Concurrent Streaming, Viewer Transfers, and Audience Loyalty in a Creator Ecosystem: A Minute-Level Longitudinal Study

A new longitudinal study by Maxwell Shepherd provides an unprecedented, minute-level look at audience behavior within a live streaming creator ecosystem. Analyzing a massive dataset of 2.9 million viewership observations across 7,762 streams from 18 affiliated channels over 3.3 years, the research quantifies the complex dynamics of competition and loyalty. The study reveals that when multiple creators from the same organization stream simultaneously, average viewership per stream plummets from 14,377 to just 6,057 viewers as the number of concurrent streams rises from 1 to 9. However, after controlling for scheduling patterns, the residual correlation is a more modest ρ = -0.165, indicating that poor timing accounts for much of the raw drop.

The research also introduces metrics for tracking viewer movement and loyalty. Algorithmically detected 'viewer transfer' events—where audiences move from one ending stream to another starting one—show a median efficiency of approximately 50% across 3,243 candidate events. Perhaps most impactful for talent management is the finding that audience loyalty is a creator-level trait, not an organization-level one. Within the same network, creators' 'competition resistance' scores—measuring how well they retain viewers when others go live—varied from 0.36 to a perfect 1.00. Other loyalty metrics like stability, retention, and floor ratio showed similar wide variation. This data provides concrete benchmarks for organizations to make smarter decisions about scheduling to avoid cannibalization, designing effective cross-promotion, and identifying which creators have the most dedicated, exclusive audiences worth investing in.