Lumina Sales is live.See it

Decide what counts as an emergency before the phone agent does

7 min read

By Lumina Software

  • Voice reception
  • Service business
  • Operations
  • After hours
  • Dispatch

Every service business that takes calls after hours eventually gets the same kind of call. A homeowner has water coming through the ceiling at 11pm. A restaurant's walk-in cooler stopped holding temperature an hour before close. A tenant smells gas. The caller wants a person, and they want one now.

When a phone agent answers that call, it has to make a decision the business may never have written down: is this an emergency? If yes, who gets woken up, and how? If no, what does the caller hear, and when will someone call back?

Most of the after-hours setups we review have an answer for the easy cases and nothing for the edges. The owner knows a burst pipe is urgent and a dripping faucet is not. But the phone agent does not know what the owner knows. It only knows what someone wrote into its instructions. If nobody wrote it down, the agent guesses, and guesses at 11pm are where businesses lose customers or burn out their on-call tech.

This post covers how to write that decision down before the phone agent has to make it.

Write the emergency list in the caller's words

Start with a list of situations that justify waking someone up. Keep it short and concrete. "Urgent issues" is not a rule. "Water actively leaking from a ceiling or wall" is.

Write each item the way a caller would describe it, not the way a tech would. Callers do not say "supply line failure." They say "water is spraying under my sink" or "the pipe under the house broke." A good list includes a few common phrasings for each situation so the agent can match what it hears.

A plumbing company's list might look like this:

  • Water actively leaking and the caller cannot shut it off
  • Sewage backing up into the home
  • No water at all to the property
  • Water heater leaking from the tank

An HVAC company in a cold climate would add no heat when the outside temperature is below a set number. A property manager would add lockouts and broken exterior doors. The list depends on the trade, which is why it has to come from the business and not from a template.

Then write the opposite list: things that sound urgent to the caller but can wait until morning. A slow drain. A running toilet. A thermostat that shows the wrong time. These are worth naming because callers often push hard on them, and the agent needs permission to say "we will call you first thing tomorrow" without escalating.

Anything that does not fit either list should have a default. For most businesses we work with, the safe default is to escalate. A tech who gets woken up once for a judgment call is annoyed. A customer whose basement flooded because the agent decided it could wait is gone, and may leave a review about it.

Put safety calls on their own path

Some calls are not service emergencies. They are safety emergencies, and the business is not the right first call.

If a caller reports a gas smell, the agent should tell them to leave the building, call 911 once they are safely outside, and then call their gas utility, before anything else. Gas utilities publish this advice themselves. For example, PG&E's gas leak guidance tells people who smell gas to leave the area, call 911, and then call the utility. The same goes for anything involving fire, sparking electrical panels, or someone who is hurt.

Write these as a separate rule that runs before the emergency list. The agent does not need to decide whether a gas smell is a dispatch. It needs to give safety instructions in plain words, then take the caller's details and notify the on-call person so the business can follow up.

Keep the safety wording fixed. Do not let the agent improvise it. Write the exact sentences you want said, have the owner approve them, and test that the agent uses them word for word.

Decide how the on-call person gets reached

Once the agent decides a call is an emergency, it has to reach a human. This is where a lot of setups quietly fail.

A text message to the on-call tech's phone is the most common approach, and it is also the easiest to miss. Phones are on silent. Texts from an unknown number get filtered. The tech is driving.

A better pattern has three parts:

  • A named on-call person for each night. Not "the team." One person, with a backup. The schedule lives somewhere the agent can read it, and someone updates it when shifts swap.
  • A call, not only a text. The agent places a call to the on-call person, reads a short summary of the emergency, and offers to connect them to the caller. A text with the same details follows so the tech has it in writing.
  • An escalation timer. If the on-call person does not answer or acknowledge within a set number of minutes, the agent tries the backup. If the backup does not answer, it tries the owner. The timer and the order are written down, not left to the agent.

Pick the number of minutes with the people who will be on call. Five minutes feels long to a caller with water on the floor. It also might be how long it takes a tech to pull over. The right answer is the one your team will actually meet.

Tell the caller what happens next

The caller does not see any of this. All they know is what the agent says before the call ends. A vague "someone will be in touch" leaves them guessing whether to call a competitor.

For an emergency, the agent should say who is being contacted and roughly when the caller should expect a call back. If the business has a policy about after-hours rates, the agent should say that too, in the words the owner approved. Nobody likes a surprise on the invoice, and a caller who hears the rate up front and still says yes is a caller who will pay it.

For a call that can wait, the agent should say so plainly and give a time. "We will call you tomorrow between 8 and 9" is better than "first thing." Then the business has to keep that promise, which means the morning team needs a list of every deferred call waiting for them when they arrive.

Either way, send the caller a text with what was agreed. It gives them something to look at when they are standing in a wet hallway wondering if anyone is coming.

Review the edge cases every week

The emergency list is never done after the first draft. Real callers describe things nobody predicted, and the agent's choices on those calls tell you where the rules are thin.

Once a week, pull every after-hours call and sort them into three groups: escalated, deferred, and unclear. Read or listen to the unclear ones first. For each, ask the owner one question: what should the agent have done? Then add that answer to the list, in the caller's words.

Also look at the escalated calls the on-call person thought were not worth waking up for. Those are signs the list is too broad. And look at any deferred call that turned into a complaint the next morning. Those are signs it is too narrow.

After a few weeks the unclear pile gets small. That is the point where after-hours calls stop being a source of stress for the owner and start being something the business handles the same way every night.

The short version

A phone agent cannot know what counts as an emergency unless the business tells it. Write the emergency list and the can-wait list in the words callers use. Put safety calls on their own path with fixed wording. Name one on-call person each night, call them instead of only texting, and escalate on a timer. Tell every caller what happens next and text it to them. Then review the unclear calls each week and fold the answers back into the list.

Written by Lumina Software. Have a question? Book a call

Keep reading