How Wooga Uses Data Science to Supercharge User Acquisition
With your experience, what tips you would give to advertisers to help navigate the programmatic side of UA?
Patience is a virtue when it comes to programmatic advertising. First, it takes some time for AI models to understand user behaviors and be able to look for others who perform similarly to the ones hitting or surpassing your monetization or retention goals.
Second, it is common to find yourself with cohorts of users that initially look rather weak, especially if you are looking into their monetization, but that stick with your game and start producing results later on in their life cycles. This happens more with programmatic than with any other type of advertising. We have had more than a few cases where we stopped programmatic campaigns due to their initial numbers looking subpar only to resume them again some months later, as they overachieved long term goals.
Put on your futurist hat and think about how mobile advertising will change in the next 2–5 years? What changes do you foresee in the mobile ecosystem.
Currently there are two main trends that I am seeing in the mobile ecosystem:
- A shift towards user privacy that is being led by both Google (with the phasing out of 3rd party cookies on the Web) and Apple (with the app tracking transparency framework). Advertisers will have to learn how to deal with the loss of data. As campaign targeting and optimization accuracy decrease, it will be paramount to find ways to balance that out; I believe the first of those will be observed by the increase in importance of data scientists.
- Business consolidations which are happening on several levels in the industry — publishers buying publishers, small or big (ex. Zynga acquiring Peak Games and Rollic), advertisers acquiring advertisers (ex. Digital Turbine buying Fyber and AdColony), advertisers arming themselves with full mobile suits (ex. Applovin purchasing mobile measurement company Adjust and Machine Zone gaming studio). This will increase the importance of having a strong brand and deep pockets. The bigger your own network is, the more first party data you will have as a publisher — critical in our new world of privacy. This new reliance on 1st party data will mean fertile soil for internal cross promotion for the bigger brands.
Performance metrics, like CTR, CVR, 1-DAY, 3-DAY, 7-DAY, etc., are something UA folks have relied on, but they don’t always tell a complete story. Can you share a story of where a media source had metrics that were unusual and didn’t match true performance?
Those metrics help marketers understand the initial performance of a cohort of users as well as aid data scientists in forecasting the lifetime value of these cohorts. However no one can guarantee a perfect accuracy of a forecast — specially when we work with products and consumers that are constantly evolving.
I have seen many cohorts of users that outperformed their original forecasts. They looked inauspicious at first, but after weeks, in some cases even months after install, they started monetizing and became profitable within their life cycles. Typically, marketers cannot afford to wait to decide whether a campaign is worth the money invested in them or not. So while one should still act on early signals, reviewing campaigns later is important to confirm the original assumptions materialized — and if not, rectify the earlier decision.
Sticking with metrics — tell me a metric that isn’t universally available that you’d really love to have.
Qualitative uninstall rate — or understanding of why a user uninstalled. That might be impossible to have, however with that information one could both try and improve their product to prevent further similar churn as well as develop the best messaging to speak to the users when trying to bring them back again to their app (re-engagement).
How do you view incrementality — particularly when dealing with non-traditional methods? Do you see pros/cons in industry-standard approaches when it comes to incrementality?
Being able to measure the lift that advertising spend provides in conversion rates of re-engaged user cohorts is necessary to make sure you aren’t burning your money. Depending on the size of the retargeting, you will need to adapt the proportion of your holdout group, so that you can have statistical significance in your comparison in a timely manner. If you cannot measure incrementality you’d be taking a bet on whether or not you are winning — or at least by how much.
Let’s try a lightning round. Separately give me quick thoughts on each of these:
1) App Store Optimization
Key for the user journey. A store page that is not optimal will make the users you brought via paid ads drop before installing your app.
2) Investment in New UA Methods
If you depend on only one means of advertisement you will miss the chance to hit audiences that are active elsewhere (TV, blogs, influencer followers etc.).
3) Programmatic UA
Growing in importance — You cannot depend on social and search only to advertise your app online.
4) DSP Differentiation
Is real! Each DSP has its own optimization models that produce different results to your campaigns. Learning to deal with these will produce positive results.
5) iOS 14
Every advertiser’s challenge in 2021. Marketers will have to relearn their game: craft messages that speak to both new and existing users; convince users they want to be tracked; and understand holistically whether their paid efforts are successful.
