AI Answering Services for Plumbers, Electricians and Swiss Home-Service Companies

A running toilet and a burst pipe, booked the same way

Two calls come in on the same Tuesday morning. One is a toilet that has been running on and off for a couple of weeks — mildly annoying and pushing up the water bill, but nothing anyone would call a crisis. The other is a burst supply pipe under a bathroom floor, and water is already finding its way through the ceiling of the room below. Both callers describe their problem as “urgent”, in roughly the same tone, because to each of them, in that moment, it genuinely is.

A business with three vans and two technicians on the road cannot afford to treat these two calls as interchangeable. Yet that is exactly what happens when an AI answering service is built to log “urgent” as a tick-box rather than something to actually establish. The running toilet gets the next free slot this afternoon, because the caller sounded stressed. The burst pipe waits behind it, because nobody asked the one question that would have told the two calls apart: is water actively running right now, or has it already stopped?

This is the real case for a properly built AI answering service in a Swiss trades business, and it is narrower than most sales conversations make it sound. The value is not that every call gets answered — a voicemail greeting technically does that too, in the sense that it produces a sound. The value sits in the handful of questions asked before anything gets booked, because a business with a small, fixed number of technicians has almost no spare capacity to absorb a dispatching mistake. This article sets out what those questions are, what a shallow version of each one looks like next to a working one, and where a script like this earns its keep — and where it is more machinery than a small operation actually needs.

Why three vans change the maths

Compare this to a business where a missed detail is merely awkward. A hotel that books a room for the wrong date can usually move the guest, apologise, and absorb the cost of a free breakfast. An office that transfers a call to the wrong department loses two minutes. A plumbing, electrical, heating or locksmith business running two or three technicians does not have that kind of slack. Each technician is one finite, non-fungible unit of capacity for that day — once a slot is booked it is booked, and once a van is on the road to the wrong address, there is no quiet corner of the schedule to absorb the error.

This is also where the Swiss skilled-trades labour shortage earns a mention it rarely gets in this context. A team already stretched thin cannot treat a wasted callout as a line in a monthly report the way a larger company might. It is a missed afternoon for the customer who actually needed help, and often a technician standing at the wrong door, phone in hand, waiting for someone in the office to establish what should have been settled before the van left.

The six questions a working script asks before anything gets booked

None of these six questions is exotic. What matters is that a script actually asks them, in a form specific enough to act on, before a technician is dispatched or a slot is promised.

• Service area or postcode, asked before anything else is discussed — Swiss street names repeat across neighbouring communes, and confirming the postcode early tells the business whether the job is even within reach before anyone gets excited about a same-day slot.
• Job type in the caller’s own words, mapped to a short list of categories — enough to route the call to the right specialist without pretending to diagnose the problem over the phone; an electrician does not need a blocked-drain description read out in full.
• A specific urgency signal, established by a direct follow-up question rather than inferred from tone of voice — “is water actively coming out right now, or has it stopped?” tells a dispatcher far more than a caller’s claim that something is “pretty urgent”.
• A photo or short video, requested by text message rather than described down the phone — a caller’s idea of “a bit of a leak” and a technician’s idea of the same leak are not reliably the same job.
• Access instructions — entry code, key location, parking restrictions, a dog in the garden — the detail that turns a fifteen-minute job into a forty-minute one before a single tool comes out of the van.
• A realistic availability window, checked against actual technician capacity rather than promised to keep an anxious caller happy — “someone will call you back within the hour to confirm a time” beats an optimistic slot nobody can actually fill.

Good script, bad script: the same questions, two different outcomes

The difference between a script that helps and one that quietly creates work rarely shows up in the greeting — it shows up in what happens the moment a caller says something ambiguous. The examples below are constructed illustrations of the pattern, not transcripts of a real call, but the shape repeats often enough in practice to be worth naming precisely.

• Service area — Bad: the caller asks for a same-day visit and the assistant confirms a 2pm slot without ever asking where the job is; the office only discovers the address sits outside the service area once a technician is already fifteen minutes into the drive. Good: the assistant asks for the postcode or street and town first, confirms the job is within range, or is honest that it sits at the edge and promises a call-back within a set time instead of guessing.
• Urgency — Bad: the caller says “there’s a leak, it’s pretty bad”, and the assistant marks the job urgent based on tone alone; “urgent” now means nothing more specific to the technician choosing which of two jobs to reach first. Good: the assistant asks whether water is actively coming out right now or has slowed, and whether it is reaching a room below — a specific detail a dispatcher can actually act on, not a mood.
• Photo — Bad: nobody asks for a photo; the office finds out on arrival that “a bit of rewiring” is a scorched socket needing immediate isolation, and the van is carrying the wrong parts. Good: the assistant asks the caller to send a quick photo or short clip to a given number, so the technician can see the actual job — and bring the right parts — before leaving the yard.
• Availability — Bad: the assistant books “sometime this afternoon” to sound accommodating, without checking who is actually free; the office then has to call the customer back to walk the promise back once the schedule is checked. Good: the assistant offers a window it has actually checked against the day’s bookings, or commits to a call-back by a specific time rather than a slot nobody can guarantee.

