@minh_••••KEEP
“vào lệnh long BTC x10, ăn non r… tuần trước cháy tài khoản”
why Posts his own leveraged entries and a liquidation last week. Trades his own capital, daily.
Telegram audience mining
And next to every one of them, the sentence explaining why they made the list, quoted from what they said themselves.
“vào lệnh long BTC x10, ăn non r… tuần trước cháy tài khoản”
why Posts his own leveraged entries and a liquidation last week. Trades his own capital, daily.
“looking for influencers and traders to bring onto our exchange, DM open”
why Exchange BD stating the need out loud. Buys audience as a job.
“VIP room 99$/mo — 87% winrate, limited slots, DM now”
why Sells signals to the same people you want. A competitor, not a lead.
Three rows out of a chat with eighteen thousand members. The other 17,997 were read too, which is the part nobody else does and the only reason the three above can be trusted.
Collection
Seven passes, and the order is the point. Each one throws away work the next would otherwise waste money on.
Who counts, and who does not. Exchange affiliates but not signal sellers. Traders with capital but not airdrop farmers. This sentence is what the classifier is judged against later, word for word.
A keyword grid built in the language your audience types in, crossed with city and community patterns. Twenty seed terms become a hundred queries.
Search results lie. Every chat is sampled for messages a day, unique authors and spam share. Half to two thirds are dropped before one contact is harvested.
Participants plus what they themselves wrote. A handle with no history can only be guessed at. Accounts without a username are dropped, since nothing can reach them.
Each candidate read against your brief in the original language. Back comes a verdict, a reason quoting their own behaviour, and a summary. Rejects ship too.
Every handle ever classified sits in one index. Nobody in your base was messaged for another client, and nobody known is paid for twice.
One message per person, built from their own stated need. Where data is thin it stays general instead of inventing a detail.
Before an event, its chat fills with people introducing themselves: who they are, what they sell, who they want. That is a filled-in questionnaire nobody had to ask for, and it expires. We run those chats on the event calendar, two weeks out and one week after.
Who we find
Audiences are not equally dense. Traders are everywhere once you know which chats to read. Serious affiliates are not. Each is quoted on its own instead of blended into one flattering number.
Ranges, not guarantees. A tighter brief lowers the count and raises the quality, and that trade is better agreed before the first invoice than after it.
Delivery
Shipped columns: handle, name, company, role, segment, source chat, what they want, what they offer, summary, verdict, reasoning.
| Handle | Verdict | Reasoning |
|---|---|---|
| @lan_•••• | KEEP | Accumulating monthly since 2023 and asking how to move funds off the exchange into a wallet. A long horizon holder with real capital. |
| @marcel_••• | KEEP | Runs affiliate partnerships for an exchange and asks the room for partners. States the need himself, so no inference required. |
| @tuan_•••• | THIN | Three messages, all one word reactions. No evidence of trading anything, so the brief cannot be confirmed either way. |
| @airdrop_••• | DROP | Only ever posts referral links and farms testnet tasks. No capital behind the account, so volume without value. |
Sending
Delivery is an infrastructure problem, not a copy problem. Accounts belong to your campaign alone, are warmed before they carry anything, and run well under the ceiling rather than at it.
Throughput
Scaling means adding warmed accounts, never pushing existing ones harder.
Size this against how many replies your team can answer. Sending more than you can handle is how a good base gets wasted.
Measured
Ours, from campaigns we ran ourselves. Your reply rate depends on your offer, and showing you the mechanism beats quoting someone else's conversion.
Next
Send the profile you are after. Back comes the realistic monthly number for that audience, the chats it would come from, and a sample of qualified rows with their reasoning, so the targeting can be judged before anything is agreed.
Message us on Telegram