A channel with 18,000 subscribers was holding 11% ERR consistently. Then it dropped to 7.8% in one week — same content, no complaints, nothing obvious. Turned out 600 bot accounts had joined over three nights and never opened a single post.
What spam bots actually cost you
Telegram calculates engagement rate as a share of your total subscriber count. Every 200 inactive bot accounts in a 10,000-subscriber channel costs you roughly 2% ERR. At 1,000 bots that's 10% gone — silently, without any notification.
Ad buyers check ERR before placing orders. A visible drop with no explanation creates doubt that doesn't go away just because you cleaned the bots out later. The reputation damage outlasts the attack.
Bots arrive in two ways. The first is a targeted attack: a competitor pays a service $3–5 per thousand ghost subscribers. The second is passive — your channel ends up in traffic reseller databases and starts receiving accounts from shared pools.
How TGuard identifies spam bots
TGuard maintains a collective database across all connected channels. Every time an admin bans a user — manually or through automation — it's recorded. An account banned in five or more channels in the network gets marked as a known attacker.
This is documented ban history, not behavioral profiling. Five independent admins in different channels all decided this account was a problem. That's a pattern, not a coincidence.
When the antivirus fires
TGuard tracks how many known attackers join your channel within a one-hour window. When three or more show up in that window, the channel owner gets an alert.
The alert includes the exact count of suspicious accounts and the list of invite links they used to get in. If the attack came through multiple links, all of them are listed — so you immediately know which ones to revoke.
Auto-ban vs. notify-only
By default, the antivirus notifies and leaves the rest to you. You can enable automatic blocking: when the threshold is hit, TGuard bans all confirmed attackers in that batch without waiting for your input.
Regular subscribers aren't affected. Only accounts from the attacker database get banned. Like any antivirus, there's a small false positive rate — about 1–2%.
Users who solved your channel's captcha are excluded from the check — even if their account appears in the attacker database.
Works for join requests too
If your channel uses join requests instead of open subscriptions, the antivirus adapts. TGuard auto-rejects requests from known attackers before they ever enter your subscriber list — the bot doesn't just get removed, it never gets in.
How that queue works on the owner's side — where requests live, what the requester sees, and how to clear hundreds at once — is covered in Telegram Join Requests.
Why this catches what captcha misses
Captcha is effective against cheap bots that can't interact with a button. Professional attacks use "aged" accounts — ones with real activity history and a normal-looking profile. They pass captcha without any trouble.
TGuard identifies these accounts through their ban history across the network, not their behavior at the door. Even if an attacker passed your captcha, the network database already has their record from five other channels.
Anti-bot protection vs. a spam filter
"Anti-bot" gets thrown around loosely. Half the tools that claim it are keyword filters that delete messages after they're posted — useless here, because the accounts flooding a channel rarely post anything at all. They just sit in your subscriber count. A message filter has nothing to filter.
Real anti-bot protection has to act at the join, not the message. TGuard checks every account against the network ban record the moment it tries to enter, and against the invite link it arrived through. An account with a record in five other channels doesn't get a second look — it's flagged before it ever counts toward your total. That's the whole difference between cleaning bots out later and not letting them in.
How to enable it
- Open @channel_guardian_bot and select your channel.
- Go to Security.
- Find Antivirus and turn it on.
- Optionally, enable automatic blocking to have attackers banned without manual action.
Frequently Asked Questions
- What counts as a known attacker in TGuard's database?
- An account that has been banned in five or more channels connected to the TGuard network. The classification is automatic — every admin who bans a user contributes to the shared database. No manual setup required on your end.
- Can the antivirus accidentally ban a real subscriber?
- The false positive rate is about 1–2%, similar to any antivirus. TGuard only acts on accounts in the confirmed attacker database. Users who solved your channel's captcha are additionally excluded from the check, even if they appear in the database.
- How do I know if my channel already has spam bots in it?
- The clearest signs are an unexplained drop in engagement rate (ERR) with no content changes, a sudden subscriber spike without any active ad campaign, and low views relative to total subscribers. TGuard's Cleanness Report shows a breakdown of suspicious accounts in your current subscriber base.