Spotify’s goal is listening time. By investing in mixing human curation with machine learning, they created franchises like Discover Weekly and Spotify Wrapped. Algorithms analyze listening history to serve music people didn’t know they loved. Personalized playlists now drive over 30% of total listening time, and the viral nature of Wrapped boosted app downloads and made it a phenomenon.
LinkedIn optimizes for session frequency to shed its “job-search only” reputation. The “People You May Know” (PYMK) feature leverages weak ties to trigger connection requests, while the feed algorithm now prioritizes “knowledge and advice” over viral clickbait. Users with strong networks return more often; leads generated here are 6x more likely to convert than on other platforms, justifying premium ad rates.
TikTok’s model creates immediate engagement to drive ad revenue. The “For You” feed doesn’t rely on your social graph; it relies on “signal density.” This interest-graph model keeps users glued for ~23 hours/month. Consequently, narrative-driven ads on TikTok see 70% higher ROI than standard social ads.
While their metrics differ, retention is no longer about features; it is about how well you know the user.
