It’s also numbers. YouTube has given creators tools to literally benchmark thumbnails. You can just see which one does better. The vast majority of people unfortunately are susceptible to the same patterns. In the end, you need people to click on your video or they can’t watch it (which is the point here, to inform people). So here we are.

For example, thumbnails with faces work much better than without (doesn’t really matter what or who the face is). I find that monumentally stupid and weird, it just is what it is.