Why the Monthly Report Looks Fine
Most phone reporting is built from averages, and an average is arithmetic dominated by the majority of easy calls. The customers you are failing are a minority — which means they are, by construction, invisible in every mean you calculate.
A worked example. Five hundred calls last week. Four hundred and fifty answered almost immediately. Fifty waited more than four minutes, and thirty of those hung up.
| The report says | Which sounds | But actually |
|---|---|---|
| Average wait: 40 seconds | Fine | Describes nobody. 450 people waited seconds; 50 waited minutes |
| 94% of calls answered | Excellent | 30 people gave up — probably the most commercially significant fact of the week |
| Average call length: steady | Reassuring | Says nothing whatever about whether anything got resolved |
Stop asking “how did the phones perform?” Start asking “how many people had a bad experience, and when?” The first question has an answer you cannot act on. The second one names a Tuesday and an hour.
Question 1: Who Gave Up, and When?
Ask for: the number of abandoned calls, and a breakdown of how long each waited before hanging up.
An abandoned call is someone who wanted to reach you badly enough to ring, waited, and gave up. In a sales context that is revenue gone. In service it is a customer more likely to leave, more likely to complain publicly, and quite likely to ring again — so you carry the cost twice.
ACXPA's 2026 self-reported Australian figures put voice abandonment at an average of 9% and a median of 5%. The gap matters: a minority of operations with very poor abandonment are pulling the average up, so if you are above nine per cent you are not in the middle of the pack, you are in the tail of it.
Abandons under 30 seconds
A design problem. The menu is too deep, the greeting too long, or there is no sign a human exists. People are leaving before they ever reach a queue. Fix the front door, not the roster.
Abandons at 1–3 minutes
An expectation problem. People will wait if told what they are waiting for. Position announcements, an estimated wait, or a callback offer all move this number without adding a single person.
Abandons past 3 minutes
A staffing problem, in specific intervals. No menu redesign touches this. Either more people in those windows, or a callback so nobody has to hold at all.
Check two things before comparing yourself to anything
Many systems exclude abandons under about five or ten seconds on the basis that they are misdials. That is reasonable, but it changes the number materially, so a benchmark comparison is meaningless unless both sides count the same way. And check whether a caller who accepts a callback offer and hangs up is being logged as an abandon — that records a success as a failure and can make a genuinely good change look like a regression.
Question 2: What Did the Worst Tenth Experience?
Ask for: the 90th percentile wait time. The sentence you want completed is "ninety per cent of callers waited less than ___".
This single number replaces the average as your experience measure, and it is the one that predicts complaints. Your worst-served tenth is where escalations, negative reviews and churn come from — and they are exactly the population that an average is built to hide.
| Average wait | 90th percentile | What you actually have |
|---|---|---|
| 40 seconds | 90 seconds | A consistent queue. Everyone gets a similar experience. Healthy |
| 40 seconds | 6 minutes | A queue that is fine most of the time and badly broken for a tenth of your callers. Same average, completely different business problem |
| 90 seconds | 2 minutes | Uniformly a bit slow. Easier to fix than the row above, and much less damaging |
For context on where Australian sectors sit, ACXPA's 2026 Best Practice Report showed self-reported speed of answer varying enormously by industry — utilities reported the slowest at a 227-second average, while banking and finance carried the highest median at 79 seconds. Two things follow. Sector benchmarks are context rather than targets. And where a sector's average and median diverge sharply, the distribution is skewed — which is precisely the situation percentiles exist to describe.
Question 3: How Did We Do by Half-Hour?
Ask for: calls offered, answered and abandoned, in half-hour intervals, for a normal week.
This is the question that most often changes someone's mind, because the answer is nearly always the same and nearly always a surprise.
What you will almost certainly find
Performance is excellent for most of the day and collapses in two windows: the first hour of trading, and the half hour or so after lunch. Those two intervals contain most of your abandoned calls and most of your long waits — and a daily or monthly figure averages them into invisibility. Every unhappy caller you have is concentrated in about ninety minutes of the day.
Then look at what your team is doing in those two windows. In most businesses the answer is: the morning meeting, and the post-lunch handover. The two intervals where the queue fails are the two intervals when the fewest people are on the phones, and nobody has ever connected the two facts because nobody had the interval report.
