Skip to content

Understanding YouTube

A recent interview with a representative from YouTube aimed at content creators, emphasised how we’ll get more exposure for our videos if we concentrate on satisfying the audience, rather than the algorithms. Now, they would say that, wouldn’t they? But it makes sense, in terms of the way we approach optimising our videos: YouTube designs the algorithm to serve the audience, so understanding audience preferences will help the algorithm favour our content. We can all agree that as time goes on, YouTube will steadily get better at delivering what its audience wants.

‘Watch time’ is important, of course. If viewers are leaving after a few seconds, the conclusions will be inevitable. But YouTube wants to focus on ‘long-term audience value’ too. It’s not hard to design a video that everyone will watch to the end, but the site doesn’t want viewers to be disappointed and go away. How does it determine this value metric? The answer is in data analysis. The ‘long-term audience value’ of a video may be hard to pin down in pure data terms, but correlating data flagged up by much-appreciated videos may be easier to measure. Comments and likes are obviously good, but viewers returning to a channel may be an even more significant indicator.

If you’re a frequent YouTube viewer (and I must say, I watch nearly as much there as ‘regular’ TV nowadays), you’ll probably have thought about what you’re presented with, and why. What keeps us there (and keeps YouTube’s advertising revenue rolling in) is how good the site is at offering suggestions as to what to watch next.

Thinking about what it’s doing is the key to making our own videos work harder.