October 7th, 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.
Branch recently hosted its first-ever Agency Advisory Council; a close-knit group of mobile marketing and analytics leaders from top agencies. Branch brings this group together to discuss important ongoing industry developments and trends, share strategies and leading practices, and to provide input and feedback on Branch’s overall strategy and product roadmap.
No surprise, the topic du jour was how changes in iOS14 and Apple’s intent to limit access to the IDFA would affect clients’ mobile marketing and measurement strategies. Although Apple gave us all a reprieve in September, the topic remains top of mind to help agency clients plan for a transition in 2021.
Quick note: This article does not provide in-depth explanations for many of the concepts behind these IDFA-related agency discussions. For more on the fundamentals related to the loss of IDFA in upcoming Apple iOS releases, the use of Apple’s SKAdNetwork, and steps you can take to prepare, visit Branch’s iOS14 landing page and read related articles on our blog.
Branch opened the conversation by setting the context for what is really affected here. With so much hype around the potential loss of IDFA, it often gets lost that paid, app-to-app campaigns are the primary mobile media vehicle impacted by the loss of the IDFA. Don’t get me wrong, app-to-app campaigns are a big driver of mobile marketing and return on ad spend, but it’s also important to point out that there’s an array of paid and owned channels (web-to-app, email, owned banners, QR codes, etc.) that are more-or-less unaffected by the change — and where Branch deep linking and attribution still function as they always have.
Our discussion focused initially on the issue of retargeting and how this capability will be affected. Jon Goodman, President of Range Digital Marketing, noted that his team is taking a close look at enabling clients to leverage their registered user base for retargeting “We’re assuming that the SANs [self-attributing networks like Facebook and Google] will continue to view email addresses and phone numbers from registered users as part of a custom audience and we can retarget as we normally would. That said, we’ll miss out on users who installed but did not register.”
Liz Emery, Senior Director of Mobile App Strategy at Tinuiti, said that her team isn’t assuming retargeting will no longer be an option, but that how it works may change, “SANs are not providing a definitive stance that retargeting will go away or a particular product won’t work. But we’re talking about the changes in how those models are powered. There are different inputs you can use with your Facebook and Instagram retargeting. Same thing with Google for re-engagement campaigns.” But she also adds that, “We’re having much harder conversations on the display side given how traditional display retargeting has worked. How all this will change is anyone’s guess right now.”
Emery went further to say that the team is looking at a three-pronged approach to help Tinuiti’s clients prepare: “One, we’re keeping a close eye on how SANs are changing how they’re powering retargeting. Two, we’re helping clients look at what first-party data they have and how that can be used for maximum impact on each [advertising] platform. Lastly, we’re looking at ways to optimize conversion events from a CRM perspective.”
Heads were nodding on the subject of CRM, leveraging owned channels, and fostering engagement beyond paid media. “This is becoming a larger and larger conversation with clients. How they should be looking at other parts of the funnel and getting them to think about first party data differently,” Emery added.
As we know, the concept of moving budget from iOS-targeted campaigns to Android is always a possibility. On this subject, our agency advisors felt this may help in the short-term, providing an option for clients to increase their Android investments during the transition to see how some of these changes play out and how the market reacts.
Finally, web-to-app campaigns are also seen as a viable alternative more and more clients are asking about. “Mobile web inventory is coming up a lot in discussions with clients versus neglecting a big iOS audience because we can’t see the granularity we want,” said Courtney Hirose, Group Director of Performance Media at Horizon.
Rounding out the conversation on SKAN workarounds, Branch also notes here that an interesting rationale is shaping up around using the web to onboard your customers and then send them to the app. Matej Horava, Product Manager at Branch, points to a compelling recent post from Analyst and Strategy Consultant Eric Seufert, where he demonstrates how using the web for your onboarding flow can save money and enable more granular attribution for app installs. However, this comes with caveats like adding friction in the onboarding process that can cause dropoff as well as not being able to optimize for in-app events.
Agency Advisory Council members also got an early, inside look at the work Branch is doing with SANs and non-SAN ad partners that will be taking a SKAN-only approach and partners that plan to use both SKAN and Branch’s device-level attribution (based on predictive modeling and Branch’s vast link graph that sees anonymous touches across billions of users). We also discussed how Branch works and is evolving its products, integrations and dashboard views to support all cases.
As topics shifted to working with SKAN campaigns and what it means to have much less granularity in attribution and campaign performance data, the group discussed how they’re preparing to test in this environment. Addison Wheeler, Assoc. Ad Operations Director at PMG Advertising Agency, emphasized the importance of a good baseline. “For one of our retail clients, we know how app revenue breaks down for Android versus iOS. We know we’ll start seeing our attribution data erode as we stop getting iOS data but Android will remain consistent. So we can look at a lot of these factors together, set a wide net and test different scenarios to isolate learnings.”
While testing will be critical, several participants indicated that they’ll be doing this on a small scale early in the process. Without a clear understanding of what SKAN data may or may not be able to tell us, the strategy here is to make lighter investments and run smaller tests early in order to just to see how the reporting looks and what insights can be gleaned.
Conversion values are fundamental to Apple’s SKAdNetwork solution and its ability to provide advertisers with attribution data at an aggregate level. Following app installs that resulted from an SKAdNetwork campaign, the advertiser passes these conversion values (single integer values from 0-63) back to Apple in order to help the advertiser understand, at a very high level, which campaigns drove what kinds of downstream in-app events and behaviors.
How best to use conversion values in SKAN for campaign measurement and optimization remains a bit of a gray area, but agency leaders are beginning to formulate their strategies and tests. Sorting out the best way to use conversion values likely depends a lot on clients’ industry category and the types of events and outcomes that are most important. Addison Wheeler from PMG highlights a common approach that uses conversion values to track purchase amounts: “Our retail client is looking at using conversion values to bucket different purchase amounts. We haven’t yet determined the ranges. Working within the 63 values for ecommerce, our goal is to get it as granular as we can get and as close to what we normally see as possible.”
The group also briefly discussed using conversion values to track user journeys or life cycle events. There was general agreement that this may not be as useful for ecommerce, but could be useful for some clients, depending on their industry category and how they measure success in their apps.
Lastly, other comments and advice around how to prepare included tactical things like keeping tabs on partner documentation for how campaign setups will be done in SKAN and how many campaigns can be run simultaneously. Further, it was noted that agencies should be having conversations with clients about impacts to their dashboards and analytics.
The industry has now long-relied on device-level data to understand and improve campaign performance. Among our council group, we sense determination to continue relying on device-level data for as many touchpoints across the user journey as possible going forward, regardless of the fate of the IDFA. As the session wound down, Branch shared its long-term vision with this group for supporting multiple attribution methods: aggregate data from SKAN, deterministic matching of device IDs like IDFA (where still available) and GAID, as well as our predictive modeling for attribution.
Each method will have its strengths and weaknesses. For example, the advantage of SKAN is full deduplication (Apple will give credit to last-click and they see everything registered with SKAN) but you have the issue of less granularity based on aggregate data, and the limitations of its numeric conversion values (0-63). Branch’s strength will continue to be the use of tracking links and predictive modeling to tie upstream clicks to downstream events. In this way, running both attribution methods in parallel will have its benefits; SKAN as an effective currency (buy and bill on unduplicated attributed events) and Branch tracking links and predictive modeling to drive more value through optimization of spend across channels and ad partners.
For more information about Branch’s agency programs and partnerships, contact Bill Westcott, VP Solution Partners, email@example.com.