After-hours calls: on-call tech vs answering service vs AI

What an after-hours call is actually worth

One analysis of plumbing service tickets found 68% of emergency calls come in outside normal business hours. Pipes burst in the evening, a dead furnace gets noticed at midnight, and the garage door quits at 6:45 AM with the car inside. After-hours is not the leftover slice of your demand. For emergency work, it is the main event.

It is also the most profitable slice. After-hours emergency tickets commonly bill $350 to $800 and up, usually with a trip premium the customer pays without blinking. Miss two a week and you have quietly handed a competitor $36,000 to $83,000 a year.

Voicemail is a leak, not a coverage plan

First, the non-option. Roughly 80% of callers who hit voicemail hang up without leaving a message, and for emergency callers the real number is close to all of them. Nobody with a flooding basement narrates the problem to a machine and waits politely. They hang up, dial the next company on the list, and the first business that answers wins the job.

Voicemail's $0 monthly cost is an illusion. Its real price is those hang-ups at $350 to $800 each. If your after-hours plan is a greeting you recorded in 2019, you do not have an after-hours plan. You have a referral service for your competitors.

The on-call rotation: cheap on paper, paid for in people

The default answer for most shops: one tech carries the phone each week. Answer speed is decent when the tech hears the phone. Booking is where it falls apart. A tech woken at 1 AM does not open the schedule and book a Tuesday tune-up. He says call the office in the morning, and a meaningful share of those callers never do. The rotation answers calls; it does not capture them.

The hidden cost is worse. The call that does the damage is not the 2 AM burst pipe - that one pays. It is the 12:50 AM price-shopper asking what you charge for a faucet swap, waking your best tech for nothing, three nights in one week. Senior techs do not usually quit over wages. They quit over sleep.

Live answering services: real humans, real limits

A live answering service puts a warm human voice on every call, which matters to a panicked caller. The pricing model is where you need to squint: $1 to $2 per minute, with monthly minimums. A five-minute emergency call runs $5 to $10, and at typical field service volume the bill lands between $500 and $1,500 a month - and the minimum means slow months cost you anyway. An in-house overnight hire starts around $37,000 a year in salary alone, before taxes and benefits, for one shift five days a week.

Capability is the bigger limit. Most services work from a generic script, cannot see your schedule, and take messages rather than book - so the caller is recorded, not captured. Emergency triage depends on a third-party operator following your written instructions at 3 AM.

AI answering: flat cost, instant pickup, honest caveats

An AI receptionist answers on the first ring, every time, including three calls at once during a cold snap - exactly when humans and rotations get buried. Pricing is a flat monthly fee, typically a fraction of a per-minute service and nowhere near the cost of a hire. Wired into your actual calendar, it does the one thing the other options cannot: book the non-emergency caller into a real morning slot on the spot, ask structured triage questions, and page a human only when the answers say emergency.

The honest caveat: it is only as good as its configuration. No system knows out of the box that a rumbling water heater is urgent and a running toilet is not. You have to teach it your triage rules, service area, and pricing script, then test it with role-played calls before it takes a real one.

The four options side by side

• Voicemail. $0 a month, but 80% of callers hang up. No booking, no triage. Real cost per missed emergency: $350-$800.
• On-call rotation. No new line item, human answer within minutes - when the tech hears it. Weak booking; triages everything by waking a person. Hidden cost: a $15,000-$30,000 replacement when it burns someone out.
• Live answering service. $1-$2 per minute plus minimums, often $500-$1,500 a month at service-company volume. Warm and human, but usually takes messages instead of booking, and triage rides on the individual operator.
• AI receptionist. Flat monthly fee well below the other paid options. Instant pickup, real booking, rule-based triage - provided you invest the setup hours and hard-wire true emergencies to a human.

The real fix: separate triage from truck rolls

Whichever option you choose, the principle is the same: answering the phone and rolling a truck are two different decisions, and fusing them makes after-hours coverage expensive. Only a true emergency should wake the on-call tech. Everything else becomes one of two things: a booked appointment in tomorrow's first slots, or a structured message with name, number, address, and the problem in one line.

1. Write the emergency list with your techs. If they would genuinely drive out at 2 AM for it, it goes on the list. Cap it at five to eight scenarios.
2. Decide what a non-emergency becomes. Hold two or three protected morning slots for overnight bookings so "first thing tomorrow" is a promise you can keep, not a scramble.
3. Put the triage layer in front of the on-call phone, not behind it. The tech's phone rings only after something - an operator with clear instructions or an AI asking your triage questions - decides the call qualifies.
4. Track two numbers every week. After-hours calls answered, and tech wake-ups. The first should climb; the second should fall until it matches your emergency list.

Run that for a month and the comparison stops being theoretical. You will see how many $350-$800 tickets were arriving at midnight all along - and how few needed to cost a technician's sleep.