An audit sheet for your own inbound phone line
Most businesses cannot answer how many calls they missed last month, or how long the people who hung up actually waited, about their own phone line at all. That is not a knock on any one office -- a dialler report usually shows a single answer-rate number and stops, which hides every one of the questions that actually decide whether the number means anything. This sheet is the walk-through: what to pull from your own reports, the definitional trap each measurement hides, and what our own floor found on the identical question, so you can see the shape of a complete answer before you go looking for one.
Run it against your own call reports, not ours. Every line below names the measurement to take, not a target to hit -- our own numbers are printed so you have something to compare the shape of your answer against, including the lines where our own number is the unflattering one.
No signup, no email wall, nothing to fill in first. The CSV opens straight into a spreadsheet with a column for the number you pulled from your own reports.
How to use it
Work down it against whatever your dialler or phone system can export -- a call detail record, a queue report, a carrier CDR. Most of these lines need nothing fancier than filtering and counting rows; none of them needs a data warehouse.
- Write down the denominator with the number. A rate with no n attached cannot be checked later, including by you.
- Pull the raw counts before the percentage. A spreadsheet formula hides which population it divided; the counts do not.
- Expect at least one unflattering answer. Our own floor got one on almost every line here -- see the number next to each measurement.
There is no scoring on this sheet. A composite score would hide which measurement moved it, and the whole point is to be able to point at one line and say which question your own line has not answered yet.
What our own floor measured this against
Every first-party figure below comes from one floor's own inbound calls over a fixed window, read straight off the dialler's own tables. Nothing here is an industry benchmark and nothing refreshes automatically -- it is one system, measured, so you can see what a complete answer looks like before you build your own.
Two of the lines below also appear on the vendor due-diligence sheet, because the same measurement matters to both readers. The documents differ in who is asking: that sheet is questions you put to somebody else before you sign; this one is measurements you take on a system you already run.
What "answered" actually contains
Three measurements, and they come first because a manager who only counts answered calls has not yet asked what those calls were. An answered call is not automatically a value metric -- it is a call somebody picked up, and picking up is the cheap half of the question. Run these three against your own reporting period before trusting any answer rate that follows.
Read this before you use the section
These describe one floor of 89,858 inbound arrivals over the 90 days from 2026-05-03 to 2026-08-01, at a mean answered-call length of 23.8 seconds. They are DESCRIPTIVE of what one line produced, never a target: a low hangup share does not itself mean a healthy line, and a high one does not itself mean a broken one. Read them beside your own composition, not as a bar to clear.
The two ways to count "answered"
Three measurements about what a queue report can and cannot see. A rate is a fraction, and an inbound floor keeps two different bottoms for it depending on which table produced the report -- every call that arrived, or only the calls that reached a queue. The gap between them is where calls go missing from a report that looks complete.
-
Run your answer rate twice -- once against every call that arrived, once against only the calls that reached a queue -- and check how far apart the two numbers sit.
54.3% of inbound calls reached a live agent (48,761 of 89,858 that arrived); 61.3% of the calls that reached a queue did (48,761 of 79,553).
What your own line got
-
Find the calls that were routed toward an agent queue on your own line and left no queue record at all.
6,529 in-group-routed calls left no queue record, invisible to any figure computed on the queue log alone.
What your own line got
-
Find the calls that never entered a queue on your own line because they were routed straight to an extension.
3,776 arrivals (4.2%) were routed to an extension and never entered a queue.
What your own line got
Read this before you use the section
Publish both denominators or neither: the queue-record rate flatters because it silently drops calls that arrived and never reached a queue at all, and the arrival rate is the number a caller actually experiences. Check both on your own line before quoting either one alone.
Whether the pooled figure is the number you actually get
Two measurements about dispersion. A single pooled answer rate across every line a floor runs can be dominated by whichever line carries the most volume, and averaging in a line nobody staffs is a different mistake that moves the average in the other direction. Both have to be checked before a pooled number means anything about any one line.
-
Do not read a single pooled answer rate as what a typical line on your own floor gets -- check the median against the pooled figure.
Across the five staffed in-groups on our own floor the median answer rate is 49.2%, against a pooled 61.3% carried by the two largest groups (84.9% of the volume).
What your own line got
-
Before you average every queue on your own line, check which of them ever had an agent assigned at all.
