Every trade business has the same phone problem, and it is not the one people describe.
The problem is not that you miss calls. It is that the calls you miss are not all the same call. A burst pipe at eleven at night, a landlord chasing a compliance certificate, and someone thinking about a bathroom in March all arrive on the same number, and by the time you come down off the roof they are three identical missed-call notifications.
That is a sorting problem. Most AI receptionist setups are built as booking systems, which is why they disappoint in trades.
The trade call is triage
In a lot of businesses the phone is a diary. Someone rings, they want a time, you give them one. Photography has almost exactly that shape, where the whole call turns on a date.
Trades are different. The first question is not when, it is how bad.
Because the answer determines everything after it. If there is water coming through a ceiling, nothing else on the call matters and somebody needs to be woken up. If someone wants a quote for a rewire, the right outcome is a good set of notes and a call back on Tuesday. Same phone, same greeting, opposite handling.
A system that treats both as bookings will either fill your calendar with emergencies it should have escalated, or page you at midnight about a bathroom renovation. Both destroy trust in the thing within a fortnight.
You have to define urgent, and nobody wants to
This is the part that gets skipped, and it is the reason most of these projects underperform.
An AI receptionist can ask about water, gas, power, heat and safety. It can branch on the answers. What it cannot do is decide your threshold, because your threshold is a commercial judgement about when you are willing to leave your dinner.
So it has to be written down. Not in general terms, in specifics:
What always escalates. Gas smell. Water actively entering the building. No power to the whole property. Anything involving a fire, a shock or a fall.
What escalates only in hours. No hot water. A single dead circuit. A blocked drain that is not backing up.
What never escalates. Quotes. Compliance paperwork. Anything a tenant is reporting that the agent has not authorised.
What you do about repeat callers. If someone rings three times in an hour, that is information, and most scripts ignore it.
Writing that list takes an afternoon, and it is the same exercise as mapping any process before you automate it. It is also the deliverable that survives if you ever change vendors, which is worth remembering when you are being sold a platform.
The caller usually cannot describe the problem
Worth designing around, because it separates a script that works from one that reads well.
People ring a plumber to say the toilet is broken. That covers a running cistern, a blockage, and a cracked pan leaking into the subfloor, which are three different jobs with three different urgencies and two different vans.
The fix is not more questions. It is better ones, asked in the right order. Open description first, so the caller says it in their own words and you get the detail nobody would have thought to ask for. Then the narrow safety questions, because those are the ones you need reliable answers to. The general principle is in the piece on what an AI receptionist should ask, and trades are the clearest case for it.
One more thing that matters here: get the callback number in the first ten seconds, before anything else. Job-site reception is unreliable, and a call that drops halfway through is only recoverable if you already have the number.
After hours is where this actually pays
Most trades evaluate an AI receptionist against the working day. That is the wrong window.
During the day you get to the phone eventually, or a partner does, or the caller tries again. It is not ideal, but the loss is limited. After five, and on weekends, the loss is total: an emergency call that rings out goes to whoever answers next, and that is usually a larger outfit with a call centre or a competitor paying for after-hours cover.
The arithmetic is straightforward enough to do yourself. Count the after-hours calls you currently miss over a fortnight, apply your callout rate to the ones that were genuinely urgent, and compare it to a monthly fee. For a lot of trades the after-hours calls alone settle it and the daytime coverage is a bonus. The full working is in the post on whether one of these is worth it.
Be honest with yourself in one direction, though. If you are not actually willing to get out of bed, do not buy a system that wakes you up. Set it to take good notes overnight and call people back at seven. That is a legitimate configuration and a lot of trades would be happier with it.
Quotes and callouts are two different calls
The other thing worth splitting, because they need opposite handling.
A callout wants speed and address accuracy. Where, what, how bad, how do we get in, is there a dog, is there parking. Short, structured, escalate or book.
