Digital marketing and user acquisition trends are evolving at a rapid pace. With changes in costs, privacy regulations, ad formats, and increased competition, mobile marketing and UA managers have to adapt very quickly.

Amidst all the uncertainties and competitive market space, how do game developers gain a competitive edge?

The answer is predictive technologies. Prediction uses artificial intelligence (AI) and machine learning (ML) techniques to identify the likelihood of future outcomes based on historical data. The key development here is that it produces useful insights that marketers can use to deliver tangible business value like delivering a higher return on ad spend, faster.

To address the complexity of predictive analytics and its importance, we spoke in depth with Claire Rozain, UA Lead at Rovio, a Finland-based mobile-first gaming company that creates, develops, and publishes mobile games, including the ever-popular Angry Birds series.

Claire began her professional career as a marketing assistant in Antidot and since then worked with Geosys, ITinSell, Mappy, Match, Pixel United, and Gameloft. She’s also the founder of Puzzle Society, an invite-only community of those who are passionate about puzzle mobile games.

As a UA manager, is prediction something that you're currently using?

Prediction has always been important to me. UA is not something you're doing just for spending money - you expect returns on investments, hope to minimize the Delta of the risk, and spend money in a way that makes sense: not only for one single event but for long-term conversion and retention that will allow the game to grow and scale.

Without prediction, you can't go further than the data you get. With prediction, you can step up your strategy and forecast user LTV. Not only sticking to past facts makes user scoring way easier, for example.

When it comes to prediction, you look at certain metrics, which ones are key, and do they signal different things for you?

The more I'm doing UA, the more I value Cost Per Retention. It's a really interesting metric when you want to assess the stickiness of your game and elaborate good strategies. I also like to have a close look at predicted LTV, Investment Projection, and Level Funnel. In the end, those are more product-oriented than UA-oriented, but they’re key to successful games.

Do you find that depending on what area of gaming you're in, you're looking at different KPIs, or are they very similar?

I get to speak with many people from different game genres. When I'm brainstorming with people from hyper-casual, their KPIs are different from the ones I use at Rovio - because the business model is not the same. When you monetize through IAP, the strategies and revenues are different from when you monetize through IAA. You need to take those facts into consideration in order to predict the right data at the right moment for the right games, but also for the right business plan. So yes, the KPIs vary. But at the end of the day, it's the same for everyone: it’s not just growing the game to grow the game. It's growing the game to be profitable, entertain users, and perfect the user experience.

In any case, the first thing I look at is the payback. How much time did it take to create the game? How many resources and investments? Based on that I can calculate a profitable ROAS. You have to keep in mind that it's more than just the cost of acquiring a customer, there's a cost that is associated with producing a game too.

In terms of data science and prediction, to what extent do you feel like data science and non-SKAN measurement are going to replace traditional measurement?

The industry evolves so quickly, it's difficult to predict anything right now.

But in my opinion, UA managers will still need to work (maybe even closer) with data scientists. Blending data is going to be more and more important, which is great - we relied on accurate deterministic data for too long.

Do you feel since iOS 14.5, UA managers or UA teams have had to work much more closely with data science teams, whereas before it was quite siloed?

Yes. At Rovio, we have data scientists that are part of the UA team so for us, it’s normal to team up.

Nowadays onboarding data scientists in UA teams is a must for any app studio - we need to evangelize data science and share knowledge on how it evolves - not only for the UA team but for all people, especially key stakeholders, that are looking at the paid installs.

In this new era, UA Managers need to deep dive into analytics and be less impulsive and emotional. When you're doing prediction, you have to rely more on data.

In terms of succeeding in predictive modeling or embracing predictive modeling. Do you have any tips or things that you would look out for when it comes to embracing this new world of UA and predictive modeling?

The keys are patience, proper thinking, and the right human resources that have a good understanding of the company. And be up for some good challenges.

External solutions can be of great help, as creating predictive modeling is very complicated. In any case, following the trends and staying informed is very important.

How do you feel privacy has impacted predictive modeling? Since we're going to be moving towards a more privacy-centric universe, do you think it is a better case for more companies to lean into predictive modeling, or why not?

When I think of predictive marketing, it's not something that is new to me - nowadays it’s just becoming more and more important because of privacy. Prediction is in a sense, a way to back up the loss of data and has to be at the center of any business strategy now.

I personally think it's great that we have more privacy. As a user, I don’t want companies to access my data and identify who I am.

Our job is to understand our users and what they want, to always perfect the entertainment.

Now that predictive modeling is growing, privacy is growing. What key trends do you think are going to emerge in 2022 or 2023? When it comes to measurement and optimizing.

Facebook just launched IPA Attribution, which is all about data aggregation. So I think the trend of aggregation is not going to slow down any time soon.

The holistic trend is going to grow as well, with incrementality measurement, even though it’s still quite complicated to achieve. I believe we’re on the road to finding new processes to optimize the understanding of our aggregated data.

These were some valuable insights from Claire as to how predictive modeling is a key aspect of mobile marketing in the data-privacy era. As users get more aware of how their data is being used, prediction is a way to back up the loss of data.

Want more information on how to grow your gaming app and stay competitive in a dynamic market? Download our 2022 User Acquisition Guide for Gaming Apps here.

‍