"Is an AI receptionist worth it?" is the right question and it deserves a better answer than a vendor's yes. The honest answer is: it depends on three numbers you already have, and there are four situations where it is clearly not worth it. This is the sum, the research behind it, and the cases against, written by a vendor who would rather you did the maths than took our word for it.
The three numbers
1. How many calls do you miss? There is no official UK statistic for this, so the research is commercial and the ranges are wide. A 2025 study of 142 UK SMEs found 47 per cent of initial calls unanswered; a 2017 survey of 300 micro-businesses (mainly trades) put it at a third; a reasonable cross-sector average is about a quarter. Count your own: look at last week's missed-call log and divide by the calls received. If you have no log, assume one in five and you are being generous to yourself.
2. What is a new customer worth? Not one appointment; the customer over the time they stay. A salon client who visits every six weeks at £60 is worth about £500 a year. A physio episode of care is five and a half sessions on average. A dental new patient is years of check-ups and the occasional crown. A plumbing customer calls you back for the boiler. Put a number on it, then halve it to be safe.
3. What share of callers would have booked? Not every missed call is a customer. Some are suppliers, some are wrong numbers, some would not have booked anyway. Use one in ten and you are being conservative.
The sum
Fifteen calls a day, one in five missed, one in ten of those a customer worth £150: that is 3 recovered enquiries a day, roughly 65 a month, about 6.5 new customers, close to £1,000 a month. Against that, a flat-rate AI receptionist at £149 a month (or a per-minute plan that lands somewhere between £50 and £400 depending on your volume, which is its own problem, covered in AI receptionist pricing in the UK).
Change the inputs to your own and the answer moves, but it does not move to "no" for most service businesses with a ringing phone. It moves to "no" in the four cases below.
When it is not worth it
Very low call volume. If you take a handful of calls a week, the sum above collapses and a per-call plan (IONOS publishes £39 ex VAT for 30 calls) or a well-written voicemail with a same-day callback habit costs less. Be honest about the volume.
Calls that need a human above all else. A bereavement service, a mental-health practice, a complaint line. The AI can answer and hand over, but if every call needs empathy and judgement from the first word, pay for the people. AnswerConnect UK publishes £395 a month for 350 minutes; Moneypenny quotes. That is the price of trained humans and it is fair. The full comparison is in AI receptionist vs a human answering service.
A diary the AI cannot book into. If the product can only take a message, you have bought a better voicemail. The value is in completing the booking, so check the integration list by name before you sign, and check whether the tier you are looking at can book at all (several entry tiers cannot; RingCentral's UK page still lists booking as "coming soon").
An owner who will not do the hour of setup. The AI answers from the facts you give it: services, prices, hours, what to do with an emergency, who to transfer to. Skip that hour and it will answer badly, and you will conclude it does not work. It is an hour.
Where it loses to a person, honestly
A person reads a voice. A person notices the caller is frightened, or angry, or elderly and confused, and changes register. A person can make a judgement call nobody wrote down. A person greets the walk-in, takes the card payment and makes the tea. An AI receptionist does none of those. It answers on the first ring at 07:50 and 19:20 and during the 11:00 rush, it never has a bad day, it books the slot and sends the reminder, and it costs a fraction. Both statements are true. The businesses that get the most from it keep the person and take the phone off them.
What "worth it" should include
The subscription is the small number. Check what it buys:
- Booking into your own diary, with reschedule and cancel on the same call.
- Reminders and deposits, because the no-show you prevent is worth more than the call you answer.
- Missed-call text-back, so the one call it could not take still comes back.
- WhatsApp and text, not just voice.
- Chasing what you are owed. Remi's PaidUp follows up unpaid invoices until they are paid, which for a trade or a private practice can be worth more than the reception itself.
- A clear AI disclosure and a warm transfer to a human when a caller asks.
- A data processing agreement under UK GDPR and a stated hosting region.
If the product does the first three, the sum works. If it does all seven, it is an employee.
The short version
Count your missed calls, put a value on a customer, assume one in ten converts, and compare with the subscription. For a business with a ringing phone and a diary to fill, the answer is usually yes by a wide margin; for a business with very few calls, calls that need a human first, a diary that will not connect, or no time to set it up, the answer is no, and a human service or a good voicemail is the better spend.
Remi is £149 a month ex VAT with 1,000 minutes included, then 15p a minute, one-week free trial, no setup fee. The trial exists so you can run this sum on real calls instead of ours.