What does Apple’s AdAttributionKit mean for Mobile Gaming?
By Mariam Ahmad 11 July 2024
“This is going to be a period of massive change and thrash within mobile.”, as put by Upptic's latest deep dive into Apple's AdAttributionKit.
As advertisers and developers navigate this new reality, AdAttributionKit brings both exciting prospects and significant challenges. While there is the fact that it enhances privacy by limiting data granularity and managing first-party data, critics worry about its impact on ad targeting precision and campaign measurement.
We take a look at the new framework, and what it might mean for the mobile gaming industry.
AAK vs. SKAdNetwork: More familiar than new
Apple's WWDC24 brought a surprise twist: AdAttributionKit takes centre stage, seemingly replacing the never-materialized SKAdNetwork 5. Launched in March for alternative app stores (thanks to the EU's Digital Markets Act), AAK is now Apple’s new foray into ad attribution tech.
AAK builds on SKAdNetwork (SKAN) 4's foundation. You'll still see three postbacks, conversion values, and privacy-preserving features like crowd anonymity. Apple wants advertisers to think of AAK as a refined SKAN with new features, including:
Universal App Store support: Initially for alternative stores, AAK now works seamlessly with the App Store. You can even use both frameworks concurrently (with Apple deciding the winner for each conversion).
Creative flexibility: AAK lets you showcase your ads in various formats - custom click ads, video view-throughs, and even in-app overlays.
Testing made easy: AAK's developer mode removes time randomization and shortens windows, allowing for faster in-app testing across all conversion windows.
The re-engagement revolution: AAK empowers re-engagement campaigns, bringing back users who already have your app installed.
Deep linking power: Paired with re-engagement, AAK utilizes universal links to direct users to specific in-app screens, maximizing campaign effectiveness.
Since AAK operates similarly to SKAN, the current workflows remain intact. It's more of an upgrade, offering enhanced functionalities without disrupting attribution and measurement.
Enhancing privacy while measuring effectiveness
AdAttributionKit enhances user privacy by minimizing the amount of user-specific data collected and shared. It employs a technique called differential privacy, which adds statistical noise to the data, making it difficult to trace back to individual users. This method enables advertisers to understand campaign performance through aggregated data without accessing personal information.
For example, advertisers can determine how many users installed an app after viewing an ad or how many in-app purchases were made, but they won't receive detailed information about individual user behavior. This approach ensures that user privacy is maintained while still providing valuable insights for campaign optimization.
Impact on mobile gaming monetization strategies
The introduction of Apple’s AdAttributionKit marks a significant shift in how mobile game developers can approach monetization. By providing a privacy-preserving method of ad attribution, this tool could have profound effects on ad revenue, forcing developers to rethink their monetization strategies.
Changing monetization strategies: AdAttributionKit changes the game by limiting the granular, user-specific data that advertisers have traditionally relied on. Without the ability to track users across apps with the same level of detail, advertisers may need to adjust their targeting methods. This shift could lead to a greater emphasis on contextual advertising, where ads are served based on the content the user is engaging with, rather than their past behavior. Furthermore, developers may need to invest more in first-party data and direct user engagement. By focusing on building a loyal user base and gathering consented data from their own apps, developers can create more personalized and effective ad experiences within the bounds of privacy regulations.
Potential effects on ad revenue: The impact on ad revenue will likely be mixed. On one hand, the reduced ability to target users precisely could lead to lower conversion rates and, consequently, lower ad revenue. The introduction of Apple’s ATT framework already led to a 40% drop in ad spend, according to some reports. On the other hand, by ensuring a more privacy-focused experience, developers might build greater trust and loyalty among users, potentially leading to increased long-term engagement and spending. Moreover, with less reliance on invasive tracking, advertisers might be more willing to pay premium prices for ad placements in trusted environments, partially offsetting the potential revenue decline.
