What an AI receptionist should ask before labeling a plumbing emergency

Every caller with a plumbing problem believes it is an emergency. The person with a dripping bathroom faucet and the person with water pouring through a ceiling light fixture both open with the same words: 'I need a plumber right now.' If your front desk — human or AI — takes that framing at face value, everything becomes an emergency, which is the same as nothing being one.

Triage is the skill of asking a few questions that separate real risk from real inconvenience, without making a stressed caller feel interrogated. Good dispatchers do this instinctively. An AI receptionist has to be built to do it deliberately, and the difference between a good one and a bad one shows up exactly here.

This guide lays out the questions that matter, why each one changes the outcome of the call, and how to evaluate whether an answering setup — AI or otherwise — actually protects your customers and your schedule.

Triage starts with risk, not scheduling

The natural instinct of any booking system is to jump to 'when works for you?' That is backwards for emergency work. The first job of intake is to establish whether damage is actively happening, because that answer changes everything downstream: the advice the caller gets, whether the call escalates to you immediately, and whether the appointment is 'first available' or 'next Tuesday.'

A risk-first flow also protects you legally and reputationally. A caller who mentions a gas smell and gets booked for a Thursday appointment is a story that ends badly for everyone. The intake layer must be able to recognize the handful of situations where the right answer is not an appointment at all.

The five questions that separate emergencies from everything else

1. Is water actively flowing where it should not be?

This is the single most important question in plumbing intake. Active water — a burst pipe, a failed supply line, sewage backing up into living space — means damage is accumulating by the minute. Anything active gets emergency handling: immediate shutoff guidance, escalation, and the first available truck. A leak that has stopped, or one contained in a bucket, is urgent but not an emergency.

2. Can you reach your shutoff valve?

The follow-up that turns a panicked call into a controlled one. A caller who can close the main turns an active emergency into a stabilized situation — still high priority, but no longer minute-by-minute. Intake should be able to give simple guidance: where the main usually is, what direction to turn it, and to shut off the water heater if the house has no water flowing to it. If the caller cannot find or turn the valve, that fact alone raises the severity.

3. Do you smell gas or see water near electrical?

Two situations where the right response is not 'we can be there in an hour' but 'leave the house and call the gas company / do not touch the panel.' These are rare, but an intake flow that cannot recognize them is unsafe. Any mention of gas smell, hissing near a gas line, or water contacting outlets, panels, or fixtures should trigger immediate safety guidance and a live escalation — never a routine booking.

4. Do you have working water, toilets, or heat?

Loss of essential service is the middle tier of severity. No water to the house, no working toilet, no hot water in winter — none of these are flooding the kitchen, but none of them can wait until next week either. The honest classification is 'urgent': same-day or next-morning response, ahead of routine work, behind active emergencies.

5. Is the problem getting worse, stable, or intermittent?

Trajectory is the tiebreaker. A slow drip that has been the same for a month is routine. The same drip that started yesterday and doubled today is a pipe telling you its plans. Asking 'is it getting worse?' catches the calls that look routine on the surface but deserve a faster slot.

What each answer should actually do

Questions are only useful if the answers drive different outcomes. A triage flow needs a real decision behind it:

• Active water or safety risk: give immediate stabilization guidance, escalate to the plumber live, and book the first available emergency slot.
• Essential service lost, situation stable: book the earliest urgent window, flag the job as urgent in the summary, and tell the caller exactly when to expect the truck.
• Routine repair: book a normal appointment with honest timing, capture full intake details, and confirm by text.
• Sales or estimate shopping: capture the lead with the job details and let the plumber decide the follow-up — do not burn an emergency slot on a price comparison.

Notice what is not on that list: 'take a message.' A message is not an outcome. Every call should end with the caller knowing what happens next and when.

The mistakes that break triage

Asking too many questions

A caller standing in water will tolerate about four questions before they hang up and dial the next shop. Triage flows fail when they are designed like insurance forms. The five questions above sound like a lot written down, but most calls resolve after two — the first question eliminates the emergency path for the majority of callers, and routine intake can proceed at a calmer pace.

Treating the caller's word 'emergency' as the classification

Callers self-classify based on stress, not risk. Some of the calmest callers have the worst problems — the retiree who mentions, offhandedly, that the water heater 'has been leaking on the gas valve for a while.' Classification has to come from the facts gathered, not the adjectives used.

Booking emergencies like routine work

If your calendar treats a burst pipe and an annual inspection as interchangeable slots, triage accomplished nothing. Emergency classification has to connect to scheduling: a protected same-day capacity, a defined escalation path, or at minimum an immediate text to the plumber with the job details.

How to test any answering setup — including an AI

Whether you are evaluating an AI receptionist, an answering service, or your own office's habits, run the same test. Have a friend call after hours with three scripts:

• A true emergency: 'water is spraying under my kitchen sink and I can't make it stop.' The call should produce shutoff guidance, an escalation, and an emergency booking — not a message.
• An urgent-but-stable problem: 'we have no hot water.' The call should produce a same-day or next-morning booking and a clear expectation, without emergency dramatics.
• A routine request: 'I'd like a quote on replacing a toilet.' The call should produce complete intake and a normal appointment, without the caller being able to talk their way into an emergency slot.

Score the results on outcomes, not politeness. Plenty of answering setups sound friendly and fail all three scripts. The ones worth paying for get the classification right and the next step booked before the caller hangs up.

Triage is not about complexity — it is about asking the few questions that carry real information, and wiring the answers to real decisions. Get that right and every caller gets the response their situation actually deserves.