DEV Community

Adel Atya
Adel Atya

Posted on Originally published at adelatya.com

After-hours calls: what home-service businesses lose overnight

About one HVAC call in ten to one in seven arrives off-hours: 9.8% to 14.1% of inbound calls, depending on the month, in ServiceTitan's 2025 data. No plumbing figure exists that I could verify.

Below: what that dataset shows, how to price your own overnight loss in one evening, and when an AI agent is not worth it.

How many home-service calls arrive after hours?

Roughly one in ten to one in seven, for residential HVAC. ServiceTitan counted every call outside Monday to Friday, 8am to 5pm in each shop's local time. Its 2025 numbers:

  • June peak: 14.1% of inbound calls came off-hours.
  • October low: 9.8%.
  • Weekday evenings and early mornings: 9.3% in June, 6.8% in October.
  • Weekends: 4.9% at the August peak, 2.95% in October.

Demand bunches up between 5pm and 9pm. A call at 3am is rare. A call at 7pm, while the owner eats dinner, is the usual one.

Two cautions. The data covers HVAC only, and ServiceTitan says it is not nationally representative. I found no published plumbing figure. Higher figures for "home services" circulate in vendor blogs, but I could not trace them to a primary study, so I leave them out.

Is the loss really overnight, or is it during the day?

Possibly during the day. Invoca's 2026 home services report found a person answered 52% of inbound calls to home service businesses. Leave out calls under 15 seconds and that rises to 65%.

If your shop answers only half its calls, the gap may be bigger at noon than at midnight. Count both before you spend anything, using the method in how many calls a plumber misses.

Why can an overnight call be worth more than a daytime one?

Night and weekend jobs often bill higher.

Housecall Pro lists emergency diagnostic fees of $149 to $259, usually separate from the repair, and says nights, weekends and holidays often run 1.5 to 3 times the standard hourly rate.

Voicemail loses callers. Moneypenny's 2016 call report found 69% of callers who reached voicemail hung up without leaving a message. The figure is old, so treat it as a direction, not a measurement. A homeowner with water on the floor will likely dial the next company on the list.

What does it look like when you price your own number?

Use your phone log, not an industry average. Here is an illustration with made-up figures, not a client result.

  • You log 12 real customer calls off-hours in a month.
  • You answered 3 yourself. That leaves 9 missed.
  • Of those, 3 were emergencies and 6 could wait until morning.
  • Emergencies close at 80% on a $400 ticket. Routine calls close at 25% on a $250 ticket.

That is 3 × 0.8 × $400 = $960, plus 6 × 0.25 × $250 = $375. Call it $1,335 a month, or about $16,000 a year, before you subtract the cost of covering those calls.

Now change the inputs. At 4 real missed calls a month, all routine, the loss is about $250. No software pays for itself there.

What should happen to each overnight call?

Not every call deserves the same response. Sort them before you pick a tool.

  1. Emergency (burst pipe, no heat in a freeze, gas smell). Needs a human or a dispatch slot within minutes. A gas smell should get a safety instruction first, not a booking.
  2. Urgent but not dangerous (water heater out, AC down in a heat wave). Can book the first morning slot, if the caller hears that promise tonight.
  3. Routine (quote request, maintenance, "do you service my area?"). A text and a morning callback is enough.
  4. Noise (spam, vendors, wrong numbers). Log it. Never wake anyone.

Check how your shop handles each tier today. In my view tiers 2 and 3 are the ones most likely to slip, because a callback at 8am can reach someone who already hired another company.

What can an AI phone agent do with those calls?

It answers in about a second. The plumbing and HVAC page describes what mine does on an emergency: it asks the safety questions first (is the water off, is there standing water, is anyone at risk), then books the emergency slot or transfers straight to your on-call number, depending on the rules you set. It quotes the dispatch fee and any flat-rate items you give it, and nothing else. Urgent, angry or out-of-rules calls go to a human, and it says so out loud.

On my Blue Ridge Plumbing demo, a missed call gets a text in 8 seconds and an AI callback in 52.

Emergency handling is the path I test hardest. My suite runs 38 simulated callers, 114 calls per run, for about $0.25 a run. Its first run caught 5 missed emergency handoffs in 41 plumbing calls on a cheaper model. See how I test phone agents.

When is an AI agent not the right choice?

  • Your log shows only a few real off-hours calls a month. The worked example above shows when that stops paying. A text-back or a better voicemail greeting is enough.
  • Your daytime answer rate is the real problem. Fix the noon calls first. A voice agent costs $900 to build and $450 a month with 500 minutes (pricing), so $6,300 in year one.
  • You already run a live on-call line that people answer. Your emergency tier is covered.
  • Your jobs need a person to price them on the phone. The agent quotes only the fees you give it.
  • You want a machine to handle a true emergency alone. Transfer those to a person.

Run the arithmetic honestly. In the example above, $6,300 against a $16,000 yearly loss looks good, but halve the recovery rate, because an agent will not rescue every caller, and it is still ahead. At 4 missed routine calls a month, it is not.

Checklist for this week

  • Export 30 days of call logs and mark calls outside your hours.
  • Sort the real customer calls into the four tiers above.
  • Count how many you answered, and how many got a callback within an hour.
  • Price tiers 1 and 2 with your own close rate and ticket, then halve the result.
  • If the figure beats the cost of an AI receptionist, call the demo agent and say your sink is backed up.

Originally published at adelatya.com.

Top comments (0)