Kabam chose Aarki for a user acquisition campaign to broaden the user base of their newest role-playing game (RPG) simulation app, Shop Titans. The campaign objective was to deliver a broad audience reach and attract highly engaged users that would generate a strong return on investment (ROI) at a low cost per install (CPI). The campaign also aimed to deliver global scale through Aarki’s large data repository and machine learning capabilities.
The combination of machine learning and creative optimization guaranteed the success of the campaign. We at Aarki had a clear objective of increasing brand awareness and acquiring high lifetime value (LTV) users who would drive strong engagement and retention rates.
Machine Learning Optimization
We started the campaign with the machine learning process that consisted of an Install Optimization phase followed by an Event Optimization phase. During the Install Optimization phase, which aimed to drive a large volume of new app installs at the lowest possible cost, our team leveraged intelligent suppression lists to reach high LTV users for the app. Unlike traditional suppression lists where inactive users are being excluded, the intelligent suppression model bids on all users not suppressed and benefits greatly from the post-install behavior of inactive users.
Once the Install Optimization phase was complete, we proceeded to the Event Optimization phase. Leveraging Kabam’s first-party data, we built and launched the Multiplicative Model to optimize the campaign to drive in-app purchases. This Multiplicative Model allowed us to make predictions derived from non-attributed event data.
We simultaneously captured short-term fluctuations and bias in the bid stream and conducted deep learning into user profiles. The model was used to predict post-install events and LTV of a user at bid time, which boosted the overall performance of the campaign. As a result, install probability was close to install rate and in-app purchase probability was close to purchase rate, indicating successful model calibration.
Once we had the predictions, it was time to design the creative. To better understand users’ motivation, we started with creative research to discover what makes the app unique and which features users find compelling. With our deep category expertise and client insights, we designed builder and gameplay ad creatives featuring the game’s UI and mechanics.
We also used game messaging and assets in all creatives to properly describe the game and inspire the target audience to play it. We focused on creating customized creative for each distinct audience segment.
The multi-faceted nature of the Shop Titans app, which includes shop and character customization, resource management, trading, crafting and questing, allowed our creative team to compose personalized messages featuring each of these aspects. As we proceeded, we conducted champion/challenger testing to identify the best performing creatives. Once we identified the champion creatives, we increased the media spend allocation and ramped up the campaign performance by focusing on the most engaged audiences. Our team set new challenges for the champion creative again and again to keep the winning creative fresh and compelling.
The campaign surpassed all client expectations including Cost Per Install, Return on Investment and volume metrics.