There’s a number in every community proposal you should ignore: member count.
A Telegram group with 40,000 members and 30 messages a day is worse than one with 900 members and 400 messages a day. Exchanges, investors and partners all check this now. A dead server with a big number attached tells them exactly what happened — you bought members.
The 48-hour test
Here’s the cheapest diagnostic we know. Stop posting for two days.
If the channel keeps talking, you have a community. If it goes silent, you have an audience with a chat box attached — and the fix isn’t a community agency, it’s better content and a clearer reason for people to show up.
Most teams fail this test. That’s fine. Knowing which one you are decides what you should actually buy.
What a community programme is for
Community work is worth it when the thing you’re building needs people to do something together. Governance participation. Liquidity provision. Bug reports. Content creation. Regional translation. Referrals that come from genuine belief rather than a payout.
Community is a distribution channel you own. That’s the real argument for it. Every other channel — paid, press, KOLs — you rent. When a campaign stops, the traffic stops. A community that’s genuinely alive keeps working between campaigns, and it’s the only asset in marketing that appreciates while you sleep.
But it only becomes that if there’s something to gather around.
Signals you need it
People are already talking about you somewhere you don’t control. A Telegram group somebody else started. A subreddit thread. That’s demand, and it’s the strongest signal there is.
Your product improves with more participants. Marketplaces, DeFi protocols, games, DAOs, anything with network effects. More people genuinely makes it better, so gathering them has compound value.
You’re about to launch a token. Distribution decisions, governance and the first months of holder behaviour all run through community. Doing this at TGE and not before is a mistake we see constantly.
Support volume is rising and you’re answering the same question forty times. A community with informed regulars turns that into peer support.
You’re going multi-region. Local-language communities in Hindi, Arabic, Vietnamese, Turkish or Bahasa aren’t a nice extra — they’re often where the actual users are.
Signals you don’t
You have no product yet. A community with nothing to do becomes a price-speculation channel within a week, and you’ll never get it back.
Nobody on your team wants to be in there. Community can’t be fully outsourced. Someone from the actual team has to show up, take the hard questions and be a person. If nobody will, buy moderation instead and be honest about it.
You want the member count for a deck. Say the quiet part out loud and skip the spend.
Discord or Telegram
We get asked this weekly, so here’s the short version.
Telegram works for announcement-led, price-aware, mobile-first communities. It dominates in India, the Gulf, Turkey, Vietnam and across Eastern Europe. It’s fast, it’s loud, and moderation is a permanent job.
Discord works for builders, gamers and anything that needs structure — separate rooms, roles, searchable history, bots that do real work. Its strength in the West is genuine. Its weakness is that new members land in a wall of channels and bounce.
X is where the conversation gets seen. It’s not a community; it’s a stage. Treat it that way.
Reddit rewards being useful and punishes being promotional harder than anywhere else. Great for protocols with a technical story, brutal for anything that smells like marketing.
Pick one primary. Run it properly. Two half-staffed servers is the most common self-inflicted wound in this category.
What actually builds one
Not giveaways. Not gm campaigns. Not a bot that hands out roles for reacting to a message.
What works is boring and repeatable:
- A daily human presence from someone who knows the product and can answer a hard question without escalating.
- A reason to come back that isn’t price — a weekly call, a build showcase, an office hours slot, a testnet that changes.
- Real moderation with published rules and consistent enforcement. Scam bots in a project’s own Telegram are the fastest way to lose the members who matter.
- Regional leads who speak the language and understand the local market, not translated announcements.
- Public recognition for the people doing the work. Contributor roles, named credits, early access. Status is a stronger currency than tokens in a healthy community.
For Metarix, community engineering ran alongside KOL seeding and PR through a round that raised $780K. The point wasn’t the number of members. It was that the members who were there could explain the project to someone else — which is the only thing that makes a community worth having.
What drives the cost
Community isn’t priced by member count, and any proposal that does that is selling you the wrong thing. What changes the shape of the work:
- Number of channels and languages. Each region needs its own lead, not a translated post.
- Coverage hours. Crypto communities don’t sleep. Follow-the-sun moderation is a different staffing model to business-hours support.
- Moderation intensity. Anything with a token attracts scammers at a volume that surprises first-time founders.
- Whether governance is live. Running proposals, votes and delegate relations is a specialist job.
- How much internal participation you can commit. The more your own team shows up, the less agency time it takes — and the better the result.
The honest recommendation
If you’re pre-product, open one channel, keep it small, and let it be quiet. A slow-growing group of people who actually care is the foundation. You can’t buy that later, and every shortcut you take now shows up as a dead server with a big number on it.
If you’re post-product and people are already talking, staff it properly. That’s the moment community stops being a cost line and starts being the cheapest distribution you’ll ever own.
Common questions
What is the difference between a community and an audience?
An audience consumes what you publish. A community talks to itself when you are not in the room. The test is simple — go quiet for 48 hours. If the channel dies, you have an audience with a chat box attached. That distinction decides whether you need a community programme or a better content calendar, and most teams get it wrong because member count looks identical from the outside.
Should we use Discord or Telegram?
Telegram suits announcement-led, price-aware, mobile-first communities across Asia, the Gulf and Eastern Europe. Discord suits builder communities, gaming projects and anything needing rooms, roles and searchable history. Most projects need one done properly rather than both done badly. Running two half-staffed servers splits your regulars and doubles your moderation load for no gain.
Do paid community members ever convert?
Rarely, and the damage outlasts the campaign. Members who joined for a reward leave when the reward stops, and while they are there they drown out the people who actually care. Worse, a 40,000-member server with 30 daily messages is a visible negative signal to exchanges, investors and anyone else running diligence on you.
How do you measure community health?
Daily active posters, the ratio of member-to-member replies versus team-to-member replies, retention of members past their first 30 days, and how many questions get answered by someone who does not work for you. Member count is the one number that tells you almost nothing, which is exactly why it appears in every agency report.
When is a community programme the wrong spend?
When you have no product for people to gather around, when nobody on your team will show up daily, or when what you actually need is support staffing. A community programme requires a real internal owner. If nobody on the team can commit to being present, moderation is the honest purchase, not community building.
