How a bakery, gym, barbershop or corner restaurant can turn walk-ins into a list they own — and send offers that actually reach people.
Most “bot for business” pitches start with conversation: an assistant that answers questions, takes bookings, chats about the menu. For a shop with one or two people behind the counter, that is almost never the bottleneck. The bottleneck is reach. You have regulars who like you and would come back more often if they were reminded, and you have no reliable way to reach them that doesn't cost money per impression.
A Telegram bot solves that narrow problem well. Someone taps a link, presses Start, and you now have a direct channel to them that no feed algorithm sits in front of. What a small business needs from that bot is short: a way to collect the opt-in, a way to send an offer on a schedule, a way to follow up automatically with people who just joined, and a way to see whether any of it moved actual footfall.
In mySkua the whole join experience is one configured flow: the subscriber sends /start, picks (or confirms) a language, optionally picks a category of interest, optionally passes a channel-subscription gate if you also run a Telegram channel, and lands on your welcome offer. No scripting, no conversation tree to maintain.
Every subscriber arrives through a deep link of the shape t.me/yourbot?start=CODE. The useful part is that the payload after start= is yours to define, and it travels with the subscriber: the person is tagged with the source they came from, permanently. So you do not print one QR code — you print a different one per placement.
?start=receipt. The highest-intent moment you have: they just paid and liked it.?start=counter, with the offer written on the card, not just “follow us”.?start=window, which quietly measures passers-by who never came in.?start=delivery, aimed at people who order but have never visited.?start=ig_bio and a dated one like ?start=ig_2026_03 per campaign burst.Payloads do not have to be real campaign codes. A plain string like ig_2026_03 is resolved as a source tag and simply attaches to the subscriber; a payload that matches a campaign code additionally triggers that campaign's message and promo. That means you can start printing codes before you have finished designing the offers behind them. The mechanics of payload design, offer sequencing and what to put on the card are covered in more depth in our guide to building a Telegram lead-generation bot.
One practical note about the offline side: the QR needs a reason next to it. “Scan for our Telegram” collects nobody. “Scan — free coffee with your next sandwich, this month” collects a list.
The welcome message is the only message you can be certain every subscriber reads, because it arrives one second after they chose to press Start. Treat it as a contract rather than a greeting. Three things belong in it: what this list is, how often you will write, and what they get right now.
Something like: “You're on the list. One message a week, Thursday, with what's on that weekend. Here's your code for a free filter coffee — show it at the counter, valid until the 30th.”Stating the frequency up front is the single cheapest thing you can do to reduce people blocking the bot in week three. It also makes the relationship an explicit opt-in rather than an assumed one.
Mechanically, the welcome text lives in Bot Texts and is editable per language, with a photo or video attached and inline URL buttons underneath (menu, directions, order page). The offer itself is better built as a campaign: a campaign carries a promo code, its own per-language message, media, buttons and an active window, and it activates once per user. That one-time activation is what stops the same person redeeming the launch offer every week, and it is also what makes the activation count meaningful later.
After the welcome, the recurring habit is one broadcast. Not three, not whenever you remember — one, in a fixed slot, so subscribers learn to expect it. A Thursday late-morning send for a weekend offer works for most food and retail businesses; a Monday send works better for gyms and services.
Scheduled broadcasts run daily, weekly or monthly with a timezone attached, so the send fires at 11:00 your local time regardless of where the server sits. One-off broadcasts show a live recipient count before you send, which is a useful sanity check when you have language or category filters applied and expect 400 people but the counter says 40. The broadcasts feature page goes through targeting and delivery stats in detail.
Keep the content of the weekly send concrete: a specific item, a specific price, a specific window. “Weekend special” with no detail performs like no message at all, and costs you the same attention.
Auto-messages are drips that fire a set delay after someone joins — minutes, hours or days — and can be targeted by language, source, tag or category. They are the part that keeps working while you are busy serving people, and for a small business three of them are usually enough:
Sends that fail are retried a bounded number of times, and subscribers who block the bot are dropped from the list automatically. That matters more than it sounds: it means your recipient counts stay honest, and you are not comparing this month's results against a subscriber number that quietly rotted.
A lot of local businesses sit in genuinely bilingual streets — a Portuguese bakery with English-speaking office workers at lunch, a Hungarian gym in a district full of newcomers. Telegram tells the bot what language the person's app is set to; the bot proposes that language, the subscriber confirms it, and every later message — welcome, campaign, drip, broadcast — is served in that language automatically.
Language codes are configured per bot, so you add exactly the ones you serve. The honest constraint is editorial, not technical: every campaign, auto-message and broadcast holds a message per language, and a machine-shaped translation in the second language reads worse than not offering it at all. If you cannot keep both columns properly written, run one language and do it well.
Categories work alongside this. If you run a gym with a clear split — classes versus weights, or a shop with a coffee crowd and a lunch crowd — let subscribers self-select at join time, give each category its own welcome offer, and target drips and broadcasts at one category instead of the whole list.
You do not need a dashboard habit. You need three numbers, checked once a week:
Give each placement its own campaign code when the offers differ, and the comparison between them becomes direct. Beyond that, retention cohorts and daily active numbers tell you whether the list is a real audience or a pile of one-time coupon hunters — the cohort that still opens messages in month three is the one worth writing for.
The realistic time budget for a small business is about twenty minutes a week: write the weekly offer, glance at joins by source, done. Everything else should be set once. Drips run on their own. Scheduled broadcasts fire on their own. Bot Texts are edited in the panel rather than in code, so changing your welcome message is a text edit, not a deployment.
Two things help when you are not the only person involved. Team roles let you add admins with owner, admin, editor or viewer permissions — give the person who writes the weekly offer an editor seat and keep the bot token and settings to yourself. And one account can hold several bots with completely separate subscribers and content, which is the right structure for two locations, or for a shop and a side venture that should not share a list.
The failure mode to avoid is silence. A list you built in March and did not write to until September is worth very little; people forget who you are and block the bot. One predictable message a week beats a burst of five followed by nothing.
Do I need a Telegram channel as well as a bot? No. The channel-subscription gate is optional — turn it on only if you already run a channel and want bot subscribers to join it. For most local businesses the bot alone is simpler, and it lets you send to people individually rather than posting to a feed.
What if my customers are not on Telegram? Then this is the wrong channel and no amount of tooling fixes it. Check first: look at whether the app appears on the phones people hand you when they show a code. In cities where Telegram is normal, the QR converts; where it is not, it does not.
mySkua is currently in a private, invite-only beta, so accounts are opened by request rather than self-serve. If you run a local business and want to try this — especially if you already have somewhere to put a QR code — join the waitlist and tell us what you sell and where. We prioritise accounts where the use case is concrete.
Related: Telegram lead-generation bot guide · Broadcasts · Opt-in.
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