The cheapest improvement available to any business with a queue
Move meetings and staggered breaks out of your two worst intervals. No headcount, no spend, no new system. It routinely produces a bigger improvement in answered calls than anything else on this page, and it can be done next Monday.
One more cut worth asking for: volume by half-hour broken down by weekday. Monday is not Wednesday. A roster built on a weekly average under-staffs Monday morning every single week, permanently, and it will never show up in a report that averages five days together.
Question 4: Who Had to Ring Us Twice?
Ask for: the number of callers who contacted you more than once within seven days.
This is the cheapest available proxy for whether problems are actually being solved. Everything else measures whether calls were answered. This measures whether they were finished, which is the thing customers care about and the thing that drives your cost base.
For a sense of scale, ACXPA's Australian Call Centre Rankings put banking at roughly 32% first contact resolution in the first quarter of 2026 — meaning around two-thirds of contacts were not resolved first time. If you have never measured your own, assume it is worse than you think, because everyone does.
| What the repeat-contact number tells you | Why it beats the alternatives |
|---|---|
| It is objective and automatable | Asking agents to self-report resolution is systematically optimistic — people are poor judges of whether they solved someone else's problem |
| It is consistent over time | Survey-based measurement has a low response rate and skews to the very happy and the very angry, so the trend is unreliable |
| It breaks down by reason | A poor overall number is nearly always two or three specific issue types, not a general failing. That makes it fixable |
Its blind spot is real and worth stating: it cannot see the customer who gave up rather than ringing back. Pair it with question 1, which does see them. Producing the reason breakdown used to be expensive; automatic transcription and summarisation have made it much cheaper, and it is the step that turns a percentage into a list of things to fix.
Question 5: How Hard Is the Team Actually Working?
Ask for: occupancy — the share of logged-in time spent handling contacts — and your shrinkage assumption.
Occupancy looks like a productivity measure, so the instinct is to push it up. It behaves more like a physical limit. Sustained operation above roughly 85% produces rising errors, dropping courtesy, more sick leave and eventually attrition. And because call arrival is random rather than smooth, a team at very high occupancy has no absorption capacity — a modest volume spike collapses the queue instead of being soaked up.
Under 60%
Genuinely overstaffed for the volume, or the work is arriving in a very uneven shape that better interval rostering would smooth.
70–85%
The workable band. Enough slack to absorb a spike, enough load that people are not idle.
Above 85% routinely
Not efficiency. This is where quality falls and people leave, and it is usually a planning error rather than a deliberate choice.
Which leads to shrinkage — the proportion of paid time that is genuinely unavailable for calls: breaks, training, meetings, leave, coaching, system time. It is commonly 30 to 35% and is routinely planned at twenty. That single underestimate is the most frequent cause of a queue that is inexplicably short every afternoon despite the roster looking correct on paper.
If Your System Cannot Answer These
That is not a failure of the exercise. It is the most useful result it can produce, because it tells you where the actual limitation sits.
| Response you get | What it means |
|---|---|
| “We can only see total calls” | You are managing a queue with no instrumentation. Everything above is guesswork, including any confidence that things are fine |
| “We get a monthly summary” | Monthly averages cannot show interval failure, which is where nearly all bad experiences live. The report is real and structurally unable to find your problem |
| “The provider can run a report for us” | If a manager cannot pull it themselves in under a minute, it will be pulled twice and then never again. Self-service is the difference between a metric and a habit |
| “We don't record abandoned calls” | The single most commercially important number is not being captured. This is worth changing on its own |
| “What's a percentile?” | Fair enough — but percentile reporting is standard capability in a current platform, and managing a distribution using only its mean is a genuine handicap |
None of these five numbers is exotic. They are standard output from any current cloud platform, which means an inability to produce them is a statement about the age of the system rather than about the difficulty of the question.