Two of the seven in-groups measured here had zero agents assigned across the entire window; folding them into the average drags the all-seven median to 11.5%, which is a category error rather than a finding.
What your own line got
Read this before you use the section
A pooled rate and a median answer different questions, and the two largest groups on our own floor carry most of the volume behind the pooled figure. Check the median across your own staffed lines separately from any single blended number, and confirm which lines in the average actually had somebody assigned to them.
What happens to the calls nobody answers
Four measurements about hold time and about the calls that never reach an agent. A median wait describes only the calls that survived to be answered, and reading it as a service level says nothing about the calls that did not. Check what actually happens to the rest before publishing a wait time to anyone.
-
Check whether the calls your own line does answer actually wait, or connect the instant an agent is free.
43,926 of 48,761 answered inbound calls (90.1%) connected to an agent with no measurable hold at all.
What your own line got
-
For the calls nobody answered, check how long they actually rang before ending -- and whether your own line queues them or simply drops them.
Of the unanswered calls carrying a recorded wait, 50.1% ended within five seconds and only 38 in the entire quarter waited 30 seconds or more, against a dialer status literally called "Agent Not Available".
What your own line got
-
Read what your own dialler calls the no-agent event, in its own words, rather than a paraphrase.
Our own dialler names the state itself rather than leaving it blank: "Agent Not Available".
What your own line got
-
Remember that every wait figure you compute starts when the call reaches your switch, not when the caller dialled.
Carrier ring time before a call arrives at our own number is invisible to every instrument we own; every wait figure here starts at arrival, not at dial.
What your own line got
Read this before you use the section
A median wait describes the answered share and nothing else. It is published here beside the answer rate for that reason -- a fast median on a low answer rate is survivorship, not performance -- and it should never travel alone on your own reporting either.
Coverage split by window, not one after-hours share
Three measurements about when calls arrive against when they get answered. A single after-hours percentage answers a demand question, not a coverage one, and computing it in the wrong clock changes the number entirely. Split coverage by window before deciding whether "we cover after hours" is actually true.
-
Do not let the demand share stand in for coverage -- check the answer rate separately for each after-hours window on your own line.
We answer 59.8% of weekday-evening calls, 10.1% of weekend calls and 0.3% of calls before 08:00.
What your own line got
Read this before you use the section
Compute an after-hours share in the CALLER own local time, not the floor local time -- the two give different answers -- and keep any unresolved rows visible rather than dropping them, since dropping them silently narrows the bound. Demand and coverage are different questions and neither licenses the other.
The check nobody runs on their own line
One measurement, and it is the one most reports never take at all. Before an inbound count is read as organic demand, it is worth checking how much of it is a line dialling itself.
-
Before you read your own inbound volume as organic demand, check how much of it is your own outbound coming back.
At least 31.2% of our own inbound is our own outbound coming back -- in the week of 2026-07-25, 2,871 of 9,210 inbound calls came from a lead we had already dialled, and that is a floor, not a ceiling.
What your own line got
Read this before you use the section
This is a floor, never a ceiling, for a mechanical reason: the lookback window that finds a match is short, so a longer lookback can only find more of them, never fewer. Treat any number here as understating the true share on your own line.
What this sheet cannot do for you
It cannot see carrier ring time before a call reaches your own switch -- every wait figure any instrument produces starts at arrival, not at dial, and that is as true of your reports as it is of ours. It cannot tell you what a call was about, only how it was dispositioned, because most inbound paths carry no topic or outcome field beyond the status an agent selects. And it cannot compare your line to an industry norm, because there is no second floor in this dataset -- only what one floor of calls actually did.
It is also a snapshot. Our own figures come from a fixed window that nothing refreshes, and your own reports will drift the same way. Re-run the sheet on a fresh export rather than trusting a number you pulled once, and re-read the definitional trap on each line -- it is usually the reason two people arguing about the same phone line are actually arguing about two different fractions.
Want somebody else to run the sheet against your reports?
Send us an export and we will walk it against every line on this sheet, including the ones where the honest answer is not going to be flattering.
If you are still sizing whether a missed call is worth fixing at all, the inbound coverage calculator and the inbound answer-rate benchmark answer the arithmetic and publish the research this sheet is the instrument for.
Or skip the spreadsheet
Send us a week of your own call reports and we will run every line on this sheet against them, including the ones most vendors would rather not measure.
Book a Free Consultation