Longitudinal Content Telemetry and Attrition Dynamics in Open-Access Digital Interventions: Physical Yoga Cohorts (2014-2025)

This dataset presents longitudinal user telemetry from a non-guided digital Yoga intervention hosted exclusively on an open-access video platform (YouTube). The data spans an 11-year observation period (August 2014 – December 2025). The intervention follows the standardized curriculum of the book "Yoga en casa: Curso para principiantes" (Rollán, 2017), progressing through theoretical foundations (requirements, precautions) to practical execution (Pranayama, Asanas, and sequences). Unlike previous datasets in this series that leverage dual-platform environments (LMS/YouTube), this intervention represents a "pure" open-access algorithmic consumption framework. It serves as a real-world control group to study behavioral friction points, such as the "Theory Barrier" before physical practice and algorithmic attrition. The dataset includes granular video retention metrics, playlist attrition rates, and longitudinal view timelines, offering real-world evidence (RWE) on self-directed physical activity adherence in the absence of financial or instructional accountability. ### Dataset Structure & Reuse PotentialTo ensure full compliance with the FAIR data principles, the repository is structured as a Tabular Data Package validated under the Frictionless Data standard. It comprises interconnected CSV files covering video-level performance metrics per chapter (N=19) and longitudinal annual aggregates. Methodological caveats, the explicit bounce-filtering thresholds (Interested Views), and playlist exit rates are thoroughly documented in the accompanying README file to guide proper statistical modeling and survival analysis forecasting.