How key mobile media channels are responding to iOS14 and SKAdnetwork: Everything you need to know in one place
Last updated: 29/01/21
I was recently reading through Eric Seufert’s well put together analysis of the winners and losers in mobile advertising from iOS14. Within it, he highlighted top mobile traffic sources and how he broadly expects Self Attributing Networks (SANs – e.g. Facebook, Google, Apple Search Ads), Ad Networks (e.g. Unity Ads, Vungle, Applovin) and Demand Side Platforms (DSPs – e.g. Liftoff, Aarki, Crossinstall) to be impacted by the need for users to opt-in for IDFA access when launched in iOS14.
At the same time we were releasing our quarterly benchmark report on spend, and analysing share of wallet by Operating System (OS) to understand which channels are relying heavily on iOS revenue. We found that Facebook had the most significant iOS share of wallet with 51%, followed by Apple Search Ads (ASA) with 17%, Google UAC with 10% and Snapchat 6%.
With this in mind, it got me thinking about how key mobile ad channels are reacting to iOS14 changes. More importantly, what does that mean for your campaigns as a mobile user acquisition expert?
So, we’ve pulled together an overview of the top five SANs in terms of iOS share of wallet, summarising what they’re doing about iOS14 changes and what that means for you. We’ll continue to update it as different media channels release more information, including DSPs, Ad Networks and longer-tail SANs. You can get detailed information we’ve pulled together for each channel on the pages linked below:
Alternatively, if you’re short on time to read about individual channels here’s the TL:DR of what it broadly means for your campaigns:
- Ad Account Restructures: You need to restructure iOS14 ad accounts due to SKAdnetwork (SKAN) limitations. For Facebook, you’ll have 9 campaigns all having 5 ad sets of the same optimisation type underneath each. On Twitter you will be limited to 70 ad groups and if you have more the campaign will be paused. For Google UAC you will have 8 campaigns per iOS app. Whilst Snapchat Ad have a SKAN beta they haven’t publicly shared what the impact will be on campaign structure, but assume it will have similar constraints to Facebook, Twitter and Google due to the limit of 100 SK-campaign IDs.
- View-Through Attribution: You will likely lose view-through attribution when targeting iOS14 devices. This is likely to mean that you’re reporting reduced returns on channels like Facebook, Google UAC, Snapchat and Twitter, which typically perform best further up the funnel. Predictive modelling that we cover in more detail here is likely to become the de facto solution to investigate. The one upside here is that all channels should now be measured using the same methodology, so it is at least an apples-to-apples comparison. However, probably less effective measurement for driving full-funnel performance.
- Delayed Data: For channels Facebook, Twitter, Google and Snapchat who will rely on SKAN iOS14 data will no longer be real-time and delayed by days. Avoid knee jerk reactions based on the previous day’s performance after Apple implements these changes as you may find performance levels out a day or so later.
- Performance: Expect channels like Facebook, Google UAC, Snapchat and Twitter to go through a learning period with their algorithms for iOS14 campaigns once these changes are implemented. This could result in performance fluctuations. It’s easier said than done, but it might be worth looking to move some budget to Android campaigns during this period, if you haven’t maxed out performance metrics. Otherwise, you might be best just jumping on the rollercoaster ride for a time.
- Mobile Measurement Partners (MMPs): If you use an MMP SDK (e.g. Adjust, Appsflyer, Branch, Kochava) on iOS14 you will need to get user consent to do so using the App Tracking Transparency (ATT) framework. Also, currently Facebook will share SKAN data with MMP’s if they have Business Manager permissions. Twitter, Google and Snapchat haven’t yet confirmed publicly if they will pass SKAN data back to the MMPs but we suspect they will. The only significant question mark is for Google UAC who support via Firebase, and could use this as an opportunity to push Firebase adoption.
- Lookalike Audiences: The reach of this targeting type will be limited. Your iOS14 Lookalike Audiences may be better built using in-app events. You may have already analysed what these high-value events are in partnership with your product team or you may need to execute this analysis. Either way, you’ll need to invest time in iOS14 Lookalike Audiences from users who execute those actions.
- Post-Install Behaviours: To avoid delaying reports for channels using SKAN, you’ll want to limit the number of post-install events you track. Similar to the events you’re tracking for Lookalike Audience creation you’ll want to identify what the most important events are that are early leading indicators of monetisation and retention and focus on tracking these events.
- Facebook Audience Network (FAN): Facebook themselves have admitted without IDFAs on iOS14, FAN performance will drop. Whilst the switch from a waterfall to a header bidding model might keep CPMs afloat for publishers and developers, you will likely want to opt-out of it for iOS14 campaigns once Apple’s iOS14 changes are implemented. It may be that FAN for iOS14 becomes completely redundant.
- Apple Search Ads (ASA): There’s not a lot to see here. Campaigns are largely unaffected currently. However, if you haven’t given campaigns much love in recent times it might be time to dust them off. It’s important to now position yourself on top keywords, even the competitive ones. They may be expensive now, but they might get more competitive as advertisers try to get more out of ASA. The more history you have will serve you well and experimenting now will likely pay back when ATT and SKAN become normal. However, as an intent-based search channel if campaigns are maxed out there’s not much you can do without an increase in search volumes for your brand or category.
- Contextual Targeting: In general, what we’ve seen on the web as user-level data is limited is that advertisers moved towards more contextual targeting. It’s a good time to evaluate where your brand has opportunities to explore more contextual targeting options vs reliance on user-level targeting.
- Probabilistic LTV distribution: A big challenge with SKAdNetwork is losing the connection between media spend and actual LTV. We’ve outlined some solutions here to help fill the gaps of SKAdNetwork with probabilistic LTV distribution. Worth investigating to continue optimising to to LTV on iOS devices.
It’s easy to read this and view it as an apocalyptic moment for mobile advertising. The overriding message is “don’t panic”. This is an adjustment for mobile advertising, but not the end of it. Advertisers on the web have been through similar adjustments recently and had to change the way they think about things, but they survived 🙂
The meta thinking for the web advertisers that thrived was not to try and find loopholes. Apple will quickly shut them down. Accept the spirit of their changes and find new ways to work within them.
Here at Appsumer this kind of data complexity is what we thrive on. We’re building the most complete SKAdNetwork Reporting & Modelling solution to deliver you transparency on different metrics and a full performance view despite SKAdNetwork’s limitations. You can read more on that here or get in touch to discuss specific solutions for your challenges.
For now, give us a shout if you have any questions and check back here as we update this resource when more announcements are made in the coming months ?
P.S. Don’t forget you can get more detail that we’ve put together on specific media channels using these links: