Solutions
Teams
Built for your whole team.
Industries
Trusted by all verticals.
Platform
Use Cases
Many Possibilities. One Platform.
Home
Solutions
Teams
Built for your whole team.
Industries
Trusted by all verticals.
Platform
Use Cases
Many Possibilities. One Platform.
Pricing
Resources
Company
FAQ
Guest Post by Paul Bowen
As we roll into 2022, we take stock of the things we achieved in 2021 and set our new year's resolutions for 2022. While Tim Cook takes stock of his achievements this year over a glass of mulled wine, along with celebrating the increase in stock price of Apple by a whopping ~40%, he may also give a wry smile to his achievement of completely transforming mobile marketing on iOS.
Under the guise of a privacy friendly approach to user tracking, Apple released App Tracking Transparency (ATT), which kneecapped the ability of mobile measurement partners (MMP’s) to be able to track users for the purposes of mobile marketing optimization, attribution and measurement. Although some MMP’s are still using probabilistic attribution to match installs with campaigns, previous guidance from Apple indicates this is not a long term solution to the problems that ATT posed.
ATT created several key problems for the industry:
That said, this version of SKAdNetwork, the de facto attribution solution for iOS, was really an MVP. Before ATT had been released it had never really been stress tested in the wild. So it is that 8 months after ATT rolled out, most advertisers are still trying to figure out what ATT really means for their business.
Advertisers today have a bandaid solution to the problems posed by SKAdNetwork and ATT. Without the ability to attribute the long term impact of marketing efforts, many advertisers at scale on Facebook and Google are using blended ROAS as their new north star KPI. This is not a viable solution to growth; it’s imperative that every advertiser understands the impact of each ad network and campaign to effectively be able to allocate budgets.
2020 was a simpler time. It was simple to track the short-term returns of mobile marketing campaigns through access to user-level attribution which gave a rich record of every install, revenue event and engagements by a mobile app user that could be attributed to a campaign and ad network deterministically. This data could then be easily aggregated up at the campaign and ad network level and fed into a simple LTV model that modeled what the optimal ROAS or cost per acquisition would be for a similar ad network or campaign.
In gaming, the predominant models were the D7 ROAS model and the ARPDAU / retention models. Although these models had their flaws, they were simple to implement and therefore widely used.
With ATT, the cohort LTV model became defunct as both models relied on ad network and campaign reporting of revenue through attribution data, no longer accessible through SKAdNetwork. Therefore we need an alternative way to measure the long-term performance of marketing campaigns…
Some of the hardest problems posed by SKAdNetwork can be solved with a (marketing) data scientist. Using key available data (SKAdnetwork data, MMP attribution, cost data), the data scientist can work to solve the attribution and measurement challenges posed by SKAdNetwork. Although SKAdNetwork is suboptimal, it does have some real benefits because it’s a last-touch, deterministic, campaign-level attribution solution.
This is a real win for data scientists as these are key components that make SKAdNetwork attribution data comparable with MMP attribution. Data scientists don’t need to completely throw away their SKAdNetwork reporting and use a model that leverages tops down inputs without using the maximum value of their “bottoms up” model. An example of “tops down model” would be a Media Mix Model that tries to model causal relationships between impressions and clicks and downstream events such as installs and revenue.
Using SKAdNetwork data and MMP attribution, a bottoms up model can first model the long-term returns of campaigns or ad networks before taking into account a tops down approach for validation. However - a tops down approach is important, as it can provide key validation for the underlying model.
We’ve briefly covered the significant body of work that data scientists can focus on to solve the attribution and measurement problems raised by ATT, but it also created a significant problem with creative optimization. Without the ability to measure the performance of creative at the user-level, marketers lost the ability to test and analyze the performance of their assets in a systematic way. Prior to ATT, an industry standard approach to testing creative had emerged, although not supported by specific ad network creative testing tools, marketers were able to derive enough insights with their hacked process to determine what creative performed and what didn’t. ATT killed this creative testing process. Marketers are now left wondering what’s the best approach?
Outstanding questions include:
These attribution, measurement and creative problems are significant challenges and band aid solutions such as blended ROAS, or educated guesses, will prolong the time until when they’re solved. They can also be solved in a minor way by marketers, but for the most part, it’s absolutely necessary to either hire these marketing data scientists or partner with companies who are fully focussed on solving these problems. So set your resolution to solving these challenges in 2022 and make sure your growth function either by hiring or partnering with a data science driven company so it continues to operate at the highest level.
About Paul Bowen Paul is the General Manager of AlgoLift by Liftoff+Vungle, and is a 20 year veteran in the ad tech industry, having acted as a VP at Unity and Tapjoy previously. You can find Paul on Linkedin and Twitter.
About Paul Bowen Paul is the General Manager of AlgoLift by Liftoff+Vungle, and is a 20 year veteran in the ad tech industry, having acted as a VP at Unity and Tapjoy previously. You can find Paul on Linkedin and Twitter.
About AlgoLift
AlgoLift transforms data inputs into actions with our predictive algorithms and investment optimization technology. Our platform continuously analyzes customer data to give teams a real time understanding of user-level LTV then programmatically leverages this data to improve ad network and campaign performance.