AI chatbot for small business: is it worth it?

An AI chatbot on your website is worth it when three things are true: your site already gets steady traffic, visitors ask the same questions over and over, and leads currently wait hours for an answer. In that case a bot that answers in seconds and books the next step usually costs less per month than one extra customer is worth. If your site gets a few dozen visits a week, or every sale needs a long custom conversation, spend the money on traffic or on faster human follow-up first.

Here is how to decide with numbers rather than hope.

What an AI chatbot actually does now

Older chatbots were decision trees: click a button, get a canned answer. Current AI chatbots read your own content (service pages, FAQ, policies, price rules) and answer free-form questions in plain language. The useful ones also act:

• collect a name, email and phone number when the visitor is ready;
• ask qualifying questions and score the lead;
• book a call or appointment in your calendar;
• push the conversation into your CRM so someone follows up;
• hand off to a human during business hours.

A bot that only answers questions is a nicer FAQ page. A bot that captures and routes leads is a sales tool. Judge vendors on the second part.

What it costs

Prices from vendor pages in October 2026 (US dollars unless noted):

Three things push the real cost up:

1. Volume pricing. Per-conversation or per-resolution billing is cheap at 50 chats a month and noticeable at 2,000. Estimate your volume from current contact-form and live-chat numbers.
2. Integrations. Calendar booking, CRM sync and SMS often require a higher tier or a separate tool.
3. Set-up and upkeep. Writing the knowledge base, setting rules and testing takes a few days at the start, then an hour or two a month to review conversations and update answers.

For most small service businesses, the all-in software cost lands somewhere between US$40 and US$200 a month. A one-time set-up by an agency comes on top of that.

The break-even math

Use your own numbers. Here is the formula with an example.

• Monthly website visitors: 1,500
• Share who open the chat: 3% = 45 conversations
• Share of those who leave contact details: 30% = about 13 leads
• Share of those leads you'd have lost without the bot (they would have left without filling the form): be conservative, say one third = about 4 extra leads
• Your close rate on leads: 25% = 1 extra customer a month
• Value of a customer: $2,000

One extra customer a month at $2,000 against $100 to $200 in software is an easy yes. Now run it for a site with 200 visitors a month: you get about one extra lead every two months, and the bot is a nice-to-have at best.

The uncertain number is "leads you'd otherwise have lost". Measure it instead of guessing: compare the three months before launch with the three months after, counting form fills plus chat leads, and look at when chat leads arrive. If many come in evenings and weekends, the bot is catching people your form was losing.

Why speed matters so much

A well-known study published in Harvard Business Review in 2011 audited 2,241 US companies and found that firms which contacted a web lead within an hour were nearly seven times as likely to qualify it as firms that waited even one hour longer, and over 60 times as likely as firms that waited a day or more. The average response time among firms that did respond was 42 hours. The data is old, but the pattern still matches what businesses see: the first credible answer tends to win the job. A chatbot is one way to answer at 10 p.m.

What to train it on

The bot is only as good as what you feed it. Prepare these before you pick a tool:

1. Services and non-services. What you do, what you don't, and where you work. "We don't do X" prevents a lot of bad leads.
2. Pricing rules, not price lists. How quotes are built, starting prices if you publish them, and the exact sentence to use when a price depends on a visit.
3. Your top 30 questions with approved answers, pulled from emails, calls and reviews.
4. Policies. Cancellation, warranty, payment terms, booking lead times.
5. Qualification questions. Three to five questions that tell a good lead from a poor fit.
6. Hand-off rules. When to stop and pass to a person: complaints, legal or safety questions, anything outside the knowledge base.
7. Tone. Formal or friendly, "vous" or "tu" in French, how to greet.

Then restrict it: tell the bot to answer only from your material and to say "I'll have someone confirm" when unsure. Do not let it browse the open web on your behalf.

The risk you take on

Your chatbot speaks for your business. In Moffatt v. Air Canada (2024), a BC tribunal found the airline liable after its website chatbot told a customer he could claim a bereavement fare retroactively, which contradicted the airline's actual policy. The tribunal rejected the idea that the chatbot was a separate entity responsible for its own actions and ordered compensation of about $812. The amount was small; the principle is not.

Practical protections:

• keep prices, guarantees and legal terms out of free-form answers unless they are word-for-word from your approved text;
• review conversations weekly in the first month, then monthly;
• update the knowledge base the same day a service, price or policy changes;
• tell visitors they are talking to an automated assistant;
• check where the vendor stores conversations and for how long, since chats contain personal information.

When it's not worth it

• Low traffic. Under a few hundred visitors a month, fix traffic first.
• The website doesn't convert anyway. A chatbot won't save a slow, confusing site with no clear call to action.
• Nobody follows up. If chat leads land in an inbox nobody checks, you've paid to collect names.
• Highly custom sales. For complex B2B deals, a short form and a fast human reply may do better than a bot.
• You can't keep the content current. Outdated answers cost trust, and sometimes money.

Before and after: what to measure

Track these for three months before and three months after launch:

• total leads per month (forms + calls + chats);
• share of leads arriving outside business hours;
• median time from first contact to first human reply;
• booked calls or appointments;
• close rate on chat leads versus form leads.

If total leads and bookings don't move after three months of tuning, switch it off. A good bot proves itself in the numbers.

Where PNE fits

Our AI growth systems install the chat assistant together with the CRM pipeline and the follow-up sequences behind it, so a conversation turns into a booked call rather than a transcript. For Kaltico, a construction and consulting firm, we set up an AI assistant that answers first enquiries and FAQs and books discovery calls around the clock, connected to automated CRM follow-up. The full project is in the Kaltico case study.

Sources

• Pricing — Tidio
• Pricing — Chatbase
• Pricing — Intercom
• HighLevel AI Products Pricing — HighLevel Support Portal
• The Short Life of Online Sales Leads — Harvard Business Review
• Are you letting online leads go cold? — SmartCompany
• Airline ordered to compensate a BC man because its chatbot provided inaccurate information — Mondaq / Dentons