The true cost of a missed call, worked out properly

A missed call does not cost you one job. It costs a fraction of a customer, multiplied by how often it happens, multiplied by what a customer is worth over their whole relationship with you. Here is the arithmetic, with every assumption labelled, so you can run it on your own numbers.

Why the number is almost always wrong in both directions

Ask an owner what a missed call costs and you usually get one of two answers. Either "nothing, they call back", or the full value of an average job, as though every unanswered ring were a signed contract walking away. Both are wrong, and they are wrong in opposite directions.

The first answer ignores that a person with an urgent problem and a search results page open does not wait. They dial the next number. The second answer ignores that most inbound calls to a service business are not new business at all. They are existing customers rescheduling, suppliers, couriers, recruiters, and wrong numbers.

The honest figure sits between the two, and getting to it requires discounting twice before you apply any conversion rate. That is what the model below does. It is deliberately conservative, because a business case built on flattering assumptions collapses the first time someone checks it.

One note on what this article is and is not. Every number in the worked examples is an illustrative assumption, chosen to be plausible for that type of business. None of them is a measured industry statistic and none should be quoted as one. The purpose is the structure of the calculation, not the values.

The six inputs the model needs

1. Monthly inbound call volume

Total calls arriving at your main business number over a month, including out of hours. Pull ninety days from your phone system and divide by three, so a quiet August or a storm week does not distort the picture.

2. Miss rate

The share of those calls that never reach a human: rings out, abandoned in a queue, hits voicemail, or arrives when you are closed. Most cloud phone systems report this directly. Include the out of hours calls. They are the ones businesses most often exclude and they are frequently the largest single bucket.

3. New enquiry share

Of the missed calls, what fraction were potential new customers rather than existing clients, suppliers or noise. This is the discount that keeps the model honest. It varies enormously: a trades business advertising on Google will be high, an established clinic with a large patient list will be lower because so much inbound traffic is admin.

4. Leakage rate

Of those missed new enquiries, what share never come back at all. They do not leave a voicemail, they do not ring again, they call the next business on the list. This is the input that decides whether the whole number is large or trivial, and it is the input almost nobody measures. It is driven by urgency and by how many substitutes are one click away. A burst pipe at 22:00 leaks almost completely. A routine dental check-up leaks much less.

5. Enquiry to customer conversion rate

Of the enquiries you do answer, what share become paying customers or booked appointments. Take this from your own records. If you do not track it, that gap is worth closing before you model anything.

6. Average first job value and lifetime multiple

The revenue from the first transaction, and then a separate multiple representing repeat work and referrals over the relationship. Keep these apart. Blending them produces a single impressive number that nobody can defend in a meeting.

The model, in one table

Work down the rows. Each step narrows the previous one. The result at the bottom is monthly lost revenue, expressed twice: once at first job value and once including lifetime effects.

Step	Input	Calculation	Where to get it
1	Monthly inbound calls (C)	C	Phone system call detail records, 90 day average
2	Miss rate (M)	Missed calls = C × M	Unanswered, abandoned, voicemail and out of hours calls
3	New enquiry share (N)	Missed enquiries = C × M × N	Sample 50 missed numbers and check them against your customer list
4	Leakage rate (L)	Lost enquiries = C × M × N × L	Share of missed enquiry numbers that never appear again in your logs
5	Conversion rate (V)	Lost customers = C × M × N × L × V	Your own enquiry to booking or enquiry to job rate
6a	Average first job value (J)	Immediate lost revenue = lost customers × J	Accounts, average invoice for a first transaction
6b	Lifetime multiple (K)	Lifetime lost revenue = immediate × K	Average total revenue per customer divided by first job value

Two structural points are worth noticing. First, the model is multiplicative, so an error in any single input propagates through everything below it. That is an argument for measuring the two you can measure easily, volume and miss rate, and being conservative about the rest. Second, because it is multiplicative, halving your miss rate halves the final number exactly. There is no diminishing return hidden in the arithmetic, which is unusual and is the main reason answer rate is such a leveraged metric.

Worked example: a dental clinic

All values below are illustrative assumptions, not measured data.

Assume a two-surgery dental practice receiving 600 inbound calls per month. Assume a miss rate of 22%, which is plausible for a practice with one person on reception who is also greeting patients and handling payments. That gives 132 missed calls.

Assume 30% of those missed calls are genuine new patient enquiries, the rest being existing patients rescheduling, laboratories, and suppliers. That is 39.6 missed enquiries. Assume a leakage rate of 55%: dental demand is often non-urgent, so a slight majority give up or choose the next practice, but a meaningful minority persist. That leaves 21.8 lost enquiries.

Assume a 50% conversion from answered new patient enquiry to a booked first appointment, giving about 10.9 lost new patients per month. At an assumed first appointment value of EUR 180, that is roughly EUR 1,960 in immediate lost revenue per month.

