Home   >   Features exceeds Animoca Brands' Crazy Defense Heroes ROAS KPIs by 70% - Mobile UA case study

How the two companies joined forces to acquire more users in Japan exceeds Animoca Brands' Crazy Defense Heroes ROAS KPIs by 70% - Mobile UA case study

Who's involved?

Animoca Brands is a global mobile and blockchain game developer and publisher with offices in Hong Kong, USA, South Korea, Finland, Argentina, and the Czech Republic. Animoca Brands is a leader in the field of digital entertainment, specialising in blockchain, gamification, and artificial intelligence technologies to develop and publish a broad portfolio of mobile products.

Their leading original titles include The Sandbox, Crazy Kings, and Crazy Defense Heroes. Animoca Brands also develops mobile games based on popular intellectual properties such as Formula 1®, Marvel, Garfield, Snoopy, Thomas & Friends™, Power Rangers, MotoGP, and Doraemon. is a tech product company driving mobile user acquisition and re-engagement for brands and app developers worldwide. Using our proprietary bidder and machine-learning algorithms, we offer highly-targeted & transparent solutions designed to reach and beat retention and ROAS KPIs quickly and cost-effectively.

We strive to be more than just a vendor for our partners, but a partner that helps generate actual value, growth, and broad marketing insights that can be used across channels.

Campaign Goals

We worked together with Animoca Brands on one of their leading titles, "Crazy Defense Heroes," aiming to increase their user-base and acquire highly engaged users who would make in-app purchases in order to reach their ROAS KPIs.


When we started running UA campaigns in Japan for Animoca Brands' "Crazy Defense Heroes" on iOS in April, our initial approach was intent-based targeting. We targeted users that showed direct interest in multiple relevant genres – Tower Defense, Strategy, and Card Games. This approach allowed us to acquire users with a significantly higher than usual D1 retention rate for the genre.

Then, our machine learning models gradually learned the patterns of the users that were most engaged or made purchases on the day of their install, which allowed us to target similar users and eventually resulted in beating the defined ROAS KPIs. We were also able to reach a recoup period of 45 days.

Reaching KPIs in iOS14 Using Contextual Targeting

While the depreciation of the IDFA makes user-level targeting impossible, early testing across multiple gaming and other verticals confirmed that by focusing our efforts on contextual targeting, driven by machine learning, allows us to reach similar and sometimes better campaign performance.

Similarly to the "Crazy Defense Heroes" campaign, where we targeted users based on their interest in similar genres, the contextually based method we designed for iOS14, would target users while they're using contextually related apps - related by not just their features, but also their theme and specific mechanics.

In order to execute based on these data, we use an NLP algorithm named ELMo. Using ELMo we're able to represent each App Store's (and Google Play) app description as a vector and calculate a distance between those vectors, effectively measuring their contextual 'distance'.

Using words in context allows the model to detect nuances in the description and find contextually related apps based on the app genre, mechanics, features, themes, and more. In the case of "Crazy Defense Heroes", it meant targeting other apps that included aspects of the tower defense, strategy, and card game genres.

ELMo uses a neural network to learn associations between words and their meaning when used in different contexts, by learning in what phrases they are used inside a massive 5.5 billion token (words and their composing parts) data set. Since considering each word is crucial for understanding context, ELMo proves to be one of the most reliable options for such a complex task.

The inability to target at the user-level does not mean you need to compromise on the quality or scale of your UA campaigns. Using ML-driven contextual targeting, reaching KPIs and scale is still very much attainable. To learn more about our contextual targeting methods, you can try our Context Calculator, which shows how we use ELMo based context analysis in our UA campaigns, and contact us to learn more.

"The Animoca Brands team was cooperative and communicative. They provided us with all of the initial data we needed in order to reach and exceed their ROAS KPIs in the particular genre of Tower Defense in the competitive Japanese market," said Maor Kreichman, VP, Strategic Partnerships at

Cindy Kong, Senior Product Associate at Animoca Brands added: "It was a pleasure working with the team. They would share and discuss the plan of achieving our goals and ensure that we were crystal clear about every step. During the campaign, they reviewed the performance consistently and made optimizations in order to exceed our target ROAS."