Size Your Missed-Call Gap in One Afternoon

There is a specific conversation that happens in practices about twice a year. Someone says the phone is a problem. Someone else says it is not that bad. Nobody has a number, so the argument is settled by whoever sounds most certain, and six months later it happens again.

This guide ends that. It takes one month of your own call records and turns them into a range you can put in front of a partner, a spouse, or a vendor, with every assumption written down beside it. It takes an afternoon. It needs no new software, and it does not need anyone's permission.

The honest warning up front: a meaningful minority of practices do this and find the number is small. That is a good outcome. It costs you an afternoon and saves you a subscription, and it is the reason to measure rather than to guess in either direction.

What you are actually measuring#

The thing you want is not a count of missed calls. It is an estimate of revenue that did not arrive because a call was not answered, which is a different and smaller quantity. Most of the alarming figures in this category quietly conflate the two, and the gap between them is where credibility lives.

Four numbers get you there. How many calls went unanswered, how many of those were distinct people rather than the same person trying three times, what share of a first-time caller becomes a patient, and what a patient is worth. Multiply the four, present it as a range, and you have something defensible.

Everything below is one of those four numbers or a correction to one. The corrections matter more than the multiplication does, because the multiplication is arithmetic and the corrections are where an honest estimate separates from a scary one.

The order matters too. Each lesson makes the number smaller and the number harder to argue with, which feels like the wrong direction while you are doing it and is the entire point. A figure that survives 9 deliberate attempts to shrink it is a figure you can take to someone who does not want to hear it.

A count of missed calls is not a revenue figure. Treating it as one is the single most common error in this category.

01Pull one month of call detail records#

Every phone system worth having can export call detail records. What you are looking for is one row per call with a timestamp, a direction, a duration, and a disposition. The names differ by vendor and the data does not.

1. Log into your phone system, VoIP dashboard, or carrier portal as an admin.
2. Find the section called Call Detail Records, Call History, Call Logs, or Analytics.
3. Set the range to one full calendar month, ideally the most recent complete one.
4. Filter to inbound only. Outbound calls are a different question and will distort everything downstream.
5. Export to CSV and open it in a spreadsheet.

Pick a month that was ordinary. A month containing a holiday closure, a system migration, or two weeks of staff sickness will give you a number, but it will be a number about that month rather than about your practice.

If your system genuinely cannot export, you can do this by hand from a month of on-screen logs in about 1 hour, and it is still worth doing. What you cannot do is skip to the multiplication with a figure someone remembers.

Two data-quality checks are worth 5 minutes before you go further. First, confirm the row count roughly matches what your system reports as total inbound volume for the month, because a filtered export that quietly capped at 1,000 rows will understate everything downstream. Second, check that the timestamps are in your local time zone rather than UTC, since a 5-hour or 6-hour offset will move a large block of evening calls into the staffed bucket and make your gap look far smaller than it is.

That second one is not hypothetical. Time-zone drift in exported call records is common enough that it is worth spot-checking 3 or 4 rows against calls you remember taking.

02Split the month into staffed and unstaffed hours#

The single most useful cut in the whole exercise is the one almost no phone report gives you by default: whether anybody was in the building when the call arrived. Add one column and sort on it.

Build a small table of your actual opening hours, then classify every row against it:

The reason to separate the middle row is that it is the cheapest thing on this list to fix, and a practice that discovers most of its misses happen between 12:30 and 1:30 has learned something more useful than a total. Fix the rota before buying anything.

A standard schedule of 8 hours a day, 5 days a week covers 40 hours out of the 168 hours in a week, leaving 128 hours uncovered. That ratio is structural rather than a scheduling failure, and the reasons it behaves so differently from the staffed hours are worked through in why after-hours calls are different.

Fix the rota gaps before you price anything. They are the only bucket on this list that costs nothing to close.

Sort the unstaffed bucket by hour of day before you leave this lesson. Most practices find their evening volume concentrated in a narrow window, often the 2 hours after close rather than spread evenly across the night, and that shape matters later when you are deciding what coverage is actually worth buying. A gap that is really 10 hours a week of concentrated demand is a different proposition from 128 hours of thin, even trickle.

