In short
An AI chatbot is conversational software in a chat window on your website, app or messaging channel. Modern chatbots use a language model to understand questions and write replies, usually drawing on a knowledge base you supply. The useful ones also connect to your CRM or booking system, so a conversation ends with a record or an appointment.
Signs a chatbot will help
| Repeated questions | Your team answers the same written questions every day: prices from, service areas, opening hours, what is included. |
|---|---|
| Out-of-hours visitors | Website visitors arrive in the evening and at weekends when nobody can reply. |
| Answers that already exist | The information is written down somewhere and agreed, so the bot can be given it. |
| A clear next step | Conversations can end in something useful: a booking, a captured enquiry or a link to the right page. |
| An owner | Someone will update the knowledge base when prices, services or policies change. |
Signs it will not
Most enquiries come by phone. If your customers ring, a chatbot fixes the wrong channel. Look at where enquiries are being lost before choosing the tool. For many trades and local services, a voice agent on the phone line comes first.
Every answer depends on a site visit or a quote. A bot that can only say “someone will be in touch” adds a step without adding value. In that case, a short enquiry form that captures the right details may do the job better.
Nobody will look after it. A chatbot with last year’s prices does more harm than no chatbot at all.
Your business is responsible for what it says
Customers treat a chatbot on your site as speaking for you. In a 2024 decision, Moffatt v. Air Canada, a Canadian tribunal (the British Columbia Civil Resolution Tribunal) held the airline responsible for incorrect information its website chatbot gave a customer about bereavement fares, and rejected the argument that the chatbot was responsible for its own statements. That was a Canadian case, but the principle is a sensible one to plan around anywhere.
So limit the bot to approved information, tell it what to do when it does not know, and keep it away from anything that sounds like a binding promise: prices for non-standard work, refunds, legal or medical guidance. Tell users they are talking to an automated assistant, and make it easy to reach a person.
Setting one up well
Start with your real questions. Pull the last few months of emails, form submissions and messages, group them, and write approved answers to the most common ones. That knowledge base matters more than the choice of chatbot software.
Connect it to where the work happens. Captured details should create a CRM record with the conversation attached. Booking requests should use real availability. A chatbot that emails a transcript to a shared inbox has only moved the problem.
Keep the scope small at launch. One page, one channel and the ten most common questions is enough to learn from. Widen it once the answers are reliable.
Review conversations weekly at first. Look for questions it could not answer, answers that were wrong, and conversations where the visitor gave up. Fix the knowledge base, then the instructions.
When this matters
Relevant if written enquiries are slipping through or your team repeats the same answers all day. New Era AI’s AI agents and CRM automation services cover chat set-up with CRM write-back, and an audit will tell you whether chat or voice is the bigger gap.
Questions
Will a chatbot put customers off?
A poor one can. Customers are put off by bots that cannot answer, hide the human option or repeat themselves. A bot with good answers and an easy route to a person is usually welcomed.
Chatbot or AI voice agent first?
Look at where enquiries are being lost. If calls go unanswered, start with voice. If website visitors leave without contacting you, start with chat.
Can a chatbot take bookings?
Yes, when it is connected to a calendar or booking system and given clear rules about what it may book.