Growth
A Discord server's first 100 members are the hardest
The first hundred are the hardest and the least like the rest. Almost everything written about Discord growth assumes you already have them.
Most Discord growth advice assumes an existing community. Optimise your listing, run events, improve retention — all reasonable, all useless at eleven members, because every one of them requires a room with people in it.
The first hundred is a different problem, and it has a different solution.
The empty room problem
A new server is unattractive for a reason that has nothing to do with its quality: there is nobody in it. A member who joins and finds four people and no conversation leaves within a minute, and they were right to.
This is circular, and you cannot break it by improving the server. You break it by removing the requirement that the room be busy.
What to have ready before anyone arrives
Less than you think, and the list is short on purpose. Spending three weeks configuring a server nobody is in is the most common way a launch never happens.
Three or four channels. See below.
A one-sentence description in the server’s own name or topic, so somebody arriving remembers why they came.
A welcome message with one next action. One. Not a rules wall.
Two or three rules, in the channel where people talk, rather than a separate rules channel nobody opens at this size.
A moderation bot, configured minimally. Not because you expect trouble at twelve members, but because the first spam wave arrives without notice and configuring under pressure goes badly. Basic automod and a log channel is enough — Vetox’s eight rule types and 148 log event types are free on Starter, which means the protective layer costs nothing to have ready.
That is the whole list. Everything else — levelling, tickets, custom roles, an economy — is built for a community that exists, and adding it now produces a server that looks like an abandoned office rather than a new one.
Do not launch to the public
The instinct is to open the doors and advertise. It fails, because the first arrivals see an empty room and leave, and you have spent your one launch on an audience that will not return.
Invite privately first. Ten to twenty people who already know you, individually asked, with a reason. Not “join my server” — “I’m starting something for X, you’d be good in it.”
The difference in acceptance rate between a broadcast and a direct ask is enormous, and the people who come from a direct ask actually talk, because they came for you rather than for the topic.
What the ask should contain, in one message: what the thing is, why you thought of them specifically, and what you would like them to do when they arrive. Three sentences. A message that could have been sent to anyone reads as a broadcast even when it was not, and converts like one.
Make the room small on purpose
Twenty members in a server with three channels feels like a group. The same twenty in a server with twenty-five channels feels abandoned.
Launch with three or four channels. General, one topic-specific, one off-topic. That is enough. You can always add; you cannot easily undo the impression of emptiness.
The most common mistake in a new server is building the channel structure you expect to need at a thousand members. It guarantees that every channel is dead for the first six months.
Post as though people are watching
For the first few weeks, the server needs someone in it who is reliably present. Usually that is you.
Not announcements. Conversation — reacting to what people say, asking follow-ups, sharing things in the topic channel. A new member’s decision to stay is made on whether the room looks alive, and for the first month you are the thing making it look alive.
That presence does not have to be constant to work. A counting channel — members taking turns counting upward, one number at a time — runs on nothing but itself: no topic, no wit, no audience, and two people alternating is enough to keep a chain going between the hours you are online. Vetox’s Counting is free on every tier including Starter. The same is true of a quick match: eleven of Vetox’s forty Games are built for exactly two players, so “someone is in the server” can mean a duel rather than a conversation you have to personally carry, and every one of the forty is free regardless of tier. Neither replaces you. Both cover the hours you cannot.
This is the least scalable and most important work in the whole exercise.
The first thirty days
A rough shape rather than a schedule, because the numbers are small enough that a single week can move them.
Week one — the private invite. Ten to twenty individual asks. Expect roughly half to join and a handful to talk. Be present every day. Do not advertise anywhere.
Week two — make it a habit. Introduce one recurring thing at a fixed time: a question, a check-in, a match. It matters far more that it is predictable than that it is interesting. This is the week to notice which of your invitees actually engage — they are your first regulars, and they are worth talking to directly.
Week three — the second ring. Ask your active members whether there is one person they would bring. Not a broadcast, the same individual ask, delegated. This is where a server first grows without you doing all of the work, and it is the earliest real signal that the thing has value to someone other than you.
Week four — the first outside channel. Now, and not before, a listing or a mention in a community you belong to. The room has conversation in it, so an arriving stranger sees something.
What changes at each milestone
At twenty, the room stops being a group chat and starts needing a reason to open it. This is where the recurring thing earns its place.
At fifty, conversations begin overlapping and one channel starts feeling crowded. Add the second topic channel now, not earlier — split when the room is too busy, never in anticipation.
At a hundred, the dynamic changes qualitatively. See below.
What to avoid
Mass DMing. It converts poorly, it reads as spam, and Discord’s own systems treat it as spam. Getting the account flagged is a real risk and a permanent cost.
Invite trading. Swapping invites with unrelated servers imports arrivals with no reason to stay, which pushes conversion down and makes the room look worse, not better.
Bots before people. A twelve-member server with nine bots reads as a project rather than a community. Add capability when there is someone to use it.
Rebuilding after three quiet days. New servers are quiet in bursts. Restructuring in response to normal variance resets whatever familiarity your first arrivals had.
Where the first hundred actually come from
In rough order of effectiveness for a genuinely new server:
Existing relationships. Anyone you already talk to who fits. Highest conversion by a distance.
A community you are already part of. Not spamming — being a known contributor somewhere and mentioning what you built, once, where it is on-topic and allowed. Value first.
A thing that exists outside Discord. A tool, a guide, a video, a spreadsheet people use. A server attached to something is far easier to grow than a server that is the thing.
Server listing sites. Disboard and similar work, with a caveat: listing traffic converts badly because those visitors are browsing, not seeking. Worth doing, worth not relying on.
Paid promotion. Almost never worth it at this stage. You will buy joins from people with no reason to stay, and the join graph will look great while the active-member count does not move.
When the first bad actor arrives
It happens earlier than people expect, usually the week after a server first appears on a listing, and at twelve members it is disproportionately damaging: one person posting something ugly is a third of the day’s messages.
Two things make it survivable. Have the automod on already — a filter configured calmly last week beats one configured furiously now. And act immediately rather than proportionately. In a large server a first offence deserves a warning; in a twelve-member server, the cost of being slightly too firm is far lower than the cost of your handful of regulars watching you tolerate it.
What to measure before a hundred
Not member count. Unique people who posted this week.
At this stage that number will be small enough to count by hand, and it is the only one that predicts whether the thing survives. A server going from 40 to 80 members with the same four people talking is not growing.
Vetox’s server overview will eventually do this counting for you — a members and messages card and a seven-day chart, free on every tier, live the moment the bot joins. At this scale your own memory outpaces any dashboard, and that is a fine reason to not bother opening one yet.
The threshold
Somewhere around a hundred engaged members — not a hundred joins — the dynamic changes. Conversation starts without you. Someone answers a question before you see it. Two members talk to each other about something you did not raise.
That is the point at which the standard growth advice starts applying, and the point at which you can stop being the room’s life support.
Before it, none of that advice works. After it, most of it does.
Frequently asked questions
How many channels should a new server have? Three or four. General, one topic, one off-topic. Split a channel when it is too busy, never in anticipation.
Should I list a brand-new server on Disboard? Not in week one. Wait until an arriving stranger would see conversation rather than an empty room.
How many people should I invite privately first? Ten to twenty, individually, with a reason specific to each. Expect roughly half to join.
What counts as a hundred members? A hundred engaged members — people who post. A hundred joins with four people talking is not the same threshold.
Do I need bots at twelve members? One, configured minimally, for moderation and logging. The rest can wait for people to use them.