The Version for a Team of Four
Everything above scales down. A four-person business does not need service level targets or shrinkage models, and would be worse off with them. It needs two numbers and one habit.
| What | Why this and not the rest |
|---|---|
| Number 1: how many calls went unanswered last week, and when | For a small business this is close to the entire story. Each missed call is a real person who wanted something, and at four people you can plausibly follow up every single one |
| Number 2: how many arrived outside opening hours | Usually far more than expected. It is the number that decides whether you need an after-hours arrangement or simply a better voicemail and an automatic text |
| The habit: look at it every Monday morning, for four weeks | The pattern appears by week two and it is always specific — a day, an hour, a person on leave. Then act on the pattern rather than on the feeling that the phone is busy |
And one action that needs no report at all
Turn on an automatic SMS on every missed call. It converts a caller who gave up into a conversation you can still have, it works identically whether you were shut or simply flat out, and it requires no analysis, no roster change and no meeting. For most small businesses it is the highest-return change on this entire page.
What to Do This Week
Five questions, in the order that gets you to a change fastest.
- Monday: ask whoever administers your phone system for abandoned calls by wait-time bucket, and half-hourly performance for last week. Two requests, one email.
- Tuesday: look at the two worst intervals. Find out what the team is doing in those windows. It is usually a meeting or a break.
- Wednesday: move the meeting and stagger the breaks. This is the whole intervention, and it costs nothing.
- Thursday: turn on missed-call SMS if it is not already on. Check whether accepted callbacks are being counted as abandons.
- Next month: pull the same two reports and compare. If the interval fix worked, the abandons in those windows will have dropped visibly and nothing else will have changed.
9% / 5%
AU abandonment: average / median
~32%
Banking FCR, Q1 2026
85%
Occupancy ceiling
2
Intervals holding most of your failures
The point of all five questions is the same. Averages describe a business that does not exist — the composite of many easy calls and a few bad ones. Distributions and intervals describe the actual customers, including the ones you are losing. Ask for the tail, look at the hour, and act on the two windows where the failures live. That sequence beats benchmark-chasing in every business it gets applied to, and it can start on Monday with one email.
If you are building the queue rather than measuring one, the contact centre software guide covers the capability set, one inbox for calls, SMS and chat covers what changes when the same team handles more than voice, and AI call scoring covers the quality dimension none of these five numbers measures.
Frequently Asked Questions
What should I ask my phone system provider for first?
Two things in one email: abandoned calls broken down by how long each caller waited before hanging up, and half-hourly performance for a normal week. Those two reports produce more actionable insight than anything else available, and neither is exotic — both are standard output from any current cloud platform. The abandonment breakdown tells you which problem you have. Abandons clustered under thirty seconds are a design problem, meaning the menu is too deep, the greeting too long, or there is no visible sign a human exists, so people leave before they ever reach a queue. Abandons at one to three minutes are an expectation problem: people will wait if they are told what they are waiting for, so position announcements, estimated wait times or a callback offer move that number without adding a single person. Abandons past three minutes are a staffing problem in specific intervals, and no menu redesign will touch them. The half-hourly report tells you when. Nearly every business discovers the same thing — performance is fine most of the day and collapses in two windows, the first hour of trading and the half hour after lunch.
Why is the average wait time misleading?
Because an average is dominated by the majority of easy calls, and the customers you are failing are by definition a minority, so they are invisible in every mean you calculate. Take five hundred calls where four hundred and fifty were answered almost immediately and fifty waited more than four minutes, thirty of whom hung up. The average wait lands around forty seconds, which sounds healthy and describes nobody's actual experience. Ask instead for the ninetieth percentile wait — the number that completes the sentence 'ninety per cent of callers waited less than'. A queue with a forty-second average and a ninety-second ninetieth percentile is consistent and healthy: everyone gets a similar experience. A queue with the same forty-second average and a six-minute ninetieth percentile is fine most of the time and badly broken for a tenth of callers, which is a completely different business problem hidden behind an identical average. That worst-served tenth is where escalations, complaints, negative reviews and churn come from. Australian sector data illustrates the spread: ACXPA's 2026 Best Practice Report put utilities slowest at a 227-second average speed of answer, with banking and finance carrying the highest median at 79 seconds.
What is a normal call abandonment rate in Australia?
ACXPA's 2026 self-reported Australian figures put voice abandonment at an average of nine per cent with a median of five per cent. The gap between those two numbers matters: a minority of operations with very high abandonment are pulling the average up, so if you are above nine per cent you are not in the middle of the pack, you are in its tail. Before comparing yourself to any benchmark, check two things about how your own number is calculated. Many systems exclude abandons within the first five or ten seconds on the reasoning that they are misdials, which is defensible but materially changes the figure and makes cross-organisation comparison meaningless unless both sides count the same way. And check whether a caller who accepts a callback offer and then hangs up is being logged as an abandon — that records a successful interaction as a failure and can make a genuinely good change look like a regression. Much more useful than the headline rate is the distribution: knowing when people give up tells you whether you have a design problem at the front door or a staffing problem in specific intervals, and those have entirely different fixes.
