A bot for enquiries: when a chatbot pays off and when it stays a toy

A bot that takes enquiries is cheap and quick to build, which is why it is usually the first thing companies order. Sometimes it pays for itself in a month, sometimes it sits idle. What that depends on, and what has to exist around it.

A bot for taking enquiries is the most common first purchase in automation. It is inexpensive, takes about a week to build, and it is visible. The results, though, vary widely: for some it takes half the routine off the team, for others it lives a month and quietly dies.

The difference is not the quality of the bot. It is what happens around it.

When a bot for enquiries pays off

There are three conditions. Meet all of them and it will almost certainly pay for itself; miss one and the effect is noticeably weaker.

There are many enquiries and they are similar. A bot is strong on repeated questions: opening hours, price, availability, booking. If every enquiry is unique, there is nothing to save.
Customers already write to you in a messenger. A bot does not create a channel, it serves an existing one. Putting it where nobody writes gives you an empty chat.
Somebody follows up on the enquiry. The bot takes it and hands it on; if it then lands in a group chat and gets lost, automation has only sped up the loss.

What a bot does well

Replies instantly, around the clock. For a customer that often matters more than a perfectly precise answer two hours later.
Asks the same qualifying questions every time and forgets nothing: what is needed, when, where, how to get in touch.
Notifies the responsible person directly rather than a group chat where the message sinks.
Collects every enquiry in one place where they can be counted. Usually it is after launching a bot that a company first learns how many enquiries it was losing.

What not to expect from it

It will not sell a complex service. Its job is to hand a prepared person to a human, not to close the deal.
It will not handle rare questions. A person still answers those, and that is fine.
It will not increase the flow of enquiries. A bot works with what already arrives; new traffic is the job of advertising and the website.

A common mistake: giving the bot too long a script. People do not sit through six questions in a row — they drop out at the third. Two are usually enough: what you need and how to reach you. A person can find out the rest.

How to tell whether it is working

Counting conversations tells you almost nothing. Look at these instead.

How many enquiries reached a person with complete details rather than a bare "hello".
Time from enquiry to first reply. This is where the bot gives the most obvious gain.
Share of conversations abandoned mid-script. Rising means there are too many questions.
How many enquiries arrive outside working hours. Those are precisely the ones you used to lose.

How to check before you order one

Three questions that honestly show whether a bot is what you need right now.

How many enquiries a week, and what share of them are repeated questions? If repeats are under half, a bot is not where to start.
Who exactly will receive the enquiries and within what time will they answer? Without an owner, a bot becomes a tidy list of lost customers.
Where will enquiries be stored? If the answer is "in the chat history", you need a place to record them first and a bot second.

A bot is a good first step into automation precisely because it is cheap and quickly shows you what is really happening with your enquiries. But what pays for it is not the bot — it is the order that appears around it.