July 14th, 2020
Note: As of January 28th, 2021, Apple has changed its policies regarding attribution on iOS 14. Read about that update and Branch’s latest stance here.
Apple’s latest plans for iOS 14 will effectively render the IDFA useless for mobile app advertisers, and Apple’s proposed replacement goes by the name of SKAdNetwork. For better or for worse, this new framework will result in a massive shift of the mobile attribution ecosystem.
In recent weeks, SKAdNetwork has been at the forefront of every mobile marketer’s mind now that the API might see more mainstream adoption. The idea of SKAdNetwork isn’t completely novel — Apple introduced the concept back in 2018. Its limited capabilities and lack of real incentives meant low adoption, though murmurs arose that Apple was setting the groundwork for future iterations.
That’s exactly what happened at WWDC when Apple announced that with the release of iOS 14, device-level tracking through the IDFA will be opt-in only. In their proposed solution, SKAdNetwork, Apple assumes the role of the install measurement provider for iOS, facilitating marketing attribution in a privacy-safe way.
So how will SKAdNetwork function? When a user clicks on an ad that takes them to the App Store, the ad network can pass basic attribution parameters along with the click. Should the click result in a conversion (install), the App Store will notify the ad network accordingly. Apple will report the campaign information — network, publisher, and campaign ID (limited to 100 values for details such as campaign, creative, placement, etc.) devoid of device-level details via postbacks to the ad network. They’ll also pass the conversion value — a scale of 0-63 which the advertiser can set to give some estimates to the users’ quality.
Sounds reasonable enough, right? Now let’s take a look at how SKAdNetwork stacks up against traditional MMPs for Mobile Tracking in this new world of iOS 14.
Privacy, privacy, privacy. Apple vows to support features that are beneficial to both users and the app ecosystem. As privacy concerns grow, it’s not surprising that Apple has put the decision “to track you across apps and websites owned by other companies” in the hands of the consumer. With the move towards SKAdNetwork, Apple aims to help measure conversion rates of app install campaigns without compromising users’ identities.
Ad Fraud (for the time being). In the SKAdNetwork framework, Apple vouches for all app installs, which will dramatically reduce the likelihood of ad fraud. Driven by high incentives, intelligent fraudsters have continually innovated their malevolent processes with the market, so we’ll need to wait to see how this will play out in the medium/long term.
Install Accuracy. SKAdNetwork attribution and aggregation occurs directly from the App Store, Apple’s infamous black box, which effectively means aggregate attributed install counts will be fairly accurately depicted from this single source.
Aggregate data. Advertisers will only receive campaign aggregates with abstracted ROI. There will be no device ID or user-level visibility that was often used for things like retargeting or multi-touch attribution. Advertisers will need to find alternative solutions or shift their mindset to trust networks to report on aggregate performance.
Post-back timing. Postbacks will be generated and sent at least 24 hours after install. This will prevent the installed app from attempting to tie back in-app actions to postback activity and from identifying users at the device level. Real-time optimization will be a thing of the past, and networks will be severely hampered in their ability to optimize campaign performance.
Impression tracking no more. View-through attribution will not be supported; an ad unit must be clicked for Apple to report on the resulting actions. Gone are the days of impression tracking and measuring the impact of supporting touches.
Post-install activity. SKAdNetwork only allows you to send a single post-install event back to the ad networks (and requires delaying the initial attribution postback to do so), which means tying together in-app events to these install actions is not possible via this point solution. For marketers, measuring campaign performance has historically gone far beyond simply knowing if your app was installed (by someone out there).
Deferred deep-linking. SKAdNetwork provides no framework to defer deep link from ads. This is a huge sacrifice to user experience and conversion rates might take a hit.
Implementation woes. There are also challenges in the actual implementation of SKAdNetwork, requiring publisher apps to actively register the approved ad networks in a newly-released version of the app. This means dev work any time you need to want to work with a new ad network.
Workflow disruption. Any workflow that requires taking the device-level data behind an aggregate report (such as audience building, where that device-level data is sent to an ad network) becomes impossible without access to the underlying data.
Referee of the ads arena. While some of the mechanisms would change, the role of the MMP stays exactly the same: provide unbiased measurement that allows you to make actionable decisions (see more in the next point!).
Attribution management. With SKAdNetwork, advertisers will need to collect, store, and validate all install postbacks from each ad network, while attempting to analyze and connect them to campaign spend. MMPs plug into all of the networks to pull in cost and performance, allowing easy manipulation and visualization of all campaigns in a single dashboard.
Device-level attribution. Speaking specifically for Branch, we’ve anticipated this change for quite some time, and our link graph has always been built to function with other identifiers in cases where IDFA is not available. Our predictive matching engine already covers upwards of 90% of your mobile user base, and this attribution method will be available in iOS 14.5 and beyond when users opt into ad tracking.
Links, a likely future. For advertisers and networks that are unwilling to accept the limitations of SKAdNetwork, MMPs continue to provide the infrastructure for (a) link building to enable tracking without SKAdNetwork, and (b) user routing to support direct deep linking for re-engagement, and deferred deep linking for higher initial install and down-funnel purchase/ engagement rates.
Adjusting our integrations so you don’t have to. All current MMP integrations (both from Branch and from other attribution providers) use the IDFA to confirm device-level attribution. When the IDFA goes away, all of these integrations change based on each new specification. Our goal will be to handle these technical changes on your behalf, minimizing any updates you would need to make.
No deterministic installs. In situations where the app is already installed, there will be no change to matching accuracy; Branch will still be able to offer deep links with deterministic (guaranteed) matches. However, like all MMPs and attribution tools, we won’t be able to support deterministic matching on iOS (unless a user is logged in across your properties) for the first install, as this capability relies on IDFAs.
Shift from the Ads-focused mindset. As the ads space loses visibility, targeting capabilities, and potential value, having a point-focused ‘mobile measurement partner’ for your ads may in fact become less important.
The last point above is a great one to end on. Regardless of how our customers and partners choose to approach mobile tracking, Branch is committed to supporting all privacy-centric efforts towards providing unbiased measurement to dictate actionable decisions. We believe there is a much larger opportunity outside of app-install tracking that begs to be not only supported but promoted. In a world where install ads become relatively less cost-effective, it will become even more important to focus on other areas of your mobile tech stack, for example, web-to-app user journeys, or deep link-powered referral programs, so having a ‘mobile growth platform’ has never been more critical.