Automating a business with a Telegram bot: where to start
Companies often try to automate everything at once and end up with an unmanageable project. The right start is one process taken all the way through, not ten left half-finished.
Choosing the first process
Look for a process that repeats often, has clear steps and does not need a creative decision at every turn. Taking orders, booking appointments, answering routine questions are good candidates. Negotiating complex contracts is not.
Estimate how much time the process currently takes per week. Under a couple of hours and automation pays back slowly; a daily routine of several hours shows results within the first month.
Do not over-engineer the first version
The first version should cover the customer's main path, not every possible exception. Rare cases — under 5% of requests — can simply hand off to a human at launch instead of being scripted.
Expand the bot iteratively based on real data: after a month you can see which questions it failed to answer, and those become the next scenarios to build.
Who should be involved
The employee currently running the process by hand must take part in describing the scenario — they know the exceptions and nuances no job description captures.
Assign someone inside the company to own the bot after launch: someone needs to watch for complaints, update answers and decide when the script needs changing.
Measuring the result
Compare concrete metrics before and after: request handling time, the number of dropped inquiries, staff hours spent. A vague sense that 'it feels smoother' does not substitute for numbers when deciding on further investment.
Give the bot at least a month before drawing conclusions: the first weeks go into customers adjusting and the script being tuned against questions nobody anticipated.