A quote enquiry wants scope. What is the job, how big roughly, when are they thinking, is it a rental or owner-occupied, has anyone else looked at it. Longer, looser, and it should never end in a price.
That last part is not negotiable. Do not let a phone system quote. Trade pricing depends on what you see when you get there, and a number said on the phone becomes the number the customer remembers regardless of what your written quote says afterwards. The system can state your callout fee if that is already published, and nothing else.
Once a quote enquiry is captured properly it should flow into the rest of your admin rather than sitting in an inbox, which is the handoff between systems where most of the time actually goes. A booked callout has a second chain attached to it too, running from the confirmation through to the on-the-way and running-late messages, and the phone system is only useful if what it captures feeds that.
What to keep human
Anything that has gone wrong. A customer ringing about a job you did that has failed reaches you immediately, with no triage. That should be a hard rule the system cannot override, and it belongs alongside the rest of your operating limits.
Anything about money already owed. Disputes, payment plans, invoices someone is unhappy with. Never automate a conversation where the customer is already annoyed.
Your regulars. If you have builders or agents who put work your way, they should have a number that rings you. Sending your best repeat client through a phone tree is a false economy.
Say it is a system
Tradies worry about this more than other trades do, and the worry is misplaced.
Nobody with water coming through a ceiling cares who answers, as long as somebody does and something happens. What annoys people is finding out later that the warm chat about their kitchen was a machine. Saying so at the start costs nothing and removes the risk entirely, and modern systems sound good enough that the disclosure is the only thing distinguishing them anyway.
When this is not worth it
Three cases, and they are common.
You work entirely for two or three builders who text you. You are not missing calls, you are being handed work, and a phone system solves a problem you do not have.
You are winding down, or deliberately staying small and full. More enquiries are not a benefit if you are already turning work away.
You have someone in the office. If a person already answers reliably during hours, the honest scope is after-hours cover only, which is a much smaller and cheaper piece of work than what you will be quoted for if you ask for a full receptionist.
If you are not in one of those three, this is probably the highest-return thing you could automate, and it is the first of the four jobs I would do for a trade business. Whether it needs a build or just a subscription depends a lot on what you are already running, which is the subject of the piece on the software a trade business actually needs.
Tell me roughly how many calls you miss in a week and what happens to them now, and I will tell you honestly whether this is worth building. Send me the shape of it, and if the answer is that an off-the-shelf subscription would do the job, I will say that.
Frequently asked questions
Is an AI receptionist worth it for a plumber or electrician?
Usually yes, and for a different reason than most trades expect. The value is not in booking work, it is in sorting the emergency from the enquiry while you are under a house. If you take after-hours calls at all, or if you regularly finish a job to find three missed calls and no idea which one was urgent, the case is strong. If you work entirely on referral for a small set of builders, it is weaker.
Can an AI receptionist tell an emergency from a normal job?
Only if you define what counts as an emergency first. The system can ask about water, gas, power, heat and safety, and it can escalate on the answers, but it cannot invent your threshold. Most bad experiences in trades come from nobody writing that rule down, so the system either escalates everything or nothing.
Will it work if the caller is at a noisy job site?
Mostly, and better than it did two years ago. Modern systems handle background noise and accents reasonably well. What still causes trouble is very poor mobile reception, where the audio drops out mid-sentence. The safeguard is taking a callback number in the first ten seconds, so a dropped call is still recoverable.
Should it book jobs straight into my calendar?
For routine work, yes, if the calendar reflects reality. For emergencies, no. An emergency should page you rather than take a slot, because you are the one who decides whether to leave the job you are on. Booking and escalating are two different outcomes and the system needs both.
What does this cost for a small trade business?
A subscription product is a monthly fee plus per-minute usage, and it is the cheapest way in. A build that connects to your job management software, checks a live calendar and escalates by your own rules costs more up front and is priced as a project. Our rates are published on the pricing page.
Wondering what this would look like in your business? A short chat is usually enough to tell.
Let’s chat