Auto-messages fire a set delay after each subscriber joins your bot — minutes or days later — targeted by language, source, tag or category, with retries bounded and blocked users dropped.
Most Telegram funnels leak at the same place. Traffic arrives, the welcome message lands, and then nothing happens until someone remembers to write a broadcast. Drip automation closes that gap: every subscriber gets a sequence timed from their own join moment, so the funnel keeps working on the people who arrived at 3am while you were asleep, and on the people who arrived three months ago.
An auto-message in mySkua is one step: a delay, a set of targeting rules, and the content to send. When a person starts your bot they pass through the onboarding flow — language pick, optional category pick, optional channel-subscription gate, welcome or offer message — and their join time is recorded. From that moment the clock runs. Each auto-message whose delay has elapsed and whose targeting matches that subscriber becomes eligible, and the sender delivers it.
The important property is that the schedule is per subscriber, not per calendar. A broadcast picks a moment and reaches everyone at once. A drip reaches each person at the same point in their own journey, which is what makes the numbers comparable: if step two converts better than step three, that is a statement about the message, not about the news cycle on the day you happened to send. If the concept is new, the drip campaign glossary entry covers it in a paragraph, and the guide to Telegram drip campaigns covers how to design one.
Delivery is idempotent per subscriber per step. A step that has already been sent successfully to someone is not sent again, so pausing the bot, restarting a process or editing a step does not produce a duplicate in anybody's chat.
Delays are configured as an offset from join, expressed in minutes, and they span the useful range: a few minutes for an immediate nudge, hours for a same-day follow-up, days for the long tail of a sequence. There is no shared queue time to fight over — because every delay is relative to an individual join, sends naturally spread themselves across the whole day instead of hammering the API in one burst, which matters more than people expect once a list gets large.
Practical shape for a first sequence:
Because an international audience joins around the clock, offsets in the day range self-distribute across time zones without you modelling anything. Offsets under an hour are the ones worth thinking about hardest: that is the window where the attention your ad or deep link bought is still warm.
A step does not have to go to everyone. Auto-messages can be restricted along four axes, and they combine:
instagram_2026_03 can receive different follow-up from one who arrived from a partner channel — the same product, a different framing.The combination is where this earns its keep. “Portuguese speakers, from the affiliate source, who chose Sport” is one rule set, and it can have a completely different day-three offer from the Hungarian casino segment. More on how the segments themselves are built in audience segmentation.
Each auto-message carries a message body per language, plus optional media and optional inline URL buttons. The same shape is used by campaigns and scheduled broadcasts, so what you learn writing one carries over.
One writing note specific to Telegram: there is no soft unsubscribe. A reader who tires of your sequence does not tick a box, they block the bot, and that removes them from every future campaign as well as this one. That asymmetry should make you shorter and more useful than you would be in email. Four to six steps suits most sequences; the ceiling is redundancy, not length. If two steps do the same job, cut one.
Sending to a large list means some sends will fail, and the two kinds of failure need opposite treatment.
Transient failures — a timeout, a rate response, a momentary API problem — are retried a bounded number of times. Bounded matters: an unbounded retry loop turns one bad minute into a queue that never drains, and the point of a cap is that a step either lands or is honestly recorded as not landed.
Permanent failures are the case where the subscriber has blocked the bot. Telegram offers no delivery path to that user and only they can unblock it, so retrying is pure waste. Those users are dropped from further sends and stop consuming send capacity and stop distorting your delivery rate.
Blocks are also a signal, not just an outcome. Where they cluster tells you what to fix. A spike at one specific step points at that message — the offer is wrong, the tone is wrong, or it arrived too soon. Blocks spread evenly across every step point at cadence or at acquisition: you are either sending too often, or the traffic never wanted this bot in the first place.
Three sequence shapes cover most of what a Telegram funnel needs, and they can run simultaneously because targeting keeps them from colliding.
Onboarding. The first 48 hours. Nothing clever — set expectations, deliver whatever the ad promised, and give the subscriber one action small enough that they actually take it. If your bot has categories, this is where the category-specific welcome offer does its work, because the subscriber has just told you what they want.
Offers. Steps that carry a promo code and a deep link into a campaign. Campaigns have their own active window, their own per-language message, media and buttons, and a one-time activation per user, so a subscriber who has already claimed something is not offered the same thing twice. Give each offer step its own campaign code — this is the single change that makes performance readable later.
Reactivation. The long-offset steps aimed at subscribers who joined, did nothing, and are still technically reachable. These are the cheapest sends you will ever make: the acquisition cost is already paid. Keep them rare and keep them concrete — a reactivation message with nothing new in it is just a block waiting to happen.
A sequence you cannot measure is a sequence you cannot improve, and the fix is mostly structural rather than analytical. Three habits do the work:
The analytics side of the panel covers subscriber growth, daily active users, retention cohorts, campaign activations and delivery stats, which is enough to answer the two questions that matter: where in the sequence people stop, and which step actually produced the activation. For the reasoning behind the metrics themselves, the drip campaign guide goes deeper than a feature page reasonably can.
mySkua is a multi-tenant panel for running Telegram lead-generation bots — one account, several bots, each with isolated subscribers and content, plus campaigns, broadcasts, bot texts, languages, attribution, analytics and team roles alongside the drip automation described here. You can see the rest of the platform on the mySkua home page.
Self-serve access is currently closed: we are in an invite-only private beta and onboard teams in small batches so we can help each one get a first funnel working properly. If you run Telegram traffic and want drips, campaigns and attribution in one place, get in touch and tell us which markets and languages you work in — that context is what moves a request up the queue.
See also: Drip campaign guide · Drip campaign · Audience segmentation.
Want this on your own bots?
mySkua is in a closed beta and accounts are opened in small batches. Tell us what you run and we will come back to you.
Request access