Plumbing emergencies after midnight: how to capture without burning out
12:47 in the morning. The phone on the nightstand rings. The on-call tech (who, this week, is you) picks it up. A homeowner in Oakville says there's water coming through her kitchen ceiling and she can hear it dripping. Her husband shut off the main but she can already see the drywall sagging. Tomorrow is her daughter's birthday party. Can someone come now.
This call is why plumbing is plumbing. It's also why most plumbing shop owners are tired all the time. The math on after-hours emergencies is real money. The math on your own sleep is also real. The trick is building a system that captures the money without burning out the team that has to answer the phone at 1am.
What an after-hours plumbing emergency actually is
Three call shapes show up after midnight in residential. The first is the one above: real emergency, water somewhere it shouldn't be, panicked homeowner, has to be addressed now. This is the most valuable call shape because the alternative for the homeowner is real damage. Average ticket on these is high. Conversion is high.
The second is the "I think I have an emergency" call. Maybe it's a slow drip, maybe a hot water tank that's hissing weirdly in a way the homeowner has never heard before. The homeowner doesn't know if it can wait. Half the time it can. Half the time it can't. Either way they want to talk to someone right now to find out.
The third is the surprise. A landlord calling because his tenant just texted at 1am about a sewer backup. Or a real estate agent calling because a showing tomorrow needs the basement to not smell like a swamp.
Each shape has a different cost structure and a different right answer. Treating them the same is the mistake most after-hours systems make.
The triage problem
The hard part about after-hours plumbing is that most after-hours calls are legitimate but only some are true emergencies. If your on-call tech rolls every call, you burn him out fast and most of the rolls are routine work that could've waited until 7am.
The cheap version of triage is voicemail with a "press 1 for emergency" greeting. It works some of the time. Often the homeowner panic-mashes the keypad or just hangs up because they think nobody's home. Voicemail can't pick the urgent calls out of the routine ones because it can't ask follow-up questions.
The next version is a 24/7 answering service. Real human picks up, asks the script you gave them, decides whether to dispatch the tech. This works better but costs $400-$1,200/month for the kind of trades-shop volume we're talking about. And the script the human follows is exactly the script you wrote. If you wrote it wrong, the dispatch decisions go wrong too.
The version we built into Avidra is different. AI picks up, asks the homeowner via text or voice what's going on, and routes based on the answers. If the homeowner says "water is coming through my ceiling," the AI flags as emergency. If the homeowner says "my toilet runs sometimes but it's been like that for a week," the AI captures the lead but doesn't wake the tech up at 1am. Both calls get captured. The on-call tech only gets paged for the first one.