Why does half-hourly reporting matter so much?
Because daily and monthly averages conceal interval failure, and interval failure is where nearly every bad experience lives. A queue reporting eighty-two per cent of calls answered for the day looks like it met a target, but broken into half hours the usual pattern emerges: excellent for most of the day, and collapsing in two windows — the first hour of trading and the half hour or so after lunch. Those two intervals typically contain most of the abandoned calls and most of the long waits, which means essentially every unhappy caller you have is concentrated into about ninety minutes of the day. Then look at what the team is doing in those windows, because in most businesses the answer is the morning meeting and the post-lunch handover. The two intervals where the queue fails are the two intervals when the fewest people are on the phones, and nobody has ever connected the two facts because nobody had the interval report. Moving meetings and staggering breaks out of those windows costs nothing, requires no headcount and routinely produces a bigger improvement than anything else available. Also ask for volume by half hour broken down by weekday, since Monday is not Wednesday and a roster built on a weekly average under-staffs Monday morning permanently.
How do I tell whether problems are actually being solved?
Ask for the number of callers who contacted you more than once within seven days. It is the cheapest available proxy for resolution, and it measures something different from everything else: not whether calls were answered, but whether they were finished. That is what customers care about and what drives your cost base, since every repeat contact is a call you are paying to handle twice. For a sense of scale, ACXPA's Australian Call Centre Rankings put banking at roughly thirty-two per cent first contact resolution in the first quarter of 2026, meaning around two-thirds of contacts were not resolved first time. If you have never measured your own, assume it is worse than you think. The repeat-contact method beats the alternatives because it is objective and automatable, whereas asking agents to self-report resolution is systematically optimistic — people are poor judges of whether they solved someone else's problem — and post-call surveys have low response rates that skew to the very happy and the very angry. Its blind spot is real: it cannot see the customer who gave up rather than ringing back, so pair it with the abandonment number, which does. Break it down by contact reason, because a poor overall figure is nearly always two or three specific issue types rather than a general failing.
Is high occupancy a good thing?
No, and this is one of the most common management errors in a phone team. Occupancy is the share of logged-in time spent handling contacts, so the instinct is to push it up as a productivity measure. It behaves more like a physical limit. Sustained operation above roughly eighty-five per cent produces rising errors, dropping courtesy, increased sick leave and eventually attrition. And because calls arrive randomly rather than smoothly, a team running at very high occupancy has no absorption capacity, so a modest spike in volume collapses the queue rather than being soaked up. Between about seventy and eighty-five per cent is the workable band — enough slack to absorb a spike, enough load that people are not idle. Below about sixty per cent usually means genuine overstaffing for the volume, or work arriving in a very uneven shape that better interval rostering would smooth. Related to this, check your shrinkage assumption: the proportion of paid time genuinely unavailable for calls, covering breaks, training, meetings, leave, coaching and system time, is commonly thirty to thirty-five per cent and is routinely planned at twenty. That single underestimate is the most frequent cause of a queue that is inexplicably short every afternoon despite a roster that looks correct on paper.
We are a team of four. Is any of this relevant?
Most of it scales down, and some of it should be dropped entirely. A four-person business does not need service level targets or shrinkage models and would be worse off with them. It needs two numbers and one habit. The first number is how many calls went unanswered last week and when — for a small business that is close to the entire story, because each missed call is a real person who wanted something and at four people you can plausibly follow up every single one. The second is how many calls arrived outside opening hours, which is almost always far more than expected and is the number that decides whether you need an after-hours arrangement or simply a better voicemail message and an automatic text. The habit is to look at both every Monday morning for four weeks. The pattern appears by week two and it is always specific — a particular day, a particular hour, someone on leave — and then you act on the pattern rather than on a general feeling that the phone is busy. One action needs no report at all: turn on an automatic SMS on every missed call. It converts a caller who gave up into a conversation you can still have, works identically whether you were shut or flat out, and needs no analysis or roster change.
