How to find the dead ends in your phone menu

Ring your own number and press 3. If you hear silence, you have a dead end, and you have probably had it for a while. Nobody reports these: the caller hangs up and rings somebody else, and the only symptom is business you never hear about.

Here are the five faults worth looking for, in the order they cost you money.

Somebody added an option to the recording and never wired it up, or wired it to a queue that was later deleted. The caller presses it, the line goes quiet, and the call is over.

This is invisible from the inside, and that is the whole problem. The person who built the menu presses 1, because they know 1 works. The caller who presses 3 is a stranger who never speaks to you again.

How to find it: press every key, including the ones you are sure nobody uses. Especially those, in fact, since they are the ones nobody has tested since the day they were added.

This is the one people are most surprised by, and it is usually the most expensive.

A caller who presses nothing is not being awkward. They are driving. They are on a speaker across the room. They are on an old handset whose tones the system cannot hear, which is more common than you would think. Some of them are simply waiting to be told what to do, because your greeting sounded like the beginning of a sentence.

A menu should give up waiting after about eight seconds and send them somewhere: whoever answers the general line during the day, voicemail after hours. If it does not, they listen to nothing until they hang up.

How to find it: ring the number and put the phone down. Wait a full minute. That single test finds more than pressing every key does.

The caller presses 1, reaches the right team, and waits. And waits. If nothing is set to catch them when the team is busy or out, the call has nowhere to land and no one is aware it exists.

A queue needs a second question answered: what happens after forty-five seconds when nobody has picked up? Voicemail is a perfectly good answer. Silence is not.

How to find it: press each key at your busiest hour, not your quietest.

Somebody adds a department and updates the routing but not the recording. Somebody re-records the greeting from a document that turns out to be an old draft. Six months later the menu says press 4 for deliveries and 4 does nothing, or 7 reaches the workshop and nobody has ever been told.

This is the fault we would fix first if we could only fix one, because both halves are wrong in a way that reads perfectly on its own. The recording is fine. The routing is fine. Only together are they wrong.

How to find it: write the greeting out as text, list the keys that actually work beside it, and read them against each other. Or paste both into the phone menu builder, which does that comparison in both directions.

Press 2 for support, then 1 for hardware, and you are back at a menu again. Every exit leads to another list of options and none of them ends at a person.

This one survives every careful read-through, because each menu looks fine on its own. It is only wrong as a shape: a group of menus pointing at each other with no way out. What a menu is supposed to end at, and how a published one is versioned, is the phone menu itself. Checking it means following every path to the end, which is exactly the kind of thing people are bad at and software is good at.

Ring your published number twice. Once during opening hours and once after they close, because the after-hours path is a different flow and is usually the less tested of the two.

Each time: press every key, one call each. Press nothing at all and wait a minute. Press a key that is not on the list, twice. Write down what you actually heard, not what you expected to hear. That is the whole method, and it takes about an hour.

If you would rather do it on paper first, the phone menu builder runs all five checks on a menu you type in, and lets you press the keys as the caller before anybody rings the real number. It ships with a template that is broken on purpose, so you can see what a fault looks like before you go hunting for your own. It runs in your browser and keeps nothing.

Sometimes the faults are not faults but a structure nobody can hold in their head. If your menu has nine options because nine teams each asked for one, no amount of dead-end checking will fix what a caller experiences. Building the menu around what callers ring about, rather than around your org chart, is the version of this problem worth solving.

And a menu you cannot change yourself is a menu that stays wrong. Google's own guidance on helpful, reliable content applies as much to a phone line as to a page: the test is whether the person on the other end got what they came for. If changing a greeting means raising a ticket and waiting a week, the drift in fault four is guaranteed. That is worth considering when you look at what a phone system costs: not the monthly figure, but who is allowed to edit it.