Apple has now started to implement its App Tracking Transparency framework. Marketers now need to seek a user’s consent to use their device for ad tracking. What happens when consent isn’t given? Grant Simmons, VP of Kochava Foundry talks through an alternative approach to campaign attribution post IDFA…
For years, marketers in the mobile ecosystem have received real-time postbacks where the attributed clicks and associated app install and launch have been transmitted in under 200 ms. In a programmatic world, where marketers are making data models and ad buying decisions in real-time, that postback data is the oil that feeds the engine, and mobile measurement is a huge part of the feedback loop.
As of April 26th, Apple will begin enforcement of their App Tracking Transparency (ATT) framework which requires marketers to prompt users for consent to use their device for ad tracking. If user consent is granted (opt-in), the identifier for advertisers (IDFA) of iOS devices will be accessible. Deterministic attribution is possible, and the feedback loop remains. If users don’t consent (opt-out), the IDFA will be unavailable to marketers. Without that transparency granted by the user on both the ad and install side, there will be no insight. ATT is going to change measurement as we know it. Full stop.
A new way of measuring in town: The update ConversionValue
When user consent is not granted, measurement won’t be in real-time, and Apple will perform aggregated and delayed attribution via its SKAdNetwork. Without consent, there’s only one game in town for gathering insights: the updateConversionValue. This is a value from 0-63 that’s stored on the device and gives marketers 64 combinations to understand the relationship between paid media and conversion activity. Conversions include any downstream activity after the install; these include registrations, levels completed, purchases, features used, etc. The more you want to know, the longer you’ll have to wait for a SKAdNetwork postback because of a timing system.
Every time a conversion is reported, a 24-hour timer is started: If the container is updated again within 24 hours, a new timer starts. Once the 24-hour timer elapses, another random 24-hour timer begins. When the second timer completes, that’s when the postback occurs. The real-time feedback loop that used to happen in under 200 ms could take up to 60 days!
The upside is the updateConversionValue. This is a value that is 6-bits worth of data, housed in a byte (8-bit) container. That’s 8 bits or 64 combinations (of 0’s and 1’s), and it has enabled us at Kochava to create four conversion models to help marketers understand the user journey. We use the first 3 bits to time gate conversion activity, but the model can be modified from day 2, 3, or up to day 7 post-install. This allows marketers to close that feedback loop and get more timely reporting. We hold that measurement window open in some clever ways to get performance and tie back conversions to paid media. In short, the updateConversionValue is a big deal and a way to understand attribution of iOS campaigns in the absence of the IDFA.
Driving forward with incrementality testing
With the majority of users expected to opt-out of ATT, we’re looking at the end of the real-time feedback loop. If we use the analogy of a car, real-time postbacks have allowed us to look ahead at the road. With such delayed postbacks, we now have to look in the rearview mirror by running back tests like incrementality measurements. (Note: Many people use the terms “attribution” and “incrementality” interchangeably, but they are VERY two different things. Attribution means tying a conversion back to the initial impression/click, and incrementality is determining the impact or lift of an ad.)
With ATT and the SKAdNetwork, incrementality testing will become more important because it helps answer the question: In the absence of an ad, who would have converted anyway? If you can figure that out–you’re moving forward. The problem is that it’s not in real-time because you’re using prior data. This is cohort back-tested measurement to understand the overall efficacy of your ad.
There are some novel and more affordable ways to perform incrementality testing. Using an independent data set, such as with the Kochava Collective, we can create a comparable, unbiased control (holdout) group instead of withholding advertising from a portion of a marketer’s advertising population. This synthetic control saves marketers money because they can advertise to their entire audience, and they remove bias. It answers bigger questions in terms of what did these ads in aggregate do?
While to some extent, row-level real-time device measurement will not be possible with iOS users who don’t consent to tracking (ie, opt-out of ATT), measuring incremental lift is an extremely meaningful way to analyze campaigns.
A new measurement paradigm
A whole new world of ad measurement is coming, and there is much to consider. The updateConversionValue and conversion models are new, complicated considerations for marketers. How are you going to extract the most intelligence out of an 8-bit protocol and its 64 combinations? How will you use it to understand the user journey? It’s not an easy puzzle to solve, but it is solvable through the conversion models.
Then, flip that car around with new measurement paradigms to understand ad spend using incrementality testing to understand how those ads performed.
While measurement of iOS campaigns is changing drastically, not all hope is lost—we’re just answering the same questions in a different way.
Grant Simmons, VP, Kochava Foundry