Samsung Electronics has revealed a case study testing the practical application of artificial intelligence (AI) agents in business using games. By applying AI to marketing in the Galaxy Store, the company analyzed user behavior and recommended games, resulting in a significant increase in click-through rates compared to previous methods.
Lee Hee-jin, a partner in Samsung's MX division, presented the Galaxy Store AI marketing agent case study on September 30 at the Microsoft Industry Summit held in Seoul's COEX.
In collaboration with Microsoft, Samsung has organized data accumulated in the Galaxy Store for AI utilization and established a structure where multiple AI agents collaborate according to the workflow of actual marketers. Since last year, the company has been refining its data platform and has recently completed a proof of concept (PoC) to push for limited domestic commercialization and global expansion.
During the summit, Lee explained how AI could transform marketing strategies in the Galaxy Store, using games as a primary example. The process from user exposure to clicks, installations, and payments is directly linked to the Galaxy Store's revenue. This allows for easy comparison of marketing effectiveness when AI is applied, as conversion rates at each stage can be quantified.
However, as the scale of services in the Galaxy Store has grown, limitations in traditional marketing methods have emerged. With over 10,000 games available and an increasing number of promotional targets, it has become challenging for existing staff to analyze user behavior and design campaigns.
Complex data analysis required developer assistance, and the process of segmenting users and defining targets often relied on marketers' experience or past success stories.
Samsung introduced AI agents into this process, utilizing AI for data analysis, campaign planning, user segmentation, and message generation. They selected two representative games to validate the actual marketing effectiveness among Korean users.
Lee noted that the goal was not merely to replace marketers' tasks with AI. Samsung set a specific target to increase the click-through rate (CTR) by 15% after users were exposed to the games. Subsequently, they conducted A/B testing with approximately 1 million users in Korea, promoting the same games through both traditional marketers and AI agents.
The results exceeded initial expectations, showing more than double the improvement over the target. Lee stated, "The CTR for the group using AI agents increased by 105% compared to the traditional method."
He attributed the significant increase in CTR to the segmentation of users into detailed groups and tailoring messages to the characteristics of each group. Unlike previous generic messages like 'A new game has been released, try it out,' AI analyzed user behavior data to understand group characteristics and create tailored game recommendation messages.
According to Lee, this PoC focused on increasing CTR after game exposure. Further validation is needed to determine whether improved CTR leads to actual game installations and payment conversions. Samsung plans to develop installation prediction models and purchase conversion models to expand the application of AI.
Lee highlighted that the success of this experiment was made possible by the data Samsung has accumulated over the past eight years. He explained, "The organization of user behavior and marketing-related data in a format usable by AI has been the foundation for this PoC."
Since last year, Samsung has been working with Microsoft to ensure that existing data can be utilized by AI agents. They have organized data where the same terms had different meanings or calculations across departments, enabling AI to find and analyze the information marketers need in their actual work.
Lee stated, "Based on this PoC, we plan to achieve some level of commercialization domestically by the end of the year and then expand globally. We will verify whether actual results continue in the Galaxy Store and aim to apply this to various Samsung services in the future."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.