Count your missed calls for one week before you decide anything

A window and door fitter with eight staff runs its office from a converted garage. The owner's partner does the phones on Monday, Wednesday and Friday mornings. The rest of the week the landline rings into a voicemail box that was last emptied in spring. Asked how many calls the firm misses, the owner says "a few, maybe". Asked how many quotes it sent last month, the owner has the number instantly, because quotes live in the software and missed calls live nowhere.

That gap in knowledge is the problem. Every decision about the phone, from a cheap auto-reply to a full AI receptionist, depends on a number that almost nobody has. The good news is that the number takes one week to collect and costs nothing.

Why you cannot guess it

Two figures get quoted for how many calls businesses miss. Both come from the US, and they measure different things. Invoca, a call-tracking provider, wrote in February 2024 that about 26 percent of calls never get answered, going by the data on its own customer platform. That covers Invoca's customers in general, not trade firms. 411 Locals, an American SEO firm, monitored the calls of 85 small businesses in 58 industries for 30 days and published in January 2016 that 62 percent of them did not reach a person. Calls that went to voicemail count in that figure, and the sample is small.

Neither number tells you where your own firm stands. That depends on things no study can see, such as whether anyone is in the office at lunch and whether the mobile forwarding works in the workshop. So you count.

The one-week audit

Pick a normal week. Not the week after a trade fair, not the week the owner is on holiday. Then collect three things.

The missed-call log. Most office phone systems and every mobile show a list of calls that rang out, with time and number. Write them down each evening, or photograph the screen. If calls forward to a mobile, check that log too, and note which calls went to voicemail rather than being answered.

The voicemail box. Count the messages left. Then count the missed calls that left no message. That second number is the important one, because those callers are the ones who moved on to the next firm in the list.

The timestamps. For every missed call, note the hour. By Friday you will see a shape. In most trade firms the missed calls cluster at lunchtime, between five and seven in the evening, and on Saturday morning. Those are the hours when a homeowner has finally got round to calling and nobody is at the desk.

Strip out what is not a customer. Suppliers, sales calls, the bank, the number that rings every day at 9:04 and hangs up. Be honest about it; an owner who wants the number to be small will find reasons to discount half the list. What remains is people who wanted to buy something.

Turning the count into a number you can act on

You now have missed customer calls per week. Two more figures from your own books turn that into money.

The first is your quote conversion rate on first contact: of the people who phone about a job, how many end up as an order. Most owners know this to within ten points. The second is the average value of an order. Multiply the three together and you have the revenue that rang out per week. Multiply by 45 working weeks for the year. The calculator does this arithmetic with your own figures in two minutes, and it also lets you vary the conversion rate to see how sensitive the result is.

I deliberately give no example amount here, because it swings by an order of magnitude between trades. A window fitter with a four-figure average order and a cleaning contractor on monthly contracts get very different answers from the same number of missed calls.

What the sum leaves out is the second order. The homeowner who wanted new windows and reached voicemail calls someone else. That firm also gets the front door in two years and the recommendation to the neighbour. A missed first call is a missed customer, not a missed job.

Which fix fits which result

The count tells you which of three fixes to reach for. They are not exclusive, and most firms end up with two.

Under ten missed customer calls a week. An SMS auto-reply plus forwarding to a mobile will usually do. Anyone who rings and gets no answer receives a text within seconds: "Hello, this is Muster Windows. We are on site at the moment. What can we help with? We will get back to you shortly." Many phone systems and mobile apps can send this. The guide to missed call text back covers the set-up and what the text should say. The limits are that someone still has to read the replies and make the call back, and that landline callers get nothing. To forward only the calls nobody picks up, see the guide to conditional call forwarding. A single reply to a call attempt, without an offer in it, is not advertising under German unfair-competition law as I understand it; check with a lawyer if you are unsure.

Ten to thirty a week, clustered at lunch and after five. Forwarding on its own stops working here, because the person the calls forward to is on a ladder. The audit will show that a large share of your missed calls fall in two or three fixed windows. An AI receptionist that picks up after ten seconds during those windows, takes the enquiry and books a slot from the real calendar closes most of the gap without changing anything about how the office works the rest of the day.

Over thirty a week, or a business that trades on emergencies. At this volume the owner is the bottleneck, and every evening callback is half an hour taken from quoting or from family. The receptionist takes over first contact entirely, and the office only sees the calls that need a decision. Since 2 August 2026, Article 50 of the EU AI Act requires the provider of an AI phone assistant to build it so that callers learn they are speaking to a machine, at the latest when the conversation starts, and the duty has no exemption for small providers. Check that the greeting says so before you switch it on. In practice that sentence costs nothing and callers stay on the line more readily when they know what they are dealing with.

The AI receptionist page describes what the assistant says, what it hands to a person, and how it fits with the rest of the sequence. It is the fifth employee we install, not the first, because the employees before it deliberately produce more calls and this one is there to catch them. If you are weighing it against a human answering service, AI receptionist vs answering service compares how each works and how each is priced.

What the window fitter found

The window fitter counted for one week in September. Twenty-three missed calls, of which four were suppliers and two were the same sales call. Seventeen customers. Eleven of the seventeen rang between 12 and 2 or after 5. Three left a voicemail. Fourteen did not.

The owner had the conversion rate and the average order to hand. The yearly figure was larger than the firm's entire marketing budget, and the fix for the lunchtime and evening windows cost less per month than one of those missed orders. The decision took less time than the count.

Do the count before you talk to anyone selling a solution, including us. It is the only number in this discussion that comes from your business rather than from a study, and it is the one that settles the argument.

The phone is one part of the week. Our examples of AI for tradespeople cover quotes, scheduling, invoices and reviews, with what the AI does in each and what the owner still decides.