fully prepared to still run programmatic campaigns on iOS.
What is Limit Ad Tracking (LAT)?
Limit Ad Tracking is a default privacy feature in iOS that prevents advertisers from targeting individual users. This means that no advertisers will have access to the user's device IDFA. LAT traffic is, therefore, non-personalized traffic that can use contextual targeting for brand awareness goals.
App Tracking Transparency (ATT) framework requires app publishers to get users' authorization before tracking any app-related data that would allow advertisers to track them.
ATT goes into effect with the release of iOS 14.5 announced for Spring 2021.
Understanding SKAdNetwork (SKAN)
🔹 SKAdNetwork is the framework provided by Apple to enable performance marketers to measure attribution on their mobile programmatic campaigns while protecting user privacy. Hence, it is effectively the only real last-click MMP on iOS.
🔹 SKAN submits postbacks without IDFA or any advertising ID. Its integration is essential to receive last-click attribution and performance insights for iOS campaigns.
Data granularity, postbacks and conversions events on SKAdNetwork - what you need to know
🔹 SKAN provides aggregated data. To protect customer privacy, Apple won’t be sending back device level postbacks for SKAdNetwork Data. This means SKAN lacks funnel event data granularity.
🔹 iOS sends SKAN postbacks with a delay of up to 0-24h after the event has occurred.
🔹 iOS will send only one postback per each attributed conversion event, which will contain the highest conversion value event defined by the customer. Up to 63 conversion value definitions are possible, leading to many possible outcomes in terms of measurement.
🔹 To preserve users' privacy, iOS will not send conversion-value and source-app-id parameters on a postback until after a unique combination of parameters passes an unknown privacy threshold set by Apple for the campaign. In short: if your campaign does not achieve sufficient conversion volume, you won’t see any postbacks on your campaign dashboard at all.
What data will MMPs receive from SKAN?
🔹 MMPs will receive AdNetwork ID and Campaign ID. The postback data can look like this:
"user-agent":"Mozilla/5.0 (iPod; U; CPU iPhone OS 4_3_3 like Mac OS X; ja-jp) AppleWebKit/533.17.9 (KHTML, like Gecko) Version/5.0.2 Mobile/8J2 Safari/6533.18.5"
What happens to my retargeting campaigns on iOS?
🔹 Deterministic retargeting campaigns depend on advertising identifiers like the IDFA to target users. Therefore, the changes announced for iOS 14.5 won't make it possible to run re-engagement campaigns deterministically on these users/devices.
🔹 Retargeting campaigns on LAT users will become part of a broader ad campaign approach encompassing branding and UA.
Can I use Fingerprinting on iOS?
Fingerprinting or any other deterministic identifiers are not allowed by Apple and won't substitute for the device ID. Most recently, e.g. an attribution partner had to replace their SDK as it was supporting fingerprinting, leading to their clients' apps rejection in the App Store.
Are users who opt-out of tracking on iOS apps gone forever for deterministic advertising?
🔹 No. Each app has to ask users permission individually, and some users might choose to opt-in or opt-out the first time when being asked. However, users might subsequently see incentives to opt in and change the settings accordingly. The opt-in rate by the app is, therefore, to some extent undergoing dynamic changes.
🔹 To enable deterministic advertising, both the advertiser and the inventory supply app on the same device must have the user's permission for targeting. If only one of the two apps has an opt-in, there won't be any match of demand and supply based on the device ID on any programmatic ad exchange.
What does this mean for my programmatic strategy? - The big picture
🔹 Once iOS14.5 is rolled out and inventory transits to SKAdNetwork, deterministic retargeting campaigns won't be possible to run. When it comes to re-engagement, it is very well still possible. Due to targeting inaccuracies, we'll be able to run a blended advertising campaign leading to outcomes as awareness, acquisition and re-engagement. The budget allocation becomes something more like "holistic mobile advertising" and reminds us of old-style branding campaigns. Let's bear in mind that all of these ad campaigns have the same objective: business growth.
🔹 You'll have to adapt your measurement frameworks and implement ATT. Measuring for incrementality, for example, is far harder on LAT because there is no deterministic ad performance feedback.
🔹 For device ID populations, which are mainly prominent on Android, deterministic retargeting campaigns are still viable and will run as usual based on device IDs.
How do I effectively measure my campaigns in iOS going forward?
🔹 SKAdNetwork is the only MMP for LAT campaigns on iOS, providing partially deterministic attribution for installs only. Partly because performance can be attributed to channels but won't reveal the respective user that has converted.
🔹 If you wish to measure the ad campaigns' performance on iOS in a more comprehensive fashion, you will have to implement uplift measurement based on synthetic control groups.
How do I test for incrementality in this environment?
🔹 The traditional uplift measurement relies on IDFA user-level information. Those are methodologies as placebo ads, intent-to-treat and ghost bids/ads. For LAT audiences, such uplift tests are not anymore possible as they all rely on device IDs.
🔹 Synthetic control groups can provide valuable information about the ROI of your advertising campaigns and overall mixed media model performance. They have to be implemented on the advertiser side as they reply on back-end data, give advertisers control over measurement, and give relatively fast feedback depending on the app product, campaign incentives, and usual conversion turnaround time.
What about ad fraud?
🔹 Ad fraud is expected to rise significantly with LAT traffic. The reason is simple: the device ID was used for fraud protection mechanisms and blacklisting. As traffic becomes now "invisible" from a user ID perspective, only fraud protection based on supply information is still performing.
🔹 New anti-fraud mechanisms are supposed to come into place as demand will be high to drive campaign efficiency on LAT traffic. This could, e.g. be a special site certificate authenticating the impression as genuine.
Get ready with RevX
Trust your partner to guide you seamlessly through these changes and help you implement a strategy that works for your app business.
At RevX, we are iOS14.5 ready: we are registered with Apple’s SKAdNetwork and have fully integrated with programmatic suppliers and MMPs for postback processing, mapping and forwarding. Additionally, we have been conducting extensive tests on LAT traffic to ensure that we are ready to drive the highest possible media buying efficiency once we have a real live environment post iOS14.5. We're ready to brave these winds of change with you.
Got any more questions on how to address LAT audiences? Please drop us a note - we are happy to consult, strategize and execute towards your goals.
This article was produced with the collaboration of Martje Abeldt, Chief RevX Officer.