What operational chaos actually looks like

Illustrative scenario: Here is how a shallow script turns two ordinary calls into a bad afternoon. At 10:40, a caller who has never dealt with a plumbing problem before rings in genuine distress about a toilet that will not stop running — nothing is flooding, but she sounds close to tears. At 11:05, a second caller rings about a burst pipe under his bathroom floor; he has already shut off the mains valve himself and reports the problem calmly, because he has been through this before. A script that reads urgency from tone alone logs the first caller as the priority and the second as “standard, can wait” — the exact opposite of what a plumber would decide standing in either room.

The business’s one free afternoon van goes to the running toilet first. By the time the office realises the second address has water actively finding its way into a ceiling below, the technician is already mid-job, and the schedule has no second van to send. The calm caller — who did everything right — waits an extra ninety minutes for help that should have gone out first, and the business only avoids a genuinely expensive claim because he had already shut the water off himself. None of this required a technical failure. It required one missing question: not “how urgent does this feel to you?”, but “what is the water actually doing right now?”

The same shallow script creates the wrong-address version of this problem just as easily: two similarly named streets in neighbouring villages, a postcode nobody confirmed, and a technician who only discovers the mistake once he is parked outside the wrong house. In a larger organisation, a mis-routed job is an inconvenience absorbed somewhere in the schedule. In a business with two or three technicians and no spare capacity, it is very often the difference between finishing the day on time and calling four customers to apologise for running late.

Setting quote expectations without inventing a price

Swiss trades customers are generally used to the idea that a firm quote requires someone to actually look at the job — a phone call is a starting point, not a binding estimate. A good qualification script works with that expectation instead of fighting it: it can share a rough starting range if the business has agreed one for standard callouts, but it should say plainly that the final price depends on what the technician finds, and that anything beyond the initial assessment gets confirmed before work begins.

What it must never do is let a confident voice sound more certain about a price than the person who eventually has to stand behind it in person would be. Over-promising creates an argument at the door before any work has started — a customer quoted an illustrative “around CHF 150” during a two-minute call is unlikely to accept a bill three times that size gracefully, even when the extra cost is entirely justified by what the technician found on arrival. Under-communicating causes a quieter version of the same problem: a caller told nothing about cost until the technician is already on site may decline the work once they hear the number, and the callout — plus the slot that could have gone to another customer — is a cost the business absorbs for nothing.

What can go wrong even with a good script

A well-designed script narrows the ways a call can go sideways; it does not eliminate them. A few limits are worth knowing before anyone assumes the qualification stage is now solved for good.

• A caller under- or over-states urgency, honestly. Even a specific question only works with the information the caller actually gives it — a nervous first-time caller and a calm, experienced one can describe the same severity very differently. The fix is a second, closed confirming question, not trusting a single adjective.
• No photo arrives. Some callers are on an old handset, in a hurry, or simply do not send it. A working script needs a fallback set of specific closed questions for exactly this case, rather than quietly dropping the detail and booking on a vague description.
• The address sits right at the edge of the service area. Guessing either way creates a problem — promising a slot the business cannot reach, or turning away a job that was actually fine. The safer default is an honest “we will confirm and call you back shortly”, not a yes or no given on the spot.
• A genuinely urgent flag is raised correctly and then sits unread. A qualification script is only as useful as the escalation behind it — if an urgent call does not actually reach a person quickly, the extra precision in the questioning has not changed the outcome for the customer waiting.
• A caller’s description does not match what the assistant was configured to recognise — an unusual dialect, a mixed-language sentence, or a problem the business has never described in its own materials. The honest response is a quick handover to a person, not a confident guess dressed up as an answer.

When this is not the right fix

This kind of script earns its keep when the person answering the phone is not the same person who will do the job, and when there is more than one technician whose time can be misallocated. Plenty of Swiss trades businesses do not fit that picture, and it is worth saying plainly where the investment matters less.

• A sole trader who personally answers calls between jobs already asks these questions in their own head, in real time — a formal script adds process without adding much judgement that was not already there.
• A business running mostly on scheduled contract maintenance, with very little ad hoc emergency call volume, has less to gain from urgency-triage specifically, even if the same assistant can still help with routine booking and enquiries.
• A business that cannot commit to acting on an urgent flag within a defined time gains little from sharper triage — better questions only pay off if someone actually responds to what they surface.

The real test before you commit to one

Judging an AI answering service by whether it picks up the phone is judging the easy 90 percent of the job. The harder and more valuable 10 percent is what it does with an ambiguous answer — a caller who says “pretty urgent”, a job description that could mean two different things, an address that might or might not be in range. Before agreeing to anything, ask to hear or read a transcript of exactly those moments, for a job type close to your own trade, and check it against the six questions in this article rather than against a demo built to sound impressive in general.

The six questions above give you a qualification flow for your own trade: work through them one by one and you can see where a shallow version quietly creates the kind of afternoon described above — before it happens with a real customer and a real van.