Audit Your Phone Tree

The phone menu is the only part of a practice that every caller experiences and almost nobody inside the practice ever hears. Staff dial internally or answer from a handset. The owner calls the mobile of whoever they need. The menu was recorded at some point by a person who may no longer work there, and it has been quietly accumulating options ever since.

The result is predictable and it is measurable. Callers abandon inside menus that are too deep, choose the wrong option and land somewhere nobody monitors, or hear a greeting that says the practice is open when it closed 40 minutes ago.

This is a 2 hour audit that costs nothing. It is the fix a lot of practices should make before evaluating any product, because a menu that loses callers will keep losing them regardless of what answers at the end of it.

What a phone tree is actually for#

A menu exists to route a caller to the right person faster than a human could ask them. That is the whole justification, and it is worth stating because most menus have quietly stopped doing it and are now routing callers to whichever department existed when the menu was recorded.

The test is comparative rather than absolute. If a caller would reach the right outcome faster by simply talking to whoever picked up, the menu is costing you rather than saving you, and that is true even when every option in it technically works.

There is a reason menus grow rather than shrink. Every option was added to solve a real problem somebody had at the time, usually a department tired of receiving calls meant for someone else. Nothing in a phone system's lifecycle ever proposes removing one, so the structure only accumulates, and after a few years it describes a practice's history rather than its present.

Most menus fail that test for the majority of callers while genuinely helping a minority. The job of this audit is to find out which is which at your practice rather than to assume either.

A menu is only worth having if it routes faster than a person asking one question. Most menus have stopped clearing that bar.

It helps to know what good looks like elsewhere. A 2022 study of 6 Veterans Affairs medical centres selected for above-average primary care access records the telephone standards those sites worked to: an average speed of answer of 30 seconds or less, and a call abandonment rate under 5% [3]. A tree that adds 40 seconds of menu before anyone is even queued has spent the whole budget before the clock starts.

01Call your own practice, as a patient would#

Everything starts here and it cannot be done from inside the building. Use a mobile that is not on your system, because internal calls frequently bypass the menu entirely and you will audit something no patient ever hears.

1. Call from an outside mobile, at a time you have not planned around, and do not tell the front desk.
2. Write down the greeting word for word, including how long it runs before the first option.
3. Write every option in the order it is offered, with the exact wording.
4. Time the whole thing from ring to first option, and from first option to a human.
5. Do it again after close, and again at a weekend, because those are different recordings and one of them is usually wrong.

Take the notes by hand rather than trying to remember. You are going to be listening, timing, and pressing keys at the same time, and the wording is the part that matters most and the part memory reconstructs most generously. What you want at the end is a transcript, not an impression.

Step 1's instruction not to warn anyone matters. A front desk that knows the owner is calling answers on the second ring, which is not the experience you are auditing.

Step 5 finds more problems than the rest of this lesson combined. Out-of-hours greetings are recorded once, reference opening hours that later changed, and are never heard by anyone who could correct them. A greeting that gives Saturday hours the practice stopped offering 2 years ago is a very common finding.

02Map what exists, not what you think exists#

Now get into the phone system's configuration and write down the actual structure. It will differ from the menu you just heard, and the difference is where the problems live.

The fourth row is the one people have never looked at. A caller who presses nothing is usually an older patient on a landline or someone who did not understand the options, and what happens to them is a decision your phone system made by default rather than one you made.

A voicemail box nobody monitors is worse than no option at all. It takes the caller's time and gives them a false sense that somebody now knows.

Ask specifically who monitors each voicemail box and when. If the answer for any box is a name that hesitates, or a person who left, you have found something, and it will have been silently collecting messages for as long as nobody has been listening.

03Count the depth to a human#

Depth is the single number most strongly associated with abandonment, and it is easy to measure and easy to reduce. Count the key presses, not the menus, because a menu with 6 options is one press.

1. From the first ring, count every key press required to reach a person for your most common call reason.
2. Repeat for your second and third most common reasons.
3. Count the seconds from ring to human on each path, with no hesitation.
4. Count the same path again, hesitating realistically at each menu as a first-time caller would.
5. Write both numbers down, because the gap between them is what patients actually experience.

Do this for the top 3 reasons rather than only the first, because menus are commonly optimised for the reason the practice cares about rather than the reason patients call about. A practice whose most common call is a rebooking and whose 1-press option is new enquiries has its structure inverted, and the count is what makes that visible.

Two presses to a human is fine. Three is a lot. Four means the majority of your callers are navigating a structure built for the convenience of the practice rather than the caller, and the abandonment in your call log is not a mystery.

