Missed calls are the most expensive thing in your business

Ask a home service owner what their biggest expense is and they will name something on the P&L. Labor, fuel, parts, insurance. Reasonable answers, all of them, and all of them wrong, because the biggest number in a typical owner-operated trade business never appears on a financial statement at all.

It is the calls that rang while you were under a house.

Why this loss stays invisible

Every other expense announces itself. A bill arrives, you write a check, it shows up in the accounting. A missed call does the opposite: it produces no invoice, no entry, no notification, and no complaint. The homeowner does not call back to tell you they went elsewhere. They just go elsewhere.

And they go quickly. A homeowner with no hot water at 6:40 on a Tuesday is not evaluating vendors. They are working down a list of local numbers, and the meaningful distinction between the companies on that list is not price, reviews, or how long you have been in business. It is who picks up.

The structural problem in the trades is that the person best qualified to answer the phone is the person least able to. In an owner-operated shop the number rings to a phone in a pocket, and the pocket is attached to someone crouched in a crawl space with both hands occupied. That is not a work-ethic failure. It is a design flaw, and it is one of the clearest possible cases of owner dependency: a critical business function routed through a single person who cannot perform it during working hours.

Published estimates of how bad this is vary enormously. Industry call-tracking analyses report unanswered inbound rates anywhere from roughly a quarter of calls to well over half, with the higher figures concentrated in smaller owner-operated shops. The spread is that wide partly because the studies measure different things and partly because a lot of the published numbers come from companies selling the solution. Which is exactly why you should stop reading benchmarks and pull your own.

Work out your own number in ten minutes

Open your phone provider, your VoIP dashboard, or your call-tracking tool. Pull the last 30 days. You need five figures.

Then the arithmetic, which is deliberately simple:

Two notes on doing this honestly. First, the callback recovery figure is where owners flatter themselves. "I always call them back" usually means "I call back the ones I notice, at the end of the day, by which point the job is done and paid for by someone else." Second, this calculation counts only the first job. It does not count the service agreement that customer never signed, the replacement they will buy from whoever fixed it, or the neighbour they never referred to you. The real number is larger than what you are about to calculate.

A worked example

A three-truck residential HVAC company. These are plausible mid-range figures, not a claim about your business, and the point is the shape of the result rather than the specific total.

Average job value of $640, and a callback recovery rate of 25 percent, which is generous.

96 missed × 0.42 booking rate = about 40 jobs that would have booked. Times $640 is roughly $25,800 of demand that arrived and left. Recover a quarter of it on callbacks and you are still down about $19,300 in a month, or a bit over $230,000 a year, from a company whose owner would tell you their problem is that they need more leads.

That is the part worth sitting with. Nearly every home service business trying to fix a revenue problem starts by buying more leads. The leads they already paid for are ringing an unanswered phone.

What AI receptionists actually fix

The genuinely new thing in the trades over the last two years is that voice AI got good enough to hold a real conversation with a homeowner, and cheap enough that a two-truck shop can afford it. Five years ago the options were an in-house CSR you could not justify or a generic call center that took messages badly. That gap is where most of the losses lived, and it has largely closed.

What the current generation does well:

• It answers on the first ring, always. Sunday at 9pm, Tuesday at 7am, during the storm week when everyone calls at once. There is no queue and no bad day.
• It captures the essentials. Name, address, phone number, what is broken, how long it has been broken. Enough to call back with context instead of "hi, someone from this number called."
• It handles overflow properly. Most shops do not need every call answered by AI. They need the fourth simultaneous call answered instead of dumped to voicemail.
• It books the easy ones. Routine maintenance into an open slot, given a calendar and a set of rules, without anyone touching it.

Pricing has settled into a range that is affordable relative to the loss it prevents, and given the arithmetic above, the payback period for most shops is measured in weeks rather than months. That part of the vendor pitch is broadly fair.

And what they do not

Here is the part the vendors are quieter about, and it is the reason some shops install one and see very little change.

An AI receptionist cannot make a decision you have never made. It answers the call, takes the details, and then reaches the actual question: is this an emergency, does it go on today's board, whose route gets moved, what do we quote, do we service that neighbourhood. If the answers to those live in your head, the AI cannot book anything. It just produces a tidier queue of decisions waiting for you, and you have automated the easy half of the problem while leaving the expensive half exactly where it was.

Before you automate the phone, four rules need to exist in writing:

• What counts as an emergency. No heat below a stated outdoor temperature, active water, no power, gas smell. Name them.
• What can be moved. Which job types get bumped for an emergency, and which never do.
• Where you go and where you do not. Zip codes, drive-time limits, and the jobs you decline.
• What you quote over the phone. Diagnostic fee, after-hours rate, and the point past which someone has to look at it in person.

Four decisions. Write them down and your capture rate improves whether or not you ever buy software, because now a spouse, a part-time CSR, or an answering service can run the phone. That is the whole lesson: automation is delegation to a machine, and it fails for exactly the same reason human delegation fails. You handed over the task and kept the decision.

Three fixes, in order of cost

1. Measure it this week. Free. Pull the 30-day call log and run the arithmetic above. Most owners find the number is two to five times what they guessed, and the guess is what has been keeping it unmanaged.
2. Write the four rules. Free, about an hour. Emergency definition, what can be moved, service area, phone quoting. This alone lets someone other than you answer, and it is the prerequisite for everything else.
3. Then choose your coverage. Only now does the buying decision make sense: a part-time CSR, an answering service, an AI receptionist, or some combination. Whichever you pick, it inherits the rules you just wrote, which is why the order matters.

Skip to step three and you will have bought a more expensive version of the same bottleneck.

The home-services version of the free assessment includes a real-dollar estimate of hours lost to dead quotes and unanswered demand, alongside your dependency score. It takes about five minutes: take the home services assessment, or read more about how this shows up for trades.

Common questions

Do this week

• Pull the last 30 days of call data from your phone provider. Answered, missed, abandoned.
• Run the arithmetic: missed × booking rate × average job value × (1 − callback recovery).
• Write down what counts as an emergency, in one sentence, with a number in it.
• Write down which job types can be bumped and which cannot.
• Only then price out coverage: CSR, answering service, or AI. Hand it the rules you just wrote.