← All Articles
AI & Automation15 November 20264 min read

Should You Put An AI Chatbot On Your Website

Only if it answers from facts you approved and hands real buyers to a person fast. Everything a chatbot types is your business talking, and you wear any promise it gets wrong.

Only add an AI chatbot to your website if it answers from facts you have approved and hands real buyers to a person quickly. Everything it types is your business talking, so a bot that guesses a price or a policy is a liability with a friendly face.

For plenty of service businesses, a fast reply from a real person does more than any bot. The chatbot earns its place when the same questions arrive at all hours and nobody is around to answer them.

What a chatbot does well

Repeat questions at 10pm, like your hours, the suburbs you cover, what a service includes and how booking works.

Contact details such as name, phone, job type and suburb, collected while the buyer is still on the page.

Bookings for a call or a site visit, straight into a calendar.

Routing, so a warranty question goes to service and a new project goes to sales.

Speed is why any of this matters. A buyer who gets an answer in a minute keeps talking, and the case for replying fast is in how fast should you reply to a new enquiry.

Where it goes wrong

In February 2024, a Canadian tribunal ordered Air Canada to compensate a customer after the chatbot on its website gave him wrong information about bereavement fares. The airline argued that the chatbot was responsible for its own words, and the tribunal rejected that argument.

Closer to home, the Australian Consumer Law bans misleading or deceptive conduct in trade, and a statement on your website is your statement whether a person or a piece of software typed it. If your bot promises a price, a warranty or a turnaround you do not offer, your business made that promise.

AI models also make things up with total confidence. Ask a general chatbot about your own business, and it will happily describe a service you have never offered.

Feed it your own facts

A website chatbot should answer only from documents you wrote and approved: your services, the areas you cover, the prices you publish, your policies and your common questions. Set it to offer a call back whenever the answer is not in those documents, instead of guessing.

Your inbox is the best source for those questions. Go through the last 100 enquiries, list every question and write your real answer to each one. More on mining your inbox is in why your best content is already in your inbox.

Then test it before it goes live. Ask it 50 real questions, including awkward ones about price, refunds and complaints, and read every single answer.

Hand over to a person

The bot's job ends the moment a human takes over. Someone weighing up a $50,000 job wants to talk to someone real, so the bot should collect their details, book the call and tell them exactly when to expect it.

During business hours, a real person should pick up the conversation within minutes. After hours, the bot can promise a call at a set time the next morning, and someone on your team has to keep that promise.

Privacy

Chat logs hold names, phone numbers, addresses and sometimes details people would rather keep private. Tell visitors they are talking to a bot, keep logs only as long as you need them and find out where your chatbot provider stores the data.

The OAIC, Australia's privacy regulator, has published guidance on using commercially available AI products with personal information. Read it before you switch anything on.

How to decide

Start with your own numbers. If you get a handful of enquiries a week and someone answers each one within minutes, a chatbot adds very little. If dozens arrive after hours, or the same ten questions eat hours of someone's week, it earns its keep. Repeat work like that is where AI earns its place, and the rules we hold it to at Ignis are in how we use AI at Ignis.

What to measure

Enquiries captured after hours, compared with the same months last year.

Time from bot to human, in business hours and overnight.

Wrong answers, found by reading a sample of chats every week and fixing the documents behind each one.

Get those three right and the chatbot works like a night shift that sticks to your facts. Your team then spends the day on the buyers who are ready to talk.

David Eid

David Eid

Marketing Strategist · Founder of Ignis

Marketing strategist based in Sydney, Australia. Founder of Ignis - premium marketing that scales businesses. Our average client generates $3M+/year and 1M+ views/month.

AIWebsiteLead Response
Share

Get weekly marketing insights

No fluff. No spam. Strategies from someone who has done it. Delivered to your inbox every week.

Subscribe