Emerging monetization models: AdAttributionKit may encourage the emergence of new monetization models. For example, subscription-based models could become more popular as developers seek stable revenue streams less dependent on ad performance. In-app purchases might also see an uptick, as developers focus on enhancing the user experience to encourage spending. Another potential model is the increased use of rewarded ads, where users choose to watch ads in exchange for in-game rewards. This opt-in approach aligns well with privacy standards and can be highly effective in driving user engagement and revenue.
So, how do we prepare for the transition?
I don’t think mobile developers can start any serious preparation until we begin seeing the flow of AdAttributionKit data. It’s really up to the ad networks to open the floodgates. For now, just getting a general understanding should be good enough. There are plenty of explainers on the web, and I’ve also recorded a short video on this.
Meanwhile, we still have SKAdNetwork 4 (SKAN 4), which, even after two years, has not been fully utilized by most game publishers. I think we should make sure to master SKAN before we switch to AdAttributionKit.
It's rare to see advertisers working directly with SKAdNetwork and performing basic tasks such as:
Comparing SKAdNetwork data received directly from Apple with the data from ad networks. These should be the most basic metrics in any MMP dashboard.
Determining if ad networks are billing based on SKAdNetwork conversions and understanding the differences between other data points.
Utilizing SKAdNetwork's ‘source-app-id,’ which shows the App Store ID where the advertiser acquired a user.
Interestingly, app developers (non-gaming) have adopted SKAdNetwork much better than game developers.
The replacement of Apple’s SKAdNetwork with AdAttributionKit marks a shift in mobile advertising, promising enhanced user privacy and new opportunities for innovation. Will this transformation propel the industry forward? As advertisers and developers navigate this new reality, AdAttributionKit brings both exciting prospects and significant challenges. While it enhances privacy by limiting data granularity and managing first-party data, critics worry about its impact on ad targeting precision and campaign measurement. However, the new framework also introduces features like re-engagement support and interoperability with multiple app stores, potentially unlocking new strategies for effective ad campaigns.. As the debate over balancing privacy and performance intensifies, this period of change could ultimately lead to a more resilient and innovative mobile advertising ecosystem.
What about the drawbacks and criticisms?
AdAttributionKit was released as a successor of SKAdNetwork and offers the same functionality, plus a few upgrades, like support of measurement for alternative app stores, and re-engagement measurement. The limitations of AdAttributionKit are therefore all inherited from SKAN. The idea behind AdAttributionKit (and earlier SKAN) is to give advertisers insights into campaign performance without revealing data related to individual users. Apple’s crowd anonymity substantially limits advertisers’ access to install data and complicates ROAS and LTV measurement. Delays in postbacks mean that advertisers who are used to very quick, reactive campaign adjustments will have to find new ways to look at campaigns.
However, AdKit has many benefits which are often overlooked because many advertisers are finding comfort in using their old (probabilistic) methods of measurement. One of the notable benefits is more accurate reporting: where the “traditional” probabilistic attribution significantly over-attributes in some cases and largely under-attributes in others, AdAttributionKit is far less error-prone and reports significantly fewer false negatives and false positives.
We know it’s a tough choice to make when it comes to transitioning to privacy-first tracking (you have to wave goodbye to probabilistic individual user-level reporting and LTV tracking), but achieving performance and reaching quite ambitious target KPIs on AdAttributionKit campaigns is absolutely possible. We have seen many examples of highly successful SKAN-only campaigns (and that successfully translates to AdKit). The ability to achieve performance on SKAN is directly tied to the ability to overcome Apple privacy thresholds and maximize data that can be used for campaign optimization.
This can only be achieved with a custom configuration of AdKit’s or SKAN Source Identifiers catering to the advertiser’s specific needs and the specific campaign, and then using that custom AdKit scheme for getting more visibility into the most important campaign variables. We have a proprietary methodology for that, called Explore-Exploit cycle, which combines manual and AI-driven optimizations and allows mapping the most important campaign variables to source identifiers. Then we start small with testing targeting which we believe has the highest potential to perform, and after a quick testing phase we collect learnings to do both: train machine learning models that will ensure the custom bidder only bids for high-performing inventory, and scale the campaign.
