Every team that sells on Telegram eventually has the same idea. We have 300 customers in our chats, we have a new offer, why are we writing to them one by one?
Then someone tries a blast tool, sends 200 identical messages in an hour, and the account that held five years of customer relationships gets restricted. The idea was right. The execution was the problem.
Here is what we have learned about messaging your own customers at scale, respectfully, and how to know whether it was worth doing at all.
Rule one: message people who know you
Telegram’s anti-spam systems care most about one signal: do recipients want your messages? Replies, existing chat history and zero reports keep an account in good standing. Scraped lists and cold pitches to strangers produce blocks and reports, and a handful of those can end an account. If cold outreach is your plan, Cavyro is the wrong tool; this guide is about your own customers.
This is why we built Cavyro campaigns as warm-only. A recipient qualifies only if there is already a conversation between you. It is a product constraint, and it is also the single biggest safety feature: your audience is made of people who have literally messaged you before.
Rule two: send like a human, not like a cannon
A person on a busy day sends a message every so often, with gaps that vary. A script sends one per second. Telegram can tell the difference.
What safe pacing looks like in practice:
- Randomized spacing between sends, not a fixed interval.
- A daily cap per account, in the low hundreds, not thousands.
- An immediate stop when Telegram returns a flood warning, with the campaign paused rather than retried.
- An automatic pause if a spam signal appears, so a bad afternoon does not become a dead account.
Cavyro does all of this by default and you cannot turn it off. Slower sending is the price of doing it properly, and it is a price worth paying.
Rule three: stop when they reply
Nothing says “automation” like a follow-up that arrives after the person already answered. If you run a sequence, a reply should remove the recipient from every remaining step, immediately.
In Cavyro this is called stop-on-reply and it is on by default. The drip exists to get a conversation started. Once it starts, the human takes over.
The part almost nobody measures
Most tools tell you how many messages were sent. Some tell you who replied. Neither answers the question your accountant would ask: did the campaign make money?
To answer that, the campaign tool has to see your deals, which is why it belongs inside the CRM. Cavyro ties every reply back to the campaign that earned it, then watches what happens to that person’s deals inside an attribution window: moved forward counts as conversion, closed won counts as revenue. A finished campaign report reads like a sales channel, because that is what it is: sent, replied, deals moved, revenue won, per A/B variant.
Once you see that report, the question changes from “how many messages can we send” to “which message earns the most per recipient”. That is a much better business to be in.
A starting playbook
- Segment by tag or pipeline stage, and exclude anyone the offer does not fit.
- Write two variants and let the A/B split pick the winner instead of your gut.
- Two or three steps at most, days apart, with stop-on-reply on.
- Send during your working hours, from the account the customers already know.
- Read the revenue report before you plan the next one.
If you want the whole thing handled, Cavyro campaigns do exactly this on top of your mirrored chats, and the full plan is free for 5 days.