After-hours maintenance triage: four questions that decide whether a tech drives out

An after-hours callout costs somewhere between two and four times a daytime repair once you count overtime, drive time, and the fact that the tech has no parts store open.

Most portfolios have no written definition of what qualifies for one. What they have instead is an answering service reading a script, a resident describing the problem in their own words, and an on-call tech deciding at 11pm whether to get dressed.

The answering service is making a clinical judgment with no training

Worth being blunt about the mechanism, because it explains the cost.

A resident calls at 10:40pm and says the toilet is overflowing. The service has two choices, dispatch or log for morning. Dispatching is always the defensible choice for the service, because a dispatch that was unnecessary produces a complaint from you and a non-dispatch that was necessary produces a flooded unit and a claim.

So the incentive is asymmetric and the service dispatches. Every service, every time, until you give them a decision rule that protects them for not dispatching.

That rule is the deliverable. It is one page, and it is the highest-return document in after-hours maintenance.

Four questions, in order

The triage script should resolve in under ninety seconds.

Is water actively entering the building or the unit? Active water is always an emergency. Not a slow drip under a sink with a bucket under it, but water arriving faster than it leaves. This is the only category where the cost of waiting compounds hourly, and it is the reason water leads the list.

Is anyone unsafe? No heat below roughly 50 degrees outside, no cooling above roughly 90, gas smell, exposed wiring, a door or lock that will not secure, a smoke alarm that will not stop. Gas and electrical go to the utility or the fire department first and to your tech second.

Is the unit uninhabitable tonight? No water at all, no working toilet in a one-bathroom unit, no power. The test is whether the resident can reasonably sleep there, not whether they are inconvenienced.

Everything else waits. A dripping faucet, one burner out, a garbage disposal, an appliance, a light fixture, a slow drain that still drains. These are morning work, and telling the resident that clearly at 10:40pm with a committed morning time is a better experience than a tech arriving at midnight with no part.

The fifth question the script should ask and usually does not

Whether the resident can mitigate.

A resident who can shut off the supply valve under a toilet converts an emergency into a morning repair. A resident who does not know where the valve is cannot. Most do not, because nobody showed them at move-in.

Two sentences in the triage script asking whether the resident can locate and close the valve, plus a move-in walkthrough that points at the water heater shutoff and the breaker panel, removes a real fraction of overnight dispatches. It costs nothing.

Log every call, including the ones you did not dispatch

This is the part that gets skipped and it is the part that makes the whole thing improvable.

A dispatched call becomes a work order and shows up in your system. A call triaged to morning frequently becomes a phone note that dies with the answering service, which means your after-hours volume looks like your dispatch volume and you have no idea what you declined.

You want both, because the declined calls are where the pattern lives. Twelve declined calls from the same property about the same thing is a preventative maintenance finding hiding in an answering service log.

GetDone answers the after-hours phone with an AI agent that triages against a script like the one above, escalates through your on-call tree, and creates the work order with the recording and transcript attached. Declined calls land in the record too, with a morning commitment, which is what makes the declined-call pattern above visible at all.

What it is not is a staffed call center. If you want a person rather than an agent answering at 2am, that is a legitimate preference and it is worth naming when you evaluate.

The high-risk overlap is the part with the largest tail

Two of the four emergency categories are also the categories that turn into claims: water and anything life-safety.

That means an after-hours water call needs the same documentation as a daytime one. Photos at arrival, moisture readings, what was shut off and when, and a supervisor review. The 24 to 48 hour mold window does not pause overnight, and an overnight water event with no photos is the one an insurer asks about eight months later.

In our own portfolio before high-risk work was gated, 116 high-risk orders were open with 90 past two days, and after-hours orders were overrepresented in that set for an obvious reason: nobody documents well at 1am unless the workflow makes them.

Measure three things next quarter

After-hours calls received, after-hours calls dispatched, and the ratio between them.

If you dispatch more than about half of what comes in, the script is missing. If you cannot produce the first number at all, the answering service is running your labor budget.

Which of those three can you pull today?