If I'm asking it to play music, favouring YouTube music means playing it from there

Voice commands aren't going to give you an exhaustive list of possibilities, they're going to play the top result.

Q. "Play Diamonds by Rihanna"

A1. "Playing Diamonds by Rihanna from Youtube Music"

A2. "Playing Diamonds by Rihanna from Spotify"

Either A1 or A2 will happen, but not both. There can be only one.

You don’t always need to have a verb. The way I use my Apple TV is usually to just say the name of the content because I want to pick where it comes from.

But even if you say play, it could still ask you where from and/or confirm it got the right thing. Roku != Amazon Echo.