User Migration across Multiple Social Media Platforms

TL;DR

This study analyzes user migration from Twitter to Bluesky, Mastodon, and Threads using data from over 14,000 migrants within eight weeks of Threads' launch. It combines migration typologies, platform-level activity metrics, cross-platform relationships via active-user trends and Yule's Q, and a Bayesian loyalty analysis to reveal that most migrants retain Twitter engagement despite expressed loyalty shifts, while Bluesky often complements Twitter and Threads exhibit a transient appeal. Methodologically, it leverages platform-specific data collection, influence scoring from IR and RFr, and GPT-4-based stance classification to uncover nuanced migrants' attitudes. The findings illuminate the competitive dynamics among emerging platforms, the inertia of established platforms, and offer guidance for understanding how users navigate multi-platform ecosystems in the face of policy and ownership changes.

Abstract

After Twitter's ownership change and policy shifts, many users reconsidered their go-to social media outlets and platforms like Mastodon, Bluesky, and Threads became attractive alternatives in the battle for users. Based on the data from over 14,000 users who migrated to these platforms within the first eight weeks after the launch of Threads, our study examines: (1) distinguishing attributes of Twitter users who migrated, compared to non-migrants; (2) temporal migration patterns and associated challenges for sustainable migration faced by each platform; and (3) how these new platforms are perceived in relation to Twitter. Our research proceeds in three stages. First, we examine migration from a broad perspective, not just one-to-one migration. Second, we leverage behavioral analysis to pinpoint the distinct migration pattern of each platform. Last, we employ a Large Language Model (LLM) to discern stances towards each platform and correlate them with the platform usage. This in-depth analysis illuminates migration patterns amid competition across social media platforms.

Figures (9)

• Figure 1: The migration flow between Twitter and its alternatives: Mastodon, Bluesky, and Threads. The dashed lines represent the shift of user attention across these platforms.
• Figure 2: The trend of deleted and suspended migrants' accounts on Twitter over time. The red dashed line marks the date Twitter announced its rebranding to "$\mathbb{X}$".
• Figure 3: Network traffic analysis for July-August 2023 between Twitter and the targeted domains. Overlaps show users accessing both domains estimated by Semrush.
• Figure 4: Box plots display influence scores for migrant groups (Bluesky, Threads, Mastodon) and non-migrants (Twitter), highlighting their interquartile ranges. The red dots indicate the mean influence score for each group.
• Figure 5: Active user trends comparing Twitter with (1) Bluesky, (2) Threads, and (3) Mastodon. The blue line indicates users exclusively active on Twitter, the red line represents those only active on the alternative platform, and the green line denotes users active on both Twitter and the alternative platform. The red dashed line marks the launch date of Threads. The y-axis shows the percentage of active users relative to the total migrants in each category of active status on platforms.