A Telegram subscriber list is not one audience. mySkua stores four attributes on every subscriber — language, source, tags and self-selected category — and lets your drips, campaigns and broadcasts address them separately.
Most Telegram funnels start out addressing everyone at once: one welcome message, one offer, one broadcast to the whole list. That works while the list is small and homogeneous, and it stops working the moment two traffic sources start feeding the same bot — the person who came from a Portuguese-language channel post and the person who came from an English affiliate link are now receiving identical copy, and one of them is quietly unsubscribing.
Segmentation only works if the attributes are captured at the door — you cannot retroactively ask ten thousand people which language they read or where they found you. In mySkua they are collected during onboarding and by the link the subscriber arrived through, so every message that follows has something to aim at.
Rather than an open-ended custom-field system, mySkua keeps four dimensions, each captured by a different mechanism:
Two are declared by the subscriber, two are observed by the system. Declared attributes are accurate but cost a tap; observed ones are free but describe where someone came from rather than what they want. A funnel that uses only one kind is either annoying at the door or blind afterwards.
All four live on the subscriber record inside a single bot. Run several bots from one account and their subscribers, languages, categories and content stay isolated from each other.
Telegram exposes the client language on the first message, so the bot has a good guess before it says anything. mySkua uses that guess to render the language prompt, the subscriber confirms or corrects it, and the answer becomes a stored property rather than a one-off rendering decision.
Everything after that point reads the stored value: the channel-subscription gate, the welcome message, campaign copy, drip steps and broadcasts. Language codes are arbitrary and configured per bot, so a bot can run two markets or ten without a code change — adding one is a content task. The mechanics are covered in the guide on running a multilingual Telegram bot.
The payoff of storing rather than inferring: a drip step written only in Hungarian can be aimed at Hungarian speakers instead of going to everybody and being ignored by most of them.
Every deep link into a Telegram bot can carry a payload — t.me/yourbot?start=PAYLOAD. mySkua records it as the subscriber's source, so the channel post, the affiliate placement, the flyer QR code and the paid ad each produce a distinguishable cohort even though they all land in the same bot.
Source is the dimension that settles acquisition arguments. Two placements can produce the same number of opt-ins and wildly different activation rates; without a source attribute you see one blended number and keep paying for both. With it, growth, retention cohorts and campaign activations can be read per source.
Give every placement its own payload, even when the destination content is identical. Payloads are free, and a source you did not separate at the start is one you cannot separate later.
Language, source and category are each defined by a specific moment in the funnel. Tags are the open dimension — the place to record anything else you know that onboarding does not ask about. They can be applied automatically as subscribers arrive, which is what makes them useful at volume rather than only for hand-curated lists.
The typical use is rolling several related payloads up under one label — a set of affiliate links as one partner, a month of placements as one campaign wave — so you can target the group without listing every source string. Keep the vocabulary small: a bot with forty near-duplicate tags is harder to target than one with six deliberate ones.
A category is an interest the subscriber selects for themselves — Sport or Casino, say — at an optional step in onboarding that sits between the language pick and the channel gate. It is the only dimension that tells you what someone wants rather than where they came from or what device language they use.
Because the choice is explicit it carries more weight than an inference. Someone who tapped Sport has told you which half of your content library is relevant, and sending them the other half is a self-inflicted wound. Category targets content and drips, and also changes what the subscriber sees immediately. The step is optional per bot — if your offer is single-purpose, leave it off rather than charging a tap for a dimension you will never target.
Attributes are only worth collecting if the sending tools can read them. Three send paths use them, in different ways:
Delivery is bounded on every path: sends retry a limited number of times and subscribers who blocked the bot are dropped rather than retried forever, so a stale segment never becomes a queue of doomed messages.
The sharpest use of segmentation happens before any drip runs. Each category can carry its own welcome offer, so the first message a subscriber sees after choosing an interest is written for that interest — in their confirmed language, with its own media and inline buttons.
A generic welcome spends the highest-attention moment in the funnel explaining who you are. A category-specific one spends it on the thing the subscriber told you they care about, about a second after they told you.
Every onboarding message is editable per language under Bot Texts, media and inline buttons included, so the category-plus-language matrix is content you maintain rather than code someone has to ship.
Can I segment subscribers I have already collected? Only by the attributes captured when they joined. Source is recorded from the deep link on arrival and language from the onboarding confirmation, so people who joined through an untagged link have no source to filter on. Tags can be applied going forward — which is why it pays to decide a payload scheme before a campaign rather than after it.
Which sends can be targeted at a segment? Drips take language, source, tag or category directly. Campaigns are targeted by the link you distribute and rendered in the subscriber's language. Broadcasts are chosen by hand, with a recipient count shown before the send.
What is the difference between a tag and a category? A category is chosen by the subscriber and changes what they are shown; a tag is assigned by you and is invisible to them. Use categories for interests you want to serve different content to, and tags for operational grouping.
Does segmentation carry across bots? No. Each bot has its own subscribers, languages, categories and content, isolated from the others under the same account.
mySkua is in a private, invite-only beta — one account manages several bots, each with its own subscribers, languages, categories, campaigns, drips and team roles. If you run Telegram funnels and want to stop sending one message to four audiences, join the waitlist from the homepage and request access, and we'll be in touch as more places open up.
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