The bottom line is, with AdAttributionKit Apple is giving us a way to get - although less granular - in many ways more reliable data to optimize campaigns. Embracing AdKit and learning to overcome crowd anonymity is the only way to future-proof your marketing.
While some advertisers have embraced SKAdNetwork (now known as AdAttributionKit), others have either dabbled in it or entirely disregarded it. The key unknown is Apple's enforcement strategy. Will they abruptly cut off probabilistic attribution with an OS update, creating a major disruption for the industry? Or will they phase it out gradually through a series of updates?
The adoption of SKAdNetwork has been uneven. As of early 2024, SKAdNetwork 4.0 had an adoption rate of about 33% among advertisers, reflecting a gradual but steady increase in usage as advertisers adapt to Apple's privacy-centric framework. Despite these gains, many advertisers are still reluctant to fully transition, with 87% of mobile marketers not fully prepared for a post-IDFA future.
Apple's historical preference for gradual enforcement might suggest a phased approach. For instance, the ATT rollout gave the industry time to adjust, but still resulted in significant disruption when fully implemented. However, the enforcement of Privacy Manifests in Spring 2024, aimed at stopping fingerprinting, indicates that Apple might enforce SKAdNetwork adoption more aggressively than before.
The implications of Apple's strategy are significant. A sudden enforcement could destabilize the advertising ecosystem, forcing a swift adaptation to SKAdNetwork. On the other hand, a phased approach might provide a smoother transition, allowing advertisers to incrementally refine their strategies. Currently, SKAdNetwork is responsible for about 35% of all iOS attribution, indicating its growing importance.
Ultimately, advertisers must remain vigilant and adaptable, ready to pivot their strategies in response to Apple's decisions. The industry's ability to handle these changes will be crucial in maintaining effective advertising campaigns in a privacy-first digital landscape.
The impact doesn’t seem to be that drastic, even though it’s a nice update to what SKAdNetwork can do. In this situation, it’s less about AdAttributionKit and more about the broader industry developments. Let’s take a look at the two key features of AdAttributionKit:
Multi-store: Support for alternative app stores on iOS was added thanks to the EU’s Digital Markets Act. It remains to be seen if these stores will create opportunities for game developers or just remain as a checkbox forced by legal changes. If there are no stores, there is no need to advertise or measure anything.
Re-engagement: Remarketing is far from something that everyone is doing. Mostly, this is a strategy employed by larger publishers. Additionally, there is an alternative probabilistic solution for remarketing. We will have to see how this stacks up against what AdAttributionKit can offer. Game marketers will always choose the most effective method for them.
The impact of AdAttributionKit will be insignificant unless Apple enforces a ban on probabilistic attribution (aka fingerprinting). In the scenario where Apple does nothing and allows fingerprinting, the current status quo continues. Even though AdKit supports alternative app stores, this new privacy-preserving framework makes little difference, because - as long as they’re allowed to - advertisers will turn to probabilistic attribution before AdKit.
However if Apple does implement an enforcement and ends fingerprinting, the impact could be quite significant. We believe advertisers who have invested time into SKAN-only campaigns will perform better than those who have ignored them. We believe that advertisers who target very niche, specific, and small user groups, and are unknowingly relying on behavioral targeting capabilities of ad networks will suffer the most. At the same time, larger apps with a wide target demographic of users are more likely to be able to find performance with ad networks that specialize in contextual optimization (such as Dataseat).
The change I described above will inevitably lead to some advertisers being unable to run ROAS-positive campaigns in this new privacy-compliant world. This means that demand will initially drop. CPMs will drop, but with all free markets, other advertisers (who are prepared and have invested in the new paradigm) will be able to buy cheaper impressions that perform for them. These advertisers will fill that temporary drop in demand, and a new equilibrium will be quickly found.