Dentistry has strong retention, so assume a lifetime multiple of 6, reflecting recall appointments, hygiene visits and occasional restorative work. Lifetime value of the patients lost in that single month is therefore about EUR 11,760. Note carefully what that means: it is the future value of one month's worth of lost acquisition, not an annual figure and not a recurring one.

Clinics have a second, related leak that this model does not capture at all, which is patients who book and then do not attend. That is a separate problem with a separate mechanism, covered in our note on reducing no-shows with automated reminders.

Worked example: a plumbing firm

All values below are illustrative assumptions, not measured data.

Assume a six-van plumbing and heating firm receiving 900 inbound calls per month, with engineers on the road and calls handled by whoever is free. Assume a miss rate of 35%, which is high but entirely realistic when nobody owns the phone and a large share of calls arrive in the evening. That is 315 missed calls.

Assume 45% are genuine new enquiries, a higher share than the clinic because trades businesses attract more first-time, one-off demand. That is 141.8 missed enquiries.

Now the input that dominates this example. Assume a leakage rate of 70%. Emergency and semi-urgent trade work is the clearest case of substitutable demand there is: the caller has three more numbers on screen and a problem that is getting worse by the minute. That leaves 99.2 lost enquiries.

Assume a 35% conversion, lower than the clinic because trades enquiries include a great deal of price shopping. That is about 34.7 lost jobs per month. At an assumed average first job value of EUR 240, immediate lost revenue is roughly EUR 8,335 per month.

Assume a modest lifetime multiple of 2.5, covering repeat call-outs and the occasional boiler replacement. That puts the lifetime figure at about EUR 20,840. The plumbing case is the one where a simple change in answer rate produces the largest absolute swing, which is why it is the sector where the case is easiest to make. We wrote about the specific operational pattern in AI call answering for trades businesses.

Worked example: a law practice

All values below are illustrative assumptions, not measured data.

Assume a small litigation and family law practice receiving 220 inbound calls per month. Assume a miss rate of 18%, lower than the other two because solicitors' offices tend to have dedicated reception cover during business hours. That is 39.6 missed calls.

Assume 40% are prospective client enquiries, giving 15.8. Assume 50% leakage: legal enquiries are urgent enough that people move on, but the decision is considered enough that many will try again. That leaves 7.9 lost enquiries.

Assume a conversion rate of just 20%, which is deliberately low because legal intake filters heavily for jurisdiction, conflict, merit and ability to pay. That is about 1.6 lost matters per month. At an assumed average matter value of EUR 2,500, immediate lost revenue is roughly EUR 3,960 per month.

Assume a lifetime multiple of 1.4, mostly reflecting referrals rather than repeat instructions, since many private clients need a solicitor once. That gives about EUR 5,540.

The law example is instructive precisely because every intermediate quantity is small. Fewer than two lost matters a month sounds negligible right up to the point where you annualise it and find it is comparable to a junior salary. High-value, low-volume businesses have the most counterintuitive missed call economics, which we go into further in AI intake for law firms.

The three examples side by side

Every figure in this table is an illustrative assumption. Nothing here is a measured industry benchmark.

Input (all assumed)	Dental clinic	Plumbing firm	Law practice
Monthly inbound calls	600	900	220
Miss rate	22%	35%	18%
Missed calls	132	315	39.6
New enquiry share	30%	45%	40%
Missed enquiries	39.6	141.8	15.8
Leakage rate	55%	70%	50%
Lost enquiries	21.8	99.2	7.9
Conversion rate	50%	35%	20%
Lost customers per month	10.9	34.7	1.6
Average first job value	EUR 180	EUR 240	EUR 2,500
Immediate lost revenue per month	EUR 1,960	EUR 8,335	EUR 3,960
Lifetime multiple	6.0	2.5	1.4
Including lifetime value	EUR 11,760	EUR 20,840	EUR 5,540

Three different shapes of business, three different reasons the number ends up where it does. The clinic's cost is driven by retention. The plumber's is driven by volume and leakage. The solicitor's is driven almost entirely by the size of a single matter. If you only remember one thing from the comparison, make it this: the lever that matters most is different in each case, so copying another firm's conclusion is not a substitute for running your own inputs.

What this model deliberately leaves out

Being clear about the boundaries of a model is part of trusting its output.

It ignores the cost of the calls you do answer badly. A rushed answer between two patients, a promise to call back that nobody logs, a quote that is never followed up: these are real losses and they do not appear anywhere in the arithmetic above. In most businesses they are comparable in size to the missed call loss.

It ignores marketing waste. If you spend on advertising to generate calls and then miss a third of them, part of that spend bought nothing. The honest way to express it is that your effective cost per acquired customer is your nominal cost per lead divided by your answer rate, which is a considerably less flattering number than the one in most reporting dashboards.

It ignores reputation. Some share of people who cannot reach you form a view about your reliability and tell other people. That effect is real and we are not going to attach a number to it, because any number we invented would be exactly the kind of statistic this article is arguing against.