03Separate a missed call from an abandoned one#

Your phone system's "missed" column is doing more work than it admits. At least four different events usually hide inside it, and only some of them represent lost demand.

Drop the last row entirely and count voicemails separately, because a voicemail is a task rather than a loss. Whether it becomes a loss depends on your callback discipline, which is worth measuring on its own but is not what this afternoon is for.

The abandoned-in-queue row deserves a second look before you move on. If a large share of your misses sit there, the problem is not that nobody was in the building, it is that the people in the building could not get to the phone fast enough. That is a capacity or a routing problem, and covering your evenings will do nothing for it.

A practical threshold: if more than 30% of your misses are abandoned-in-queue during staffed hours, stop this exercise and go look at your hold times first. You will learn more in 20 minutes there than in the remaining 5 lessons here.

A voicemail is not a missed opportunity. It is an opportunity with a deadline attached.

What remains, rang-out plus abandoned, is your real missed-call count for the month. In most practices this is materially smaller than the number the phone system reports, and being able to say so is what makes the final figure survive scrutiny.

04Find the repeat callers and collapse them#

This is the correction that separates an honest estimate from a vendor stat sheet, and it is also the one most often skipped, because it always makes the number smaller.

1. Sort your remaining missed calls by caller number.
2. Group calls from the same number within a 48-hour window into one attempt.
3. Count the distinct numbers rather than the rows.
4. Note the ratio of rows to distinct numbers, and write it down.

That ratio is worth keeping. A practice where 300 missed calls collapse into 180 distinct people has a very different problem from one where 300 collapse into 290. The first has a capacity problem during busy periods. The second has a coverage problem, and persistent callers are not what is failing.

There is a subtlety here that cuts the other way. A caller who tried 3 times and then stopped is showing more intent than one who tried once, not less, so a high repeat ratio does not mean the misses matter less. It means fewer people are affected, more severely.

The 48 hours window is a judgement call and you can defend a different one. A caller ringing back 5 days later about the same thing is arguably the same opportunity, and a practice with a long consideration cycle might reasonably use 7 days. What matters is that you pick one, write it down, and apply it consistently, because a ratio computed with a moving window is not a ratio.

One thing worth noting while you have the data sorted: how many distinct numbers appear only in the missed set and never in the answered set across the whole month. Those are the callers you never spoke to at all, and in most practices that count is smaller than the raw missed figure by a wide margin.

05Count what a first appointment is worth#

You need a per-patient value, and there are two defensible ways to get one. Pick whichever your records support, and say which one you used.

The first-visit method is conservative and easy to defend. The first-year method is more realistic for practices where the first appointment is a consultation that leads to a course of treatment, which describes most of aesthetics and a good deal of dentistry.

Do not use lifetime value here. It is a real concept and it is the wrong tool for this job, because it imports assumptions about retention over several years into a question about one month of phone calls. If you want a number that survives an argument, keep the time horizon short.

Whichever you pick, use your own figure. A national average for your specialty tells you about a distribution you may not sit anywhere near, and a number you pulled from your own practice management system cannot be argued with in the way an industry benchmark can.

There is one adjustment worth making if your practice has a wide service mix. A first-time caller is not a random draw from your existing patient base, because the treatments people phone about are not the treatments people book once they are already with you. If 60% of your revenue comes from returning patients on a recall cycle, a value averaged across all patients will overstate what a new inbound caller is worth on day 1.

The fix is to compute the value from new patients only, in their first 30 days or their first 12 months depending on which method you chose. It usually moves the number down, and moving it down is the point.

06Estimate the conversion rate you actually have#

This is the softest of the four numbers and the one to be most careful with, because it is where an estimate turns into a claim. You are asking what share of a first-time caller would have become a patient if the call had been answered.

The best available proxy is your own answered-call conversion. If you can pull the number of new patients booked in the month and the number of distinct first-time callers who got through, the ratio is a reasonable starting point.

1. Count new patients booked in the month from your practice management system.
2. Count distinct answered first-time inbound callers for the same month.
3. Divide the first by the second.
4. Then discount it, for the reason below.

The discount matters. A caller who got through and a caller who did not are not the same population. The one who got through reached you at a moment when you were available, which correlates with when you are staffed, which correlates with the kind of call it was. Applying your answered rate to your missed calls without a haircut assumes those populations are identical, and they are not.

