Where your first ten viewers actually come from
Nobody comes to your first stream. Or your second. This is the part that makes people quit, and it is almost always because they are waiting for an audience to arrive rather than going to find one.
Here is where the first handful of people actually come from.
Discovery does not happen on the platform
This is the thing to understand before anything else. Twitch and YouTube are not going to hand you viewers. Their recommendation systems are built to push people toward streams that already have an audience, because that is what keeps people watching. A channel with zero viewers has nothing to recommend.
So the platform is where people watch you. It is not where they find you. Those are two different places, and almost every new streamer confuses them.
Your first ten come from somewhere else, and they come one at a time.
Other people's communities
This is the big one, and it is the one people resist because it feels like work rather than streaming.
Find three or four small streamers you actually enjoy. Not big ones. People with five or ten viewers, roughly where you are. Watch them properly, talk in chat like a person, and keep turning up. Do this for a few weeks before you expect anything back.
What happens is not that they promote you. It is that you become a regular somewhere, and the other regulars notice you, and some of them will come and see what you do. Small communities overlap heavily. Ten people who all watch each other is how most of this scene actually starts.
The failure mode is obvious and everyone can spot it. Turning up, dropping a link, leaving. Do not do that. It reads as spam because it is spam, and the community you were trying to enter will remember.
X, which is still where VTubers are found
For better or worse, X is the discovery layer for this scene. It is where model reveals get shared, where clips circulate, where artists post, and where people find new VTubers to follow.
What works there is not announcing that you are live. Nobody sees those posts and nobody clicks them. What works is posting things worth looking at. Your model. A clip. Art you commissioned. A short thought about something you are playing.
A useful habit: post once a day that is not a stream announcement, and once when you go live. The first one is what gets you found. The second is for people who already follow you.
Clips do the work while you sleep
A stream is two hours long and nobody is going to watch two hours of a person they have never heard of. A forty second clip of the best thing that happened is a different proposition entirely.
Cut two or three clips from each stream. Post them as shorts, as TikToks, on X. Most will do nothing. One in twenty will get seen by a few hundred people, and a handful of those will look at your channel.
This is slow and unglamorous and it is the single highest leverage thing a new streamer can do with an hour of their time.
Other small streamers, on purpose
Collaborate early and collaborate sideways. Not with people far bigger than you, who get asked constantly and will say no. With people at roughly your size, who are in exactly the same position and usually delighted to be asked.
Two streamers with eight viewers each who play something together end the night with two audiences that have met. Do that a few times and the overlap compounds.
Raids work the same way. Raid out every single stream, to someone still live, every time. It costs you nothing, it is the normal etiquette of the platform, and it is how people learn your name.
What does not work
Follow-for-follow. You get followers who will never watch you, which makes your numbers look better and your streams feel identical.
Self-promotion channels in Discord servers. Everyone posting, nobody reading.
Buying anything. Viewbots are detectable, they get channels banned, and a stream with forty fake viewers and no chat is more obviously dead than a stream with three real ones.
Waiting to be ready. There is always one more thing to set up. The overlay can wait.
One thing the Japanese scene does differently
On the Japanese side, debuts are treated as events rather than as the first stream. A date is announced, the model is revealed ahead of time, there is a countdown, and people arrive because they were told when to arrive.
Most English-speaking VTubers just quietly go live one evening. There is nothing wrong with that, but it wastes the one moment where curiosity is naturally highest.
If you have not debuted yet, announce a date two weeks out, post the model before it, and give people a reason to turn up at a specific time. It is the cheapest audience you will ever get.
How long this actually takes
Longer than you want. Most people who stick with it are looking at months, not weeks, before a stream feels alive.
The useful measure early on is not viewer count. It is whether anyone came back. One person who returns three weeks running is worth more than forty people who passed through once, because that person is the start of a community and the forty are not.
What to do next
Pick three small streamers this week and become a regular in one of their chats.
Start clipping. Two clips a stream, posted somewhere.
If you have not gone live yet, read the first stream checklist before you do, and set a debut date rather than drifting into it.