It also ignores the cost of the fix. Whatever you deploy, whether a person, an answering service or automated call handling, costs money, and the correct comparison is recovered revenue minus that cost. The model gives you the ceiling on what is available to recover, not the net benefit.

The one measured datapoint

Everything above is a model. Here is the one real result we can point to from our own work, and we will not dress it up beyond what it is.

At VEGNA Aesthetic Clinic in Amsterdam, the measured answer rate before deployment was 62%. After deployment it was 99%. The clinic attributes EUR 6,050 per month in recovered revenue to the change, and the no-show rate fell from 20% to 10%. The full detail is on the VEGNA results page.

What that datapoint supports is narrow and worth stating precisely. It shows that a 62% answer rate can exist in a well-run clinic without anyone experiencing it as a crisis, and that closing that gap released a five-figure quarterly sum in one practice. It does not tell you what your miss rate is, and it is one clinic in one city in one specialty. Use it to justify measuring your own numbers, not as a forecast of your result.

How to get your own real inputs

Volume and miss rate: one afternoon

Export ninety days of call detail records from your phone system. Almost every cloud provider offers this as a CSV. Count total inbound, then count the calls with no answered leg. Segment by hour of day and day of week. The out of hours block is usually the surprise.

Call records identify individuals and are therefore personal data under the GDPR, so treat the export accordingly: restrict access, define a retention period, and do not email the spreadsheet around. Our note on GDPR and AI customer service covers the same ground for automated handling.

New enquiry share: one hour of sampling

Take fifty missed numbers at random and check each against your customer database. The share that are not already customers is your N. Fifty is enough for a decision at this level of precision.

Leakage: the one worth real effort

For those same missed numbers, search your call logs for the following fourteen days. Did that number call again, or appear as a booking? The share that never reappear is your leakage rate. This is the only input in the model that genuinely requires work, and it is also the one with the largest influence on the result, so it is where the effort belongs.

Conversion and value: your accounts

Both should already exist. If your enquiry to customer conversion rate is not tracked anywhere, start there, because you cannot evaluate any front office change without it.

What to do once you have the number

The number is not the decision. The decision is a comparison between recovered revenue and the cost of recovering it, and there are several legitimate answers.

If your miss rate is concentrated in a predictable daily window, the answer may be a rota change, which costs nothing. If it is concentrated out of hours, the answer is coverage of some kind, and the options are an answering service, an on-call rota, or automated handling. If it is spread evenly through the day and driven by volume, you have a capacity problem and hiring is a genuine candidate, which we compare directly in an AI front office versus hiring a receptionist.

One discipline is worth insisting on. Whatever you change, measure the answer rate before and after using the same phone system report. Answer rate is the cleanest available proxy for the whole chain, it is measured automatically, and it cannot be argued with. If it does not move, the intervention did not work, whatever else the dashboard says.

Frequently asked questions

How do you calculate the cost of a missed call?

Multiply six numbers. Monthly inbound calls, times your miss rate, times the share of missed calls that are genuine new enquiries, times the share of those callers who do not come back, times your enquiry to customer conversion rate, times average first job value. That gives immediate lost revenue. Multiply again by a lifetime value multiple to include repeat work and referrals. Every input should come from your own phone logs and accounts, not from an industry average.

Do people leave a voicemail when a business does not answer?

Many do not, and that behaviour is the reason missed calls are underestimated. A caller with an urgent or comparison-shopping intent typically moves to the next result rather than waiting for a call back. Voicemail is left more often by existing customers than by first-time enquirers. This is why the leakage rate in the model matters more than most owners expect, and why it should be measured rather than assumed.

What is a realistic call answer rate for a small business?

It depends entirely on staffing and call pattern, so treat any single benchmark with suspicion. What is measurable is your own rate: most VoIP and cloud phone systems report answered, abandoned and out of hours calls. At VEGNA Aesthetic Clinic in Amsterdam the measured answer rate was 62% before deployment and 99% after, which gives a sense of the size of the gap that can exist without anyone noticing it.

Should I include lifetime value in a missed call calculation?

Include it, but separately and conservatively. Report the immediate first job figure and the lifetime figure as two numbers, never blended into one. Lifetime multiples are estimates with wide error bars, and a business case that depends on an aggressive multiple is fragile. If the decision only works at a high multiple, it probably does not work.

Where do I find my actual miss rate?

Your phone system. Cloud and VoIP providers expose call detail records showing answered, abandoned, voicemail and out of hours calls, usually exportable as CSV. Pull ninety days rather than one week so seasonality and staff absence average out. Call records identify individuals, so they are personal data under the GDPR and should be handled on that basis.

Is a missed call always a lost customer?

No, and any model that assumes so is dishonest. A large share of inbound calls are existing customers, suppliers, recruiters and wrong numbers, and many genuine enquirers do call back. That is exactly why the model applies two discount steps before conversion: the share of missed calls that are real new enquiries, and the share of those who never return.