A Telegram bot for booking guests to a dinner replaces the administrator's private messages: the guest points a camera at the QR code on the table, leaves a name and phone number, and the request goes straight to the manager. At the Moscow restaurant SIMACH, signing up takes 35 seconds, and one and the same bot filled two wine dinners.
Key takeaways
SIMACH is a restaurant in central Moscow that holds wine dinners in the evenings: wines from a single region, a chef's set menu and live music. The task was simple: stop gathering guests for these evenings in private messages.
Guests wrote to the administrator directly, and their contacts stayed buried in chats. Some requests got lost. After a dinner the restaurant didn't know which of the people who signed up actually came or what the guests liked, so every new evening had to be gathered from scratch.
Two things were needed. First, to collect guests in one database where the name, phone number and consent to data processing sit in one place rather than in chats. Second, to sell the next dinner to that same database: the invitation should reach those who have already come or shown interest, without manual mailing.
A restaurant loses not only individual requests but also the memory of its guests: the chat closes, and there is no list of people to write to about the next evening. The table shows the difference between the two schemes.
| What happens | Requests in the administrator's chats | The SIMACH Telegram bot |
|---|---|---|
| Name, phone number and consent to data processing | Stay in the administrator's chats | Sit in one database |
| How a request arrives | The guest writes a private message, some requests get lost | The guest leaves contacts on their own, the manager gets the request at once |
| Invitation to a new dinner | Guests are gathered from scratch every evening | The bot sends the invitation across the database, each group gets its own text |
| Repeat sign-up | Another chat with the administrator | One button, no form to fill in again |
| What is known after the dinner | Unclear who came and what they liked | A rating from 1 to 5 and the guest's comment |
Something else stands out: in the private-message scheme all the work falls on the administrator — they reply by hand and make sure contacts don't get lost in chats. With a bot the guest walks the path alone, and a person joins where needed: the manager receives a request or the phone number of an unhappy guest.
A general overview of how such bots work is in the piece on a Telegram bot for booking clients. Below is how it all came together at SIMACH specifically.
Signing up starts with a QR code: it stands on the tables and on the evening poster, so the guest doesn't need to search for the restaurant or write to the administrator.
All of this takes 35 seconds. For the guest it is one short dialogue instead of a chat with a wait for the reply, and for the restaurant it is a ready row in the database with the consent sitting next to the contact.
Tip The order of steps matters more than it seems: consent first, then name and phone. That way the database never gets contacts you can't use, and you don't have to find out later whether you may write to these people.
The same bot sold the second dinner: before the new evening it sent invitations across the database, with its own text for each group of guests.
Signing up takes one button: guests who have already left their contacts don't fill in the form again. The first sign-up came in 10 minutes after the mailing.
The point of splitting into groups is simple: to talk to each guest about their own situation. Someone who has already been doesn't need the format of the evening explained. Someone who didn't finish signing up needs an easy way back. One general mailing to everyone doesn't give that.
After the dinner the bot collects impressions on its own: the guest rates the evening from 1 to 5 and writes a few words.
If the rating is 3 or lower, the manager instantly gets the guest's phone number. They can call while the evening is still fresh, and the conversation is about a specific dinner rather than a vague impression from a week ago.
Example A guest review in the bot, 5 out of 5: "Loved everything, nothing to improve — the main thing is to keep this level". Before, silence followed a dinner for the restaurant.
The rating does two jobs. For the manager it is a signal to call an unhappy guest the same evening. For the restaurant it is an accumulating picture of what guests like, which helps prepare the next dinner.
The main result: the restaurant now has its own guest database and a channel that sells every next dinner.
If you reduce the changes to four lines, you get this:
The last point matters most. Before, after a dinner the restaurant didn't know who came or what they liked; now it has guests' answers and a list of people it can write to. That is why the second dinner was gathered not from scratch but from the database the first one built up.
Every next dinner now launches in one evening: a new poster, new texts — and the bot invites guests from the database again.
Client review "They built a great Telegram bot and made changes quickly. We hope for long-term, productive work" — Ilgam, manager of the SIMACH restaurant.
A booking bot breaks not on technology but on decisions about the scenario. Five mistakes that are easy to foresee before launch, drawn from the logic of this case.
One principle covers all five points: the bot should lead the guest along a short path and after each step leave a trace in the database that can be put to use.
The effect of a booking bot is measured by five indicators, and each of them can be counted in your venue before launch and after.
| Indicator | What it shows | Value at SIMACH |
|---|---|---|
| Time to sign a guest up | How much effort a guest spends on a request | 35 seconds |
| Time from mailing to first sign-up | How well the database responds to an invitation | 10 minutes |
| Database growth between events | How many new contacts the bot accumulates | 1.5 times |
| Number of events on one bot | Whether the setup pays off through repeat launches | 2 dinners |
| Ratings after the evening | Where quality slips and whom to call | a rating from 1 to 5, a call at 3 or lower |
Before launch, measure at least the first indicator: how much time and how many messages one private-chat sign-up takes you. Other restaurants' figures won't tell you much, while your own starting point will.
The scheme fits any format with booking and limited seats: tastings, masterclasses, private evenings. You will have to change the texts and the poster; the logic of QR, consent, name and phone stays the same.
If you want to know where a guest came from, say from the table or from the poster, a link solves it. Telegram opens a bot at an address like t.me/bot_name, and a parameter of up to 64 characters can be added to it. How such a tag works is covered in the article on deep links and the user's source.
A bot for taking enquiries and bookings costs from $1,200 on the service page. Segmented mailings and post-dinner rating collection are added to the estimate, and we give the exact price after reviewing the task.
We start with that review: where your guests come from, at which step a request gets lost and what you want to know about guests after the evening. Then we build the path from QR to sign-up, requests go to the manager, and contacts and consents land in one database. How to connect such a database to a CRM is covered in the article on integrating a Telegram bot with a CRM.
Let’s discuss your project
Tell us what you need, and we will estimate the timeline and cost and suggest a solution.
A bot for taking enquiries and bookings starts from $1,200. Segmented mailings and post-dinner rating collection are added to the estimate, and the exact price is given after reviewing the task.
The same. The QR, consent, name and phone logic fits any format with booking and limited seats: tastings, masterclasses and private evenings.
The first step is the bot taking consent to process personal data, and only then does it ask for a name and phone number. The name, phone number and consent are stored in one database, not in the administrator's chats.
After the dinner the bot asks for a rating from 1 to 5. For a rating of 3 or below the manager receives the guest's phone number and can call back while the evening is still fresh.
With one button: there is no form to fill in again. At the second SIMACH dinner, the first sign-up came in 10 minutes after the mailing.
A new poster and new invitation texts. After that the bot invites guests from the database again, which is why at SIMACH every next dinner launches in a single evening.
