Netmarble’s Vampir is a dark fantasy MMORPG designed for players who value immersive storytelling, real-time combat, and long-term progression. As with many MMORPGs, the game’s success depends not only on strong creative appeal but also on its ability to attract high-quality users who are likely to engage, retain, and spend over time. In Korea—one of the world’s most mature and competitive gaming markets—this challenge is magnified by high user expectations, intense competition, and rising acquisition costs.
Korea’s mobile gaming ecosystem is highly saturated, particularly in the MMORPG category. Players are deeply familiar with the genre, highly selective, and quick to churn if early experiences fail to meet expectations. For marketers, this creates a delicate balance between aggressive scale and strict efficiency—especially during launch, when performance signals are still volatile.
Netmarble’s core objective was clear:
achieve ROAS targets while maintaining launch volume and CPI stability.
However, reaching this goal meant overcoming several constraints at once: ensuring sufficient scale during the initial rollout, avoiding CPI inflation in premium inventory environments, and identifying users with genuine monetization potential—rather than short-term engagement alone.
To address these challenges, Netmarble partnered with Appier to gain a deeper understanding of payer behavior at a granular level. Appier’s AI models analyzed cross-app usage patterns to distinguish payers from non-payers beyond surface-level demographics.
The analysis revealed that in Korea, communication apps—particularly KakaoTalk—play a central role in daily user activity, accounting for more than half of payer engagement. While this insight may appear intuitive, it highlighted a critical strategic implication: creative and inventory strategies must be tailored specifically to KakaoTalk environments to resonate with high-value users.
When communication apps were excluded, clearer behavioral differences emerged. Payers demonstrated stronger engagement in Finance, Comics, and Games, while non-payers skewed toward Health, Lifestyle, and Entertainment categories. These distinctions enabled Appier’s models to more accurately predict purchase intent and refine targeting strategies.
Leveraging Lookalike Modeling Through Behavioral Intelligence
To build high-quality lookalike audiences at launch, Appier’s model first focused on identifying what differentiates payers from non-payers at a behavioral level—rather than relying on broad demographic assumptions.
Data analysis revealed that communication apps account for approximately 57% of payer activity, underscoring the central role of messaging environments such as KakaoTalk in the daily routines of high-value users. This insight helped define a strong behavioral baseline for payer identification and guided inventory prioritization during the launch phase.
When communication apps were excluded, further distinctions emerged. Payers showed consistently higher engagement in Finance, Comics, and Games, while non-payers demonstrated stronger affinity toward Health, Lifestyle, and Entertainment categories. These category-level signals allowed the model to distinguish monetization intent from general engagement, improving the precision of lookalike audience construction.
In addition to identifying where high-value users spend their time, Appier’s model analyzed when payers are most active, revealing a clear and structured weekly rhythm. Payer engagement consistently peaked from Friday through Saturday, particularly within Finance and Games categories—indicating a strong overlap between leisure time and spending intent. Midweek activity dips further highlighted routine-driven usage patterns, providing valuable signals for pacing and delivery optimization.
These temporal insights directly informed inventory strategy on KakaoTalk, where communication apps already accounted for approximately 57% of payer activity. Appier’s AI automatically prioritized high-performing KakaoTalk placements, including Bizboard formats, to align delivery with peak payer activity while maximizing visibility in high-intent moments.
To ensure cost efficiency during this high-impact exposure, Appier implemented a floating CPI model with an upper-bound guarantee, dynamically adjusting bids in real time while preventing sudden cost spikes post-launch. This approach enabled consistent, scalable delivery throughout the critical early phase of the MMORPG launch—ensuring that reach, timing, and cost control worked in unison.