Moving a Telegram bot without losing its users

A bot can only message people who have started it, and that permission is tied to the bot’s identity — its token — not to the software behind it. If you created the bot in your own Telegram account and handed the token to a builder platform, you can take the same token to a new backend and the username, the users and their permission to receive messages all come with it.

If the platform created the bot for you and never showed you the token, the first step is getting it. Without the token, migration means a new bot and asking every user to start it again.

Telegram does not give a bot a list of its own users. The only record of who has started your bot is in the database of whatever has been running it. Before switching anything, export the chat ids, any tags or segments, and whatever state the flows depend on — a booking reference, a loyalty balance, a language preference. Once the webhook moves, the old platform stops receiving anything new.

A bot has exactly one webhook address at a time. Migration is the moment that address changes from the old platform to the new backend, so the new flows are built, tested against a copy of the exported data, and deployed before the switch. The switch itself takes seconds; a quiet hour is still the right time to make it.

After the cutover the old platform can be kept read-only for reference, then closed. Anything it charged for per contact stops the day it closes.

An abandoned custom build is a different problem: the code exists, but nobody currently understands it. That starts with a paid one-day audit of the code and the accounts, after which it is either stabilised under a support retainer or rebuilt at the catalogue price for its bot type. The deciding factor is almost always whether you control the token and the hosting, not how good the code is.