Unpacking the exploration–exploitation tradeoff on Snapchat: The relationships between users’ exploration-exploitation interests and server log data

Diego Gomez-Zara, Yozen Liu, Leonardo Neves, Neil Shah, Maarten Bos
Event CHB
Research Areas User Modeling & Personalization

Social media users continuously decide between looking for familiar content and exploring novel content. Although prior research has extensively studied the trade-off between exploration and exploitation (E/E), little is known about how it operates on social media platforms. In this observational study, we created survey scales to measure users’ exploration and exploitation interests (EEI) on Snapchat, a social media platform, and used them to survey 1,073 Snapchat users in the U.S. between December 2020 and January 2021. We combined users’ self-reported responses with their aggregated behavioral data recorded on the Snapchat platform. We employed correlation analyses and matching procedures to analyze the relationships between users’ EEI and their activities on the app. Our results suggest that exploration, and not exploitation, triggers more behavioral activities while using this social media platform. We found that exploration was positively correlated with spending time on the app, having more content impressions, and watching content from familiar sources. Moreover, exploration was negatively correlated with communication activities. By combining users’ self-reported and server log data, this work advances our understanding of how the E/E framework applies to users on a social media platform.