On Call Bell Not live yet — in build
For contractors and trades

An answering service for contractors, reduced to four outcomes.

Every after-hours call to a trade ends in one of four places, and an answering service for contractors is worth exactly as much as its ability to tell them apart. The call stops and gets pointed at 911. It wakes somebody tonight. It waits until first thing. Or it goes on the calendar.

This page is the model, not a product tour: what each outcome has to guarantee, the one rule that decides most of the hard cases, and the five questions that will tell you whether a vendor has thought about any of it. The worked example is a plumbing rule set, because that is the one that exists here in code and printed in full; the model is the part that carries across trades.

Read this first

On Call Bell has not answered a phone call yet. The safety gate, the triage rules, the call flow and the evidence trail are written and tested offline; the telephony is the next piece of work. Nothing on this page describes how fast or how reliably a live line answers, because no live line exists to measure. There is nothing to sign up for and no demo number. Everything below is either a decided price or a number you can count in the code.

The model

What an answering service for contractors has to get right.

One: the stop is not a tier. Gas, carbon monoxide, fire, an electrical hazard, someone hurt — these are not the top of an escalation ladder, because the correct action is not to reach you faster. It is to get off the phone so the caller can dial 911. A service that transfers a suspected gas leak to a contractor at 2am has done the wrong thing quickly. The stop must refuse to book, refuse to transfer, refuse to ask a clarifying question first, and tell the caller to get outside before dialling — which is what the CPSC says to do — and where the hazard is carbon monoxide it has to fire on the symptoms as well as the word, because the CDC describes carbon monoxide poisoning as “flu-like” and that is how the caller will report it.

Two: tonight has to survive nobody answering. Waking a tech is a sequence, not an event: ring the on-call phone, say the address and the one-line problem, require a keypress before connecting anyone, and if nobody picks up, try the backup, then text the whole record and tell the caller when to expect a call back. A service that says “we escalate” and means “we forward the call” has no answer for the night the phone is on silent.

Three: first thing is a promise with a name on it. The value of the tier is that nobody gets woken. It only holds if the caller is told who will contact them and by when, and if that shows up somewhere the shop reads in the morning.

Four: scheduled has to read the job back. A request that goes on the calendar unconfirmed, with the details repeated to the caller before the call ends, is worth more than a message that has to be deciphered at 7am.

The hard part

The rule that decides most of the difficult calls.

Every fact a line extracts from a call has three states, not two: true, false, and nobody asked. Most systems collapse the third into the second, and that single shortcut is where after-hours triage goes wrong.

Take one row. Active leak, shutoff not found or not done is a wake-somebody row. If the tri-state is respected, it fires whenever the water is running and the shutoff is not known to be done — including when the question was never asked. If absent silently means false, the same row goes quiet on exactly the call where the caller was too rattled to answer a question.

The mirror row is the one that lets a call wait: leak, water off and contained needs both facts positively established before it downgrades. A de-escalation must come from something the caller actually said, never from an absence.

The general form: a missing answer must never be the reason a call gets quieter. Ask a vendor what happens to their escalation rules when the caller hangs up mid-question. If the answer is a shrug, the rules only work on calm callers, and calm callers were never the problem.

Modifiers

Three things that change the answer without changing the rule.

A rule table alone is too blunt for a real week. Three modifiers each raise a matched rule by one tier, and each can be switched off by the owner:

Vulnerable occupants. An elderly resident, a newborn, someone on home oxygen. The same no-water call is a different call in that house.

Cold snap. A manual switch the owner flips for the week, not a weather feed. Frozen pipes and no heat move up while it is on. It is manual on purpose: the owner knows what their trucks can absorb this week and a forecast API does not.

Multi-unit. One backed-up main in an eight-unit building is eight households, and the tier should say so.

One constraint makes them safe: a modifier only applies when a rule has already matched. A modifier on its own is never a dispatch reason, so “there is an elderly resident” cannot by itself wake anybody at 3am.

Buying

Five questions worth asking any vendor.

  1. Show me the escalation table. Not the marketing description of it — the rows, with tiers. If it cannot be shown, it cannot be argued with, and you will be arguing with it at 2am.
  2. Which rows can I change, and which are read-only? A vendor with no read-only rows has not thought about safety; a vendor where everything is read-only has decided your business for you.
  3. What happens on a suspected gas leak? The only good answer ends the call and names 911. Anything involving a transfer is the wrong answer.
  4. What does the system do when a question goes unanswered? See above. Absent must not read as false.
  5. What does it miss? Every detection system misses things. A vendor who cannot name a single failure mode either has not measured or is not telling you. Ours are counted and published.
Scope

What is honest to say about this one.

On Call Bell is being built for plumbing shops running one to three trucks in the US, and the rule set that exists is a plumbing rule set. It has not answered a phone call. If you are a contractor in another trade, the four outcomes above are yours to use with any vendor; the shipped table is not yet.

The eight things this product refuses to do — including the ones a general answering service would happily sell you — are listed on the small-business page.

Questions

Asked and answered.

What are the four outcomes of an after-hours call?

Stop and refer to 911; wake the on-call tech tonight; capture the job and hand it over first thing; or put an unconfirmed request on the calendar. The stop is not the top of the escalation ladder — it is a refusal to escalate, because the right action is to free the caller line for 911.

Why does "nobody asked" have to be different from "no"?

Because collapsing them makes escalation rules go quiet on exactly the calls where the caller was too rattled to answer. A rule that needs to know whether the water is off must fire when the question was never asked, and a rule that lets a call wait until morning must require both facts positively established.

Should a cold snap change the rules automatically?

Not from a weather feed. It is a switch the owner flips for the week, because the owner knows what their trucks can absorb and a forecast does not. While it is on, frozen pipes and no-heat calls move up one tier.

Is this available for contractors outside plumbing?

No. The shipped rule set is a plumbing rule set, and On Call Bell has not answered a phone call yet in any trade. The four-outcome model on this page is free to use when judging any vendor.

Sources

Where each number came from.

Keep reading
Argue with it

If you run a shop and something on this page is wrong — a rule that would wake you for nothing, a phrase the word list should catch and does not — write to hello@oncallbell.com. A person reads it. That is more useful to us than a signup, and there is no signup yet anyway.

Last checked against the code on 2026-09-06