The impact of initial experience and user attachment on application downloads: Information-seeking and -sharing applications

The purpose of this paper is to examine the dynamic effects of users’ initial experience and user attachment on downloading information-seeking (e.g. news, map, education) and -sharing (e.g. messenger, chatting, social networks) applications.

From one of the largest application stores in South Korea, 225 applications were examined through analysis of 1.5 years of download records. Logistic regression and Bayesian models including time-varying coefficients were used.

Over time, the download patterns of users of the app market become dynamic. In the initial stage, users have a tendency to download apps of similar types. For example, users who initially download information-seeking applications continue to download these types of applications more frequently than information-sharing applications. In the later stage, however, users tend to download applications that accord with their attachment intensity. Users who want to share information with others are more likely to download information-sharing applications as compared to information-seeking ones. Finally, this tendency persists with the accumulation of more experience.

This study applies existing models and theories from previous research to the app market, such as state dependence, intrinsic motivation, and time-varying coefficient models. However, this study focuses on information-seeking and -sharing applications. Therefore, further study is needed in order to extend the findings to other types of applications, including games, paid applications, and so on.

This study provides valuable information for marketing managers running application stores and app developers who want to maximize performance. In the initial stages when users download apps, app market managers must recommend apps with consideration of both the users’ attachment and their initial experience, but after app users have accumulated some experience, only user attachment should be considered as a criterion for recommendations.

This is a unique study in which users’ download behavior regarding information-seeking and -sharing applications is analyzed using long-term actual data.