Missed Calls in a Brokerage: What They Cost and How to See Them

The commercial weight of an unanswered call sits in the reason for the call, not in the count of calls. A client ringing during their renewal window has a decision in front of them and a premium movement to understand. A client ringing to notify a claim has had something happen that morning. A business ringing for a quote is very likely working down a list, and the first broker who picks up gets the conversation the other names on the list do not.

Those three calls arrive in the same inbound stream as suppliers, sales calls and clients ringing to confirm something they already have in writing. A single count treats all of them the same way, which is what makes a raw missed-call total move for reasons nobody can act on.

Timing can concentrate the effect further. Renewal cycles fall in particular weeks and a weather event falls on a particular morning, so where inbound volume rises it tends to rise in bursts rather than evenly. A monthly average will smooth those bursts out, which is why the periods worth reading are usually the busiest hours rather than the whole month.

The commercial case for responsiveness at claim time is documented rather than assumed. The 2026 Vero SME Insurance Index, published by Suncorp Group, found that 44 per cent of businesses would be willing to invest in a better claims experience, and among those who agreed, good communication throughout the process was the most cited reason at 56 per cent. Communication through a claim is delivered in conversations rather than in policy wordings, which puts a share of it inside the calls a brokerage answers or does not.

Finding a credible figure for what this costs is harder than it should be. The missed-call statistics that surface most readily on this subject are published by companies selling answering services and automated receptionists, they tend to cite each other rather than any underlying research, and the ones reviewed for this article did not disclose how the number was produced. That is a reason to build your own figure rather than a reason to leave the question alone. The inputs are already in the phone system, and where the routing design itself is the cause, call queues and ring groups are usually the first place a change pays back.

Unanswered calls are rarely spread evenly through the week. They gather in four places, and three of those sit outside the report most firms look at first.

MaxoTel's call history holds every inbound and outbound call with its time, origin, destination, status, duration and cost. It can be filtered to unanswered calls only and downloaded as a CSV file, which makes a week of unanswered calls something you can sort rather than something you scroll.

One caveat carries real weight for any firm running an auto attendant. MaxoTel's own documentation notes that where an auto attendant is in use, calls are marked as connected as soon as the auto attendant picks up. An answer rate read straight off that status can therefore count a call as answered when the caller never reached a person, so the definitions behind answer rate are worth settling before any of these numbers are used to make a decision.

The queue layer answers the second half. MaxoTel's PBX statistics break each number down into calls answered, calls that reached a voicemail and calls abandoned before anyone answered, with the average, shortest and longest wait alongside them, and chart wait times against abandonment in 30 minute blocks read against the call volume for the same period.

One field is worth reading carefully rather than assuming. The duration recorded against a call is the time the call was in progress, which on a call that passed through an auto attendant or waited in a queue is longer than the time anyone spent talking to the caller. Used as a measure of conversation length it will overstate, and the gap is widest on the calls that waited longest.

The difficulty is rarely a shortage of data. The difficulty is that the record was built for a different question. The per-call history is a billing and audit record organised one row per call, the queue statistics are an operations report organised by number and by half hour, and the two will not agree, because one counts calls and the other counts queue events. Both are correct, and a firm reading them side by side without settling which one defines a miss can end up with two numbers and no decision. A few further situational things sit between either record and an answer.

A usable cost figure starts with a definition. Settle what counts as a missed call before counting anything: whether a call that reached a mailbox counts, whether repeat attempts from the same number inside an hour count once or three times, whether internal transfers count at all, and whether direct lines are in scope. Half a page is enough, and it is what makes week four comparable to week one.

Every figure below is an assumption, not measured data, and none of it is drawn from any published source. It is here to show the shape of the calculation. The point of the exercise is to replace each assumption with your own number.

Assume a brokerage takes 250 inbound calls in a week across its main number and its direct lines, and that 92 per cent reach a person, leaving 20 that do not. Assume that of those 20, four are new business enquiries, six are existing clients and the remaining 10 are suppliers, sales calls and repeat attempts by callers already counted.

Take the new business side first, because it is the part that can be sized honestly. Assume half of the four enquiries are recovered by a callback the same day, leaving two that are not. Assume one in five new business enquiries that reach a broker converts to a policy, which gives 0.4 policies a week. Across 45 working weeks that is 18 policies. At an assumed average first-year commission of $600 a policy, the figure is $10,800 a year.

Change one assumption and watch it move. At an answer rate of 96 per cent rather than 92, the weekly unanswered count halves to 10, and if the mix of callers behind those calls is assumed to stay the same, the new business share halves with it and so does the annual figure. That proportional assumption is doing real work and is worth stating: a firm whose remaining missed calls skew towards new business would not see the figure fall that far. At an average commission of $900 instead of $600, the figure rises to $16,200. The sensitivity is the useful output here, because it shows which input is worth measuring properly first.

The six existing clients a week, roughly 270 across a year, are not converted into a dollar figure anywhere above. The temptation to do it is strong and the arithmetic would be easy. The link between one unreturned call and a renewal that later moved cannot be assumed, and a retention number built that way would not survive its first serious question. It is recorded as a count and left there. Nothing in the example establishes causation either. What the figure supports is a decision about whether the gap is worth closing, and at what cost.

The number that says whether any of this is working is the proportion of unanswered calls that turn into a conversation, and no phone system produces it on its own. It takes a small habit at the other end: a note against the callback, or a tag in the CRM, enough to connect the return call to the call it answers. Without that step the recovery rate stays an impression, and an impression is not something a decision about rostering or routing can rest on.

Worth stating plainly first, because it decides what the rest of this section can honestly claim. A call that rang out and was abandoned leaves nothing behind, so there is no recording for a call intelligence platform to analyse, and the phone system remains the source of truth for how many calls went unanswered and when. A call that ended in a voicemail message is a different case: the message is itself a recording, so it can be read and analysed like any other. Callyx.ai works on the recordings your team already produces. A voicemail message and the callback that follows it are both recordings, so both can be transcribed and analysed, while the calls that connected to nothing stay a matter for the phone system's own reporting. Connecting a particular callback to the particular message that prompted it is a matter for whatever reference the firm uses at the CRM end, and is worth setting up deliberately rather than assumed.

None of these five steps requires new software. Four of them require a decision, which is usually the part that gets deferred, and the capacity they return tends to show up in the same place the phone time already goes.

Missed calls in a brokerage are worth measuring because a small number of them carry most of the commercial weight, and because the phone system already holds enough to find them. They collect outside the queue, at the edges of the day, in the short-handed windows and at whatever overflow destination the routing was set to years ago. The per-call history shows which calls went unanswered and when. The queue statistics show what the wait looked like and how many callers gave up. Read together, and with the auto attendant caveat understood, they give a brokerage a real count rather than a borrowed one.

The cost figure is a calculation the firm builds for itself. Published numbers on this subject come from companies selling the fix, and a figure assembled from your own volume, your own answer rate and your own average commission will be less quotable and considerably more defensible. Some of the cost can be sized this way and some cannot, and saying which is which is what makes the exercise useful rather than persuasive.

What the count cannot do is say what happened next. Whether the callback was made, what the client said on it and whether the enquiry converted all sit inside conversations already being recorded. That is the layer Callyx.ai reads, alongside the numbers the phone system was producing anyway. Where the underlying cause is how calls are routed, the routing map is where that work starts, and the relationships it protects are the same ones every retention programme depends on.