Core Concepts
Users engage with short-format videos on TikTok, watching most videos until the end but skipping some before completion, indicating a balance between content recommendation and user retention.
Abstract
The study analyzes user engagement on TikTok through data donations, focusing on video consumption patterns and interaction metrics. It explores how users engage with short-format videos and the role of recommendation algorithms in content consumption. The analysis includes time spent, volume of videos watched, attention per video view, and interactions through likes. Insights reveal that users watch most videos until the end but skip some before completion, suggesting a strategic approach to content recommendation for user retention.
Introduction: Shift in social media consumption towards short-format videos.
Data Collection: Recruitment process, dataset statistics, metadata collection.
User Engagement Analysis: Attention per video view, percentage of videos watched till the end.
Temporal Analysis: Changes in volume of watched videos and time spent over time.
Following Accounts: Number of followings over time and views from following accounts.
Stats
By analyzing user engagement on TikTok using data donations:
The average daily usage time increases over users' lifetime.
User attention remains stable at around 45%.
Users like more videos uploaded by people they follow than recommended ones.
Quotes
"No participant watched until the end more than 65% of the videos."
"Participants watch most of the video's duration before proceeding to the next."