A reasonable practice is to run the calculation at your measured rate, at two thirds of it, and at one third, and present all three. That is the range, and it is more useful than a single confident number.

07Sanity check the two soft inputs#

Before you multiply anything, spend 10 minutes trying to break your own two soft numbers. This is the lesson that stops the afternoon producing a confident figure built on a bad assumption nobody looked at twice.

1. Take your distinct missed-caller count and ask whether it is plausible as people, not rows. If it implies 12 new strangers a day phoning a single-doctor practice, something upstream is wrong.
2. Take your patient value and multiply it by your actual new patients for the month. If that product is nowhere near your actual new-patient revenue, your value figure is wrong.
3. Take your conversion rate and ask what it implies. A rate above 60% on first-time inbound callers is unusual and worth re-deriving before you rely on it.
4. Write down anything that failed a check, and fix the input rather than proceeding around it.

The second check is the one that catches most errors, because it is a closed loop: the same two numbers have to reconcile against a third you already know. If they do not, you have found a definition mismatch, usually a new-patient count that includes returning patients under a new insurer or a value that quietly includes product sales.

If your inputs cannot reproduce a number you already know, they will not produce a number you can defend.

08Put the four numbers together#

The arithmetic is trivial and takes one line. Distinct missed callers, times conversion rate, times value per patient, equals monthly gap. Do it 3 times, at the 3 conversion rates from lesson 6.

1. Multiply distinct missed callers by your measured conversion rate, then by patient value. Call this the high case.
2. Repeat at two thirds of the conversion rate. Call this the middle case.
3. Repeat at one third. Call this the low case.
4. Multiply each by 12 for an annual view, and note that you extrapolated from one month.

Present the middle case as the headline and always show the low case beside it. A range with a stated method beats a point estimate with a hidden one, and anyone senior enough to approve spending will trust the range more precisely because it is less dramatic.

The ROI calculator runs this same arithmetic if you would rather not build the spreadsheet, but do the lessons above first. The calculator cannot tell you whether your repeat-caller ratio is 1.7 or 1.03, and that input changes the answer more than anything else on the page.

09Write down the range and the assumptions#

An estimate without its assumptions is a rumour with a decimal point. The last lesson is the one that makes the previous seven durable: write the whole thing down on one page, in a form someone else can check.

• Which month, and whether it was ordinary
• Total inbound calls, and the count after removing sub-10-second rows
• Missed calls split into staffed, gaps, and unstaffed
• Rows-to-distinct-callers ratio
• Patient value, and which of the two methods produced it
• Measured conversion rate, and the discounted rates used
• The three-case range, monthly and annual
• The date you did this

The last line is not filler. This number ages. A practice that adds a location, changes its hours, starts running ads, or loses a receptionist has a different gap from the one it measured, and the dated page is what tells you when to run the afternoon again.

Six months is a reasonable re-measurement interval for a stable practice, and immediately is the right interval after any change to hours or staffing. What you are building is a baseline, and a baseline nobody dated is not a baseline.

What this number is not#

It is an estimate with two soft inputs, and it should be described that way to anyone you show it to. The conversion rate is a proxy, and the extrapolation from one month to a year assumes twelve months that look like the one you measured. Neither assumption is unreasonable and neither is a measurement.

It is also not, on its own, a case for buying anything. The gap is one side of a comparison whose other side is what covering it costs, and that side has four components of its own, worked through in the real cost of a missed call.

What the number does is convert a recurring argument into a shared fact. Whichever way it comes out, the practice now has one page that says what the phone is costing and how that was worked out, and the next person who has an opinion about it has something to argue with other than a feeling.

It is worth saying plainly that the low case is a legitimate answer. A single-location practice with 20 missed calls a month, most of them repeats from 8 people, and a conservative patient value is looking at a gap that no product in this category can economically close. Finding that out in 3 hours is a good use of 3 hours.

The practices for whom this matters tend to look different in a specific way: the misses concentrate in the unstaffed bucket, the repeat ratio is near 1.0 because those callers only try once, and the patient value is high enough that a handful of recovered callers a month covers any plausible cost. If your page says that, the next question is what covering it would take, and that is a separate exercise with its own numbers.