New Zealand has about 550,000 small businesses. Most of them answer their own phone, miss a significant number of calls, and have no idea how much revenue those missed calls cost them.
That gap — between what AI can do and what small NZ businesses are actually using — is a genuine business opportunity right now.
Here’s what it looks like in practice, based on building one.
What You’re Actually Selling
The product isn’t “AI.” Small business owners don’t buy AI. They buy outcomes.
The outcome here is: every inquiry call gets answered, and the business owner gets a text message with the details within 30 seconds.
A plumber who’s under a house can’t answer his phone. An accountant deep in a client file can’t pick up. A café owner in the middle of service can’t stop. The AI handles the call; they get a text.
That’s the product. Not the technology underneath it.
The Market in NZ
The businesses most likely to pay for this are:
Trades (plumbers, electricians, builders, cleaners, painters): high call volume, frequently unavailable to answer, high job values make missed calls expensive. About 20,000 small trades businesses in NZ.
Property managers: dealing with tenant and landlord inquiries all day, often in inspections when calls come in. About 800–1,200 PM companies in NZ.
Hospitality: cafés and restaurants where the phone rings constantly during service and nobody can answer it. About 15,000 businesses.
The largest of these segments — trades — is also the most underserved by software. Tradies are late adopters, but once they’re using something that works, they rarely switch.
The Economics
Here’s the rough math:
Infrastructure cost (per client, 500–1,000 min/month):
- AI voice API: ~$20–35/month
- Phone number: $2–5/month
- Telephony (call handling): ~$15–20/month
- Server, overhead: ~$5/month
- Total: roughly $45–60/month
Pricing to clients: $149/month (unlimited calls, everything included)
Gross margin: 60–65% at $149/month
At 10 clients, that’s $1,490 MRR with ~$590/month in costs = $900/month gross profit. At 20 clients: $2,980 MRR, ~$1,180 costs = $1,800/month gross. At 50 clients: $7,450 MRR, ~$2,950 costs = $4,500/month gross.
The margins improve as you scale because some costs (server, phone number provisioning, management time) are partially fixed.
How to Get the First 5 Clients
The mistake most people make is building the product first, then trying to sell it. The right order is reversed: verify the willingness to pay before optimising the product.
Step 1: Use it in your own business
If you have any service business — even a small one — run the AI receptionist yourself for 2–3 weeks. Track: how many calls came in, how many would have been missed, how many became jobs.
Those numbers become your case study. “In 3 weeks, my AI answered 47 calls. I would have missed 12 of them. Those 12 calls were worth an estimated $3,800 in jobs booked.”
That case study is your sales tool.
Step 2: Sell to people who know you
Tradies talk to tradies. Plumbers in the same area know each other. Builders go to the same suppliers. You don’t need to cold call strangers — you need one person in your network to try it for free for two weeks, get results, and tell three other people.
The first conversation isn’t: “Would you like to buy my AI service?” It’s: “I’ve been using this for my [business], and it’s caught $X in jobs I would have missed. Want to try it on your phones for two weeks at no cost?”
Nobody says no to two weeks free.
Step 3: Convert trial to paid
At the end of two weeks, share the summary with them: how many calls came in, how many were after hours, how many turned into jobs. Most businesses that see their own missed call data are motivated to continue.
Pricing conversation: “$149/month. No contract. If it’s not working in 60 days, cancel and I’ll refund your first month.”
What Setup Actually Takes
Setting up the AI receptionist for a new client takes 1–3 hours:
- Information gathering (30 min): What services do they offer, what are typical call types, what are their prices, what’s their availability, what should the AI say and not say.
- Configuration (60–90 min): Building the conversation flow, testing with different scenarios.
- Testing and handoff (30 min): Call the number together, confirm the AI handles correctly, show them how notifications work.
That’s the setup. Ongoing maintenance is minimal — updating info if their services change, reviewing call logs occasionally.
Common Objections and How to Handle Them
“Customers will hate talking to an AI.” The majority of callers don’t complain — they care more that someone answered than who answered. Show them the alternative: voicemail, where the call-back rate for NZ customers is low.
“What about complex questions?” The AI handles simple queries (availability, pricing, booking, general information). Complex or urgent situations trigger an immediate notification to the owner, who can call back within minutes. Most inquiry calls aren’t complex.
“What if the AI makes a mistake?” The AI doesn’t book jobs or commit to prices — it collects information and notifies the owner. The owner makes every actual commitment. The AI can’t overpromise.
"$149 is a lot." The framing matters: what’s the average job worth for their business? For a plumber, one saved inquiry at $400 covers almost three months of the service. For a property manager, one saved tenancy covers the year.
What Makes This Work Long-Term
Three things determine whether this becomes a sustainable business rather than a short-term experiment:
Retention. If clients see value in the first 60 days — specifically, if they can point to calls they would have missed that turned into jobs — they keep paying. Retention in the AI receptionist category tends to be high because the pain of cancelling (missing calls again) is immediate.
Referrals. Each happy client knows 5–10 potential clients. Actively ask for introductions. “Who else do you know who’s in the same situation — can’t always get to the phone, losing jobs?”
Systematised onboarding. Every new client should be set up the same way, in the same amount of time. A documented process that takes 2 hours per client, not 8, is the difference between a scalable business and a time sink.
The Realistic Timeline
Month 1–2: Get your own business running on the system. Gather data. Month 2–3: Convert 2–3 contacts from your network with free trials. Turn them into paid clients. Month 3–4: 5 paying clients at $149 = $745 MRR. Referrals starting. Month 6: 10–15 clients at $149 = $1,490–$2,235 MRR. Business has its own momentum. Month 12: 25–40 clients = $3,725–$5,960 MRR. Approaching meaningful side income.
None of this requires advertising, a sales team, or venture funding. It requires a working product, the first few clients, and consistency.
What You Need to Start
- A working AI phone system (built or using a provider like Retell AI, Bland AI, or similar)
- A NZ phone number (from Twilio or another provider)
- A way to send notifications (SMS or email when a call comes in)
- A first client — ideally yourself
The technology is the smallest barrier. The real work is in the first sales conversations, the first client setup, and building the habit of consistent follow-up.
New Zealand’s small business market is genuinely underserved for this right now. That won’t last forever. But the window is there.