We sell one, so read this with that in mind. But a product that is wrong for you is worse for us than a sale we did not make, and there are businesses where an AI receptionist is the wrong purchase.
Here is how to tell whether yours is one of them.
When most of your calls need judgement
Some businesses take calls where nearly every conversation is different and the right answer has never been written down. A specialist trade quoting every job individually. A practice where callers are distressed. A business where the phone is genuinely the sales conversation rather than the booking mechanism.
In those cases the value is in the person, and paying per minute for a skilled human is money well spent. An AI receptionist would handle the easy fraction and hand over the rest, which is a lot of machinery for a small slice.
The test: count your calls for a week and sort them. If the pile marked "needed a person" is the biggest one, buy a person.
When you are not missing calls
This sounds obvious and it is regularly ignored.
If someone answers your phone every time it rings, within a few rings, including evenings and weekends — you do not have a problem to solve. Adding an AI receptionist changes the experience for callers who were already being looked after, and the change is not obviously an improvement.
Get the number before you shop. Your carrier or phone system can usually tell you how many calls went unanswered last month. If it is genuinely low, stop here.
When your callers mostly do not speak English
Ours speaks English today. If a meaningful share of your callers would rather speak Vietnamese, Spanish or anything else, an English-only system will handle them badly, and handling a caller badly is worse than not answering.
This is a real limitation rather than a roadmap item, and you should ask any vendor for a straight answer rather than a future date.
When the calls are mostly not about booking
An AI receptionist is at its best when the caller wants an appointment, a table, or an answer to a factual question. If your phone mostly carries suppliers, staff, deliveries, debt collectors and wrong numbers, the booking capability — the thing you are paying for — is idle.
When you cannot give it a real menu
The system needs your actual services, durations, prices and who can do what. If that information does not exist anywhere except in one person's head, it has to be written down before any of this works.
That is worth doing regardless. But it is work, and a system configured from a half-remembered list will produce bookings that do not fit the day, which people will blame on the software rather than on the list.
When somebody will not own it
Not much ownership — an hour in the first week reading transcripts and correcting the things it got wrong, then occasional attention.
A system nobody looks at drifts. Prices change and the menu does not. A service gets added and the phone does not know. If there is genuinely nobody who will read the transcripts for the first fortnight, it will underperform and you will conclude the technology does not work, when what happened is that nobody told it about the new service.
When you are about to change your booking system
If you are migrating from one scheduling system to another in the next couple of months, wait. Configure the phone against the system you are going to keep, not the one you are leaving. Doing it twice is dispiriting and produces a bad first impression of something that would have worked fine.
Where it is genuinely the right answer
For completeness, the inverse. An AI receptionist earns its keep when:
- Calls go unanswered because nobody is free, not because they are misrouted
- Most callers want to book, change, or ask something factual
- Several calls arrive at once during a rush
- The phone rings outside your opening hours
- You can hand over a real service list
That describes most salons and most restaurants, which is why we built it for them. It does not describe every business, and a vendor who tells you it describes yours without asking any of the above is selling rather than advising.
The honest test
Count your calls for a week. Sort them into "wanted to book something", "had a factual question", and "needed a person".
If the first two piles are the large majority, this is probably worth your time. If the third pile dominates, it is not, and no demo will change that.
If you do want to try it on your own traffic rather than on a vendor's line, five calls to make before you trust an AI receptionist is where we would start — all five are failures we found in our own system before a customer did.