The realistic timing in step 4 is the honest one. A caller hearing options for the first time pauses, re-listens, and sometimes restarts, so a path that takes 25 seconds when you know it takes 55 seconds when you do not.

Count the presses, then count them again while pretending you have never heard the menu. The second number is the real one.

Depth has a price you can put a number on. At the O*NET median of $22.08 an hour for medical secretaries and administrative assistants [2], the staff time is not where the loss sits, because the tree exists precisely to avoid that cost. The loss sits with the caller who hangs up on level 3, and the tree is only earning its keep if the calls it deflects are worth more than the callers it sheds.

04Find the dead ends#

A dead end is any path that does not reach a person or a monitored destination. Every practice has at least one and most have several, because options accumulate and nothing ever removes them.

• An option pointing at an extension that no longer exists
• An option pointing at a handset nobody sits at any more
• A voicemail box nobody has the password for
• A voicemail box that is full and rejecting messages
• A submenu whose only escape is hanging up
• A department that was closed or merged and still has an option
• An option for a service the practice stopped offering

Test each one rather than reasoning about it. A destination that looks correct in the configuration can still be wrong, because an extension can exist, ring, and belong to a handset that was moved to a treatment room 18 months ago. The only reliable test is following the option to its end and hearing what a caller hears.

Check the full voicemail one specifically, because it fails silently and completely. A full box takes the caller's time, plays the greeting, and then hangs up on them, which is the worst possible outcome and produces no record anywhere.

There is a diagnostic worth running while you are here. Compare the list of options against the list of call reasons from your tally sheet. Options with no matching call reason are candidates for deletion, and call reasons with no matching option are the ones causing callers to guess, which is where wrong-option traffic comes from.

An option nobody chooses and a reason nobody has an option for are the same fault seen from two ends.

The last two rows are worth being ruthless about. An option for a service you no longer provide sends interested callers to a disappointment, and an option for a merged department sends them to somebody who has to redirect them, which is the exact work the menu was supposed to save.

05Check the greeting against the clock#

The greeting is doing more work than the options and it is where the quickest wins usually are. Most greetings are too long, and length at the front of a call is expensive because the caller is deciding whether to stay.

1. Time your greeting. If it runs longer than 15 seconds before the first option, it is too long.
2. Remove anything the caller did not ring to hear: awards, mission statements, website addresses.
3. Move the opening hours to after the options rather than before them.
4. Check the out-of-hours greeting states what the caller should do, not only that you are closed.
5. Check every holiday recording, and diarise the ones that need changing each year.

Step 2 is worth being unsentimental about. The greeting is not a marketing surface. A caller has already chosen you enough to dial, so an award announcement is 6 seconds spent telling somebody who is trying to book an appointment something they did not ask for, at the exact moment they are deciding whether this call is going to be worth the effort.

Step 3 is counterintuitive and it works. Callers who ring during opening hours do not need to be told the opening hours, and callers who ring outside them have already found out. Putting hours first delays every caller to inform a minority.

Step 5 is the one that needs a diary entry rather than good intentions. Holiday greetings are recorded in a hurry, left in place, and then contradict the practice's actual state for weeks afterwards, and the caller who hears one in February has no way to know it was meant for December. Put a recurring reminder against each one on the day it should change.

Step 4 is the substantive one. "We are closed, please call back during business hours" leaves the caller with a problem and no route. Naming what to do for an urgent clinical concern, and giving one clear action for everything else, is the difference between a dead end and a handoff.

06Put the most common reason first#

Menu order should follow your actual call volume, and almost no menu does, because menus are ordered by how the practice is organised internally rather than by why people ring.

If you do not have your call reasons ranked, collect them for 1 week with a tally sheet, which is lesson 1 of writing your call handling rules. It is the same data and doing both at once saves a week.

There is an ordering subtlety worth knowing. Callers do not reliably hear past the third option, so anything in position 4 or lower is effectively invisible to a first-time caller and will be reached mostly by people who already know it is there. That is fine for a genuinely rare reason and a serious problem if your second most common reason has drifted to position 5.

Then cut. A menu with 7 options is not more helpful than one with 3; it is a longer list to hold in your head while deciding. If 2 options cover 70% of your calls, offer those 2 and route everything else to a person, which is both simpler and closer to what the caller wanted.

Watch for the department that is really a person. Practices frequently carry an option for a function that one specific member of staff performs, which works while that person is at their desk and becomes a dead end the moment they are on leave. If an option depends on one individual being available, it needs a fallback or it should not be an option.

