From my own testing, the first thing that moves the needle quite a bit is comments. I use them on pretty much every upload. When a video goes out and people start talking under it fairly quickly, it does seem to push things along. On videos where no one comments at all the difference is honestly massive. Feels like the algo reads it as a dead room.
Sharing helps as well, but lately something else seems to matter more than I realised before. The viewer coming back to your channel and watching more videos. I only started paying attention to that after chatting with a bloke on here. When someone watches one video and then jumps to another one on your channel, that signal seems quite strong. It’s basically telling YouTube the viewer found what they were looking for.
Another thing people overlook is the whole user session thing. What a lot of people do is grab a few phones, open the video when it drops, watch it, like it, then close the app. On paper that looks like engagement, but in reality it’s a pretty weak signal. The session just dies.
If that “user” stays on the platform, watches other videos, maybe browses related channels, spends time there, that view carries a lot more weight. YouTube likes sessions that keep people inside the platform. If your video is part of that session, it helps.
I’ve tried pretty much every setup over the years. Manual, automation, groups, all sorts. One thing that definitely matters now is the account history. Age range, interests, what kind of videos the account usually watches, country, all that stuff. If the account’s behaviour already lines up with your niche, the interaction seems to count more.
So yeah, comments and early activity still help, but the quality of the viewer behaviour behind it matters way more than people think. Just blasting quick views and likes isn’t the same thing anymore. The algo’s a bit smarter than that these days.