"High likes mean the algorithm will push it"

"High likes mean the algorithm will push it"

A comedy account gave me a gift: two videos, same week. Video A — 4,200 likes, 38k views. Video B — 1,900 likes, 210k views.

The high-like video underperformed the low-like one by 5x on reach. Correlation isn't causation, and here they actively diverged.

The mechanism: likes are a lagging vanity signal the system reads weakly. Completion rate is the lever. Video B was 9 seconds with an 82% completion rate; Video A was 34 seconds finishing at 41%. The longer video earned more likes from people who stayed, but lost the retention race that drives expansion.

Likes measure who already liked you. Completion measures whether strangers will. Optimize the second number and the first one follows on its own.