After-Hours Emergencies: Building Triage and Dispatch That Holds Up at 2 A.M.

A burst pipe does not wait for business hours. Neither does a tooth abscess, a furnace that quits at midnight in January, or a basement filling with water while someone stands on the stairs holding a phone. I have run on-call rotations for home services shops and a regional dental group, and the thing nobody tells you is this: the technology is the easy part. The hard part is deciding, in the first ninety seconds, who actually needs help right now and who can wait until 8 a.m.

Get that decision wrong in either direction and you pay for it. Wake up your best tech for a clogged drain that could have waited, and he is useless on tomorrow's real jobs. Send a "we'll call you Monday" to someone whose ceiling is about to collapse, and you have lost a customer and maybe earned a one-star review with your name in the headline.

So let me walk through how after-hours handling actually works when it works, and where most shops get it wrong.

First, define what an emergency actually is

Every shop I have worked with thinks they know their emergencies until you make them write it down. They don't. The on-call tech has one definition, the dispatcher has another, and the answering service the owner is paying $300 a month has no idea at all.

Sit down and build a real triage list. Not vague categories. Specific symptoms and the action each one triggers. Here is the shape of it for a plumbing shop:

• Active flooding, sewage backup, no water to the whole house: dispatch tonight.
• No hot water, single slow leak that can be contained with a bucket, running toilet: schedule first thing tomorrow.
• Pricing questions, "how soon can someone come out," general curiosity: take a message, no dispatch.

For a dental practice it is uncontrolled bleeding, facial swelling, or trauma versus a lost filling or a dull ache. For restoration it is standing water and active intrusion versus a dried-out stain someone noticed on the ceiling.

The point is that triage only works if the line between tiers is something a stranger could apply correctly at 2 a.m. with no judgment of their own. If your rule requires experience to interpret, it is not a rule. It is a guess.

The questions that do the triage

A good intake is three or four questions, asked in order, that move from "is this an emergency" to "where and who." Ask them in the wrong order and you waste the caller's patience while their kitchen fills with water.

Start with the disqualifier. "Is water actively coming in right now, or has it stopped?" That one question sorts most plumbing and restoration calls into the right bucket immediately. Then location and access. Then contact details, captured early in case the line drops, because panicked people drop calls.

I tell people to write the script so it could be read by a calm twelve-year-old. Short sentences. One question at a time. No jargon. The caller is stressed, and stress eats comprehension. If your intake sounds like an insurance form, you will get half-answers and bad addresses.

Where the after-hours model breaks

Three failure points, and I have seen all three more times than I want to admit.

The first is the human bottleneck. An answering service takes a message and emails it to a dispatcher who is asleep. The "emergency" sits in an inbox for four hours. The whole chain assumes someone is awake and watching, and at 3 a.m. nobody is.

The second is over-escalation. The service is scared of missing a real one, so it forwards everything to the on-call tech. Now your tech is fielding spam and tire-kickers all night and starts ignoring the phone out of self-defense. The night the real emergency comes in, he sleeps through it because he has been trained to.

The third is the dead end. After-hours rings to voicemail, voicemail is full or never checked, and the lead is just gone. I have watched shops lose five-figure restoration jobs this way. The customer called, got a beep, and called the next company on the list.

A system that handles emergencies well closes all three gaps: it answers instantly, it triages before it escalates, and it only wakes a human when the rules say a human is genuinely needed.

What I do with AI handling the front line now

For the last stretch I have moved the first contact to an AI agent, and it solves the structural problems an answering service never could. It picks up on the first ring, every time, in whatever language the caller speaks, and it never gets tired of asking the same triage questions correctly at 4 a.m. That consistency is the whole game. A human answering service has a bad night; the script does not.

The way I set it up with LastWorker: the agent learns the triage tiers during a short setup conversation, the same list you already wrote down. True emergency, it gathers the details and transfers or texts the on-call tech immediately with the address and the problem already summarized. Tomorrow-job, it books the slot or takes the message. Not-an-emergency, it answers the question and logs the lead. The tech only hears from it when the rules fire.

What I care about most is the escalation logic doing what it is told. The agent should transfer to a human when the situation calls for it and only then. We had to fix a case early on where it was honoring the config setting instead of the actual transfer outcome, and that distinction matters: a "transfer" that quietly fails is worse than no transfer, because everyone assumes the handoff happened. Test that the handoff actually completes. Call your own line at midnight and see what happens.

A simple escalation ladder

Write down who gets contacted, in what order, and how long you wait before moving to the next person. Keep it boring and explicit.

The backup rung matters. On-call people miss calls. They shower, they drive through dead zones, their phone dies. If there is no second name on the ladder, your "system" is one sleeping person.

Test it like a customer, not like an owner

Owners test their own setup by calling and describing the perfect emergency in clear terms. Real callers mumble, panic, and bury the important fact in sentence four. Have someone who does not know the script call in and describe a messy situation. See if the triage still lands. Do it after hours, because that is when it counts and that is when nobody is watching.

The shops that win after-hours are not the ones with the fanciest tools. They are the ones who decided, in advance and in plain language, what counts as an emergency and what happens next. The tool just executes the decision you already made. If you want to see how the per-conversation pricing works for night coverage, the pricing page lays it out, and there is no monthly fee for a phone that mostly sits quiet until the one night it earns its keep.

Write the triage list this week. Everything else is downstream of that.