On April 26th this year Apple released their iOS14.5 update, which added to the list of already existing challenges - like GDPR or Cookie-Tracking-Restrictions – for the marketing world.
To be fair, the update and its far-reaching consequences for mobile user analytics was already announced by Apple back in 2020 to give app developers enough time to make adjustments.
Nevertheless the update left a huge impact on the industry especially for marketers operating on advertising platforms like facebook and relying on their revenue coming from performance channels. We have already taken a closer look on the update when it was published and now, eight months later will review the impact it had, what we've learned and suggest how to deal with marketing optimization in the upcoming year.
What exactly do IDFA and ATT mean?
Let’s rewind: The update of iOS14.5 included the so called App Transparency Framework (ATT). Within this framework, all publishers of apps running on devices using iOS14.5 must ask their users for permission, if they want to track, share or collect their data via the Identifier for advertisers (IDFA). The data usage has to also be described on the product page in the app store.
Before iOS14.5 the IDFA came by default, enabling advertisers to anonymously identify a user’s device and collect information about their behavior while using apps and mobile websites. Marketers used this information for performance and user experience optimization.
As of today, 90% of all iPhones and 85 % of all devices released in the past 4 years are running on iOS 14. The opt-in rates for tracking are observed around 20%, varying from country to country. Other sources such as Appsflyer report higher opt-in rates around 40%, taking into account not only denied but also not determined ATT status on devices.
Whether it is 20% or 50% - at least half of the users don’t consent to be tracked across devices. This decreases the available data points dramatically for advertising platforms making optimization based on cross-device conversion data almost impossible.
The impact on social ad campaigns and reporting
The data collected through the IDFA was largely used for running precise retargeting campaigns on social platforms like facebook.
Through missing opt-ins, facebook was unable to properly track user behavior off-platform and web conversion events. For performance marketers the ability to run strategic retargeting and personalizing their ad message was made impossible. Retargeting Pools not only on facebook dropped in size, resulting in losing potential customers especially in the bottom of the funnel.
The granularity of separated ad-sets retargeting different audiences - for example abandoned cart users and product view users - addressing them strategically with different messaging is now gone. The tracking issues affected not only facebook’s former algorithmic optimization capabilities but as well the overall reporting granularity and reliability.
First of all the attribution window changed to choose between 1-day click/1-day view and 7-day click/1-day view, collecting data from iOS devices only for 24h. With this most reported results on campaign level became estimated based on modeling at the same time lacking transparency for marketers to understand how the results came about.
Further as preparation for the ATT update, Marketers were asked to set up and prioritize conversion events such as Product View, Add to Carts and Purchase Events and set up Server-Side-Tracking via the Conversions API. These actions though didn’t seem to bring much improvement in the tracking once the iOS14.5 update was in place.
Reported conversions and ROAS dropped in half on average although for most companies the overall revenue reported from their shop platforms remained flat or even further increased in proportion with marketing spend. This led to the conclusion that a “blended” marketing ROI calculation simply taking into account overall spend and overall revenue helped validate that advertising effectiveness at least with a ballpark number.
However, it not only got hard to advertise and target properly, but also to do strategic planning for budget allocation on marketing channels, because the real impact between channels, activities and revenue remained unclear.
The impact on costs and budgeting
Planning budgets got more than complicated and of course - expensive. The pandemic with greatly increasing online shopping numbers caused a huge competition and thus led to online ad costs jumping up. Budgets were shifted from OOH to online and mobile and it’s no surprise the share for android compared to iOS was much greater than the years before.
This had a huge impact on for e.g. Black Week which used to be a retargeting super booster season for companies - now with much smaller retargeting pools and higher advertising spend hurting consumer businesses.
Reactions and solutions
Apple themselves provides a SKAdNetwork API for a privacy-preserving, accurate attribution for iOS campaigns. So it is somewhat possible to link installs and conversions to campaigns and marketing channels, but not on the user level.
Facebook provides an aggregated event measurement for iOS14.5 (or later) web and app events. The platform recommends to combine ad sets to receive larger audience sizes for targeting as well as prioritizing the most relevant events, as iOS devices will only report the one event highest of your hierarchy. All functions, effects and restrictions are largely explained here.
However, there is a potential solution to the tracking incapabilities - at least for companies who have a historical database: Marketing Mix Modeling (MMM) applies statistical analysis of time-series data from marketing activities on a macro level aggregated by day - e.g. daily spend on a specific platform and daily revenue in the stores - to help show the impact of marketing activities on conversions.
Another advantage to using MMM is that other non-marketing related effects can be taken into account, such as seasonality or product changes. The advantage of MMM operating on aggregated data and by this being independent from cross-device and user-level tracking, not utilizing any personal identifiers and with this applicable regardless of IDFA restrictions or GDPR regulations. Our product Sphere applies MMM and helps you to analyze the impact from different marketing channels. It includes a budget forecasting tool to help you plan holistically and set up your marketing strategy proven by statistical analysis. We have written extensively about how MMM can help circumvent the iOS14.5 challenges in this blog article, including a case with our client FreeNow.
Given that this method is data-driven, it needs historical data to deliver a big enough volume to be significantly analyzed by the algorithm. Subsequently smaller budgets shifts or new platform experiments with smaller spends such as trying out Pinterest or Tik-Tok with a small fraction of the marketing budgets will probably not suffice to be picked up and analyzed by the software.
Moving into 2022 it is crucial for marketers to adapt to a shift of mindset when it comes to analyzing and adjusting their campaigns. A clear call is to think and plan marketing activities holistically and take care of finding an independent measurement solution. Data-driven analysis is not gone - it is just moving towards a privacy-friendly direction and demands to utilize statistical knowledge, data-science and engineering to help you out with growing your business with marketing.
As a marketer getting more creative with messaging and ad-creatives will still help to optimize further on top of the funnel improving CTRs (click-through-rates) because the insights of whether or not an ad is clicked on remains on the advertising platform and can still be optimized for.
Finally it is a good time to come back to the roots, let platform hacking and campaign optimization rest and focus on optimizing the user experience onsite and improve the products companies are selling.