Use the caller's words rather than yours. "Billing and accounts" means nothing to somebody ringing about a bill they do not understand, and "treatment coordinator" is an internal job title that no patient has ever used to describe why they are calling.

07Test the urgent path specifically#

This is the lesson that justifies the whole audit. Every other fault costs you a booking; a broken urgent path costs something else, and it is the path least likely to have been tested because nobody wants to occupy it unnecessarily.

• Call from outside and follow the urgent option to its end, at 3pm on a weekday
• Do the same at 9pm on a weekday
• Do the same on a Sunday
• Confirm a person actually answers, not a recording telling you to call back
• Confirm the number it reaches is a number somebody carries, not a desk phone in an empty room
• Confirm what happens if that person does not pick up within 60 seconds
• Confirm the greeting names what counts as urgent, so callers self-select correctly

The fifth line is the failure that most surprises practices. Out-of-hours urgent options frequently ring a handset in the building, which is a phone ringing in a dark room with nobody in it, and it has usually been that way since the system was installed.

Survey data supports drawing this line clearly for callers. Patients accept automation for administrative tasks and consistently want a human for clinical matters, with 81% saying they want a person for actual medical advice [1]. A menu that makes the urgent route the hardest to find inverts exactly that preference.

The sixth line is a rota question rather than a phone question, and it is the one that tends to get deferred. Whoever is on call needs to know they are on call, and the handset needs to be with them rather than charging in the office. Practices that discover a broken urgent path usually find the phone configuration was correct and the human arrangement behind it had quietly lapsed.

The seventh line is about self-selection and it is cheap to fix. If your greeting says "for anything urgent, press 1" without saying what urgent means, callers apply their own definition, and you get both over-use by anxious patients and under-use by people who did not want to bother anyone.

Get the urgent path wrong and the cost lands outside your practice. In a quality improvement study at an urban multispecialty practice, running from November 2016 to November 2017, 33.8% of surveyed patients reported seeking emergency or urgent care because they could not reach their provider [4]. A menu that buries the urgent option under 2 levels is one of the ways that happens.

08Re-record, then walk the whole thing again#

Changes to a phone system routinely do not do what the person making them expected, so the last lesson is verification rather than configuration.

1. Make the changes in one session rather than over a fortnight, so you know what changed.
2. Re-record the greeting yourself or with whoever has the clearest voice, in a quiet room.
3. Walk every path from an outside line again, including the ones you did not change.
4. Walk the out-of-hours version after close on the same day.
5. Have one person who was not involved walk the main path and describe what they heard.
6. Write the date on the configuration note, so the next audit knows how old this one is.

Step 1 is about attribution again, at a smaller scale. A fortnight of small adjustments produces a menu nobody can describe and a set of results nobody can explain, whereas an hour of changes made together leaves one clear before and after. Book the hour rather than fixing things as you notice them.

Step 3 catches the most common regression. Phone systems share destinations between options, so a change to one option's target frequently moves another option nobody was thinking about.

Step 2 is worth more care than it usually gets. A greeting recorded on a mobile in a corridor sounds like a greeting recorded on a mobile in a corridor, and it is the first impression of the practice for every caller who has never been. Use a quiet room, stand still, and record it 3 times, then keep the least hurried one.

Step 5 is worth the 10 minutes. Somebody who did not build it will hear an ambiguity you have stopped noticing, and their confusion is a preview of your callers' confusion.

What to keep measuring afterwards#

Two numbers tell you whether this worked, and both come from the call export you are already pulling. Abandonment inside the menu, and the share of calls reaching a person on the first attempt, both compared against the month before you made the changes.

Set the comparison up before you change anything, since the point of the audit is lost if you cannot show what it did. Pull last month's abandonment figure now and write it beside the date you made the changes, because a phone system report is a rolling window and the number you needed will be harder to reconstruct in 6 weeks.

Expect the menu abandonment number to move within a month, since it responds directly to depth and greeting length. Expect nothing else to move, because a phone tree audit does not create capacity and does not answer more calls; it stops losing the callers who were already trying.

One warning about attributing the change. If you run this audit in the same month you alter opening hours, add a member of staff, or start any marketing, you will not be able to say which of them moved the number. The audit is cheap enough that it is worth running in a quiet month on its own, purely so the before and after mean something.

Re-walk the whole tree every 6 months and immediately after any staffing or hours change, both of which silently invalidate destinations. The wider set of numbers this belongs to is in read your front office numbers, and if the audit shows your losses are outside opening hours rather than inside the menu, sizing your missed-call gap is the exercise that quantifies that instead.