AI receptionist or a person: how to compare them honestly
Search for the cost of an AI receptionist and you will meet the same kind of number everywhere: the share of calls a business misses, what each missed call is supposedly worth, how few callers leave a voicemail. Check where those figures come from and the trail ends at companies selling AI receptionists. That does not make them false. It does mean nobody should build a budget on them, and it is worth knowing that the scary number in the headline was published by the party that benefits from you being scared.
The comparison below uses numbers you can produce yourself in an afternoon.
What does each option actually cost?
The published prices are the easy half, and unlike the statistics they are checkable.
| AI answering service | Part-time person | Full-time receptionist | |
|---|---|---|---|
| Typical cost | $30-300/mo | hourly, local rates | $35,000-50,000/yr plus overhead |
| Hours covered | all of them | the hours you buy | business hours, minus leave |
| Handles the unusual | no | yes | yes |
| Improves with your business | only if maintained | yes | yes |
Both ends of that table mislead if read alone. The AI column ignores that somebody has to maintain it and that a mishandled call still costs you a customer. The salary column ignores that a person does far more than answer the phone, and comparing the two as though they do the same job is the error the whole category is built on.
How do you find out what missed calls really cost you?
Four numbers, all from your own records, none of them requiring a vendor's survey.
- Missed calls per week. Your phone system has this. If it does not, count for one week by hand.
- How many were new enquiries rather than existing clients, suppliers or spam. Usually a much smaller fraction than the marketing material assumes.
- Your close rate on enquiries you do answer. From your own history.
- Your average first-year value per client. Not the value of one job - what a client is worth over the first year.
Multiply: missed enquiries per week, times close rate, times client value. That is your annual exposure, calculated from your business rather than an industry average that may not resemble it at all.
Doing this honestly usually produces a smaller number than the vendor statistics and a more useful one. It also tells you something the averages cannot: whether your missed calls happen at 2am, in which case an answering service covers them, or at 11am while you are with a client, which is a different problem with different solutions.
What can an AI receptionist genuinely do?
The category has improved and it is worth being specific about where.
It answers every call immediately, at any hour. No queue, no voicemail, no difference between Tuesday morning and Sunday night.
It handles the repeated questions. Opening hours, location, whether you cover a postcode, what a service costs, whether you have availability this week. In most service businesses this is the majority of inbound calls.
It books appointments against a live calendar, which removes the phone tag that turns a two-minute booking into a two-day exchange.
It takes structured details and writes them somewhere you can act on them, instead of on a sticky note.
It never has a bad morning. Consistency is genuinely valuable and rarely mentioned, because it is not exciting.
What does it still not do?
Judgement. A caller whose situation does not fit any category gets handled badly by a script and well by a person who can think.
Recognising who matters. Your largest client calling with a problem sounds to a system like any other call.
Anything off-script that matters. The caller who mentions in passing that they are talking to a competitor - a person hears that, a system does not.
The rest of the job. A receptionist also greets people, chases documents, manages a diary, calms an annoyed visitor and notices that something is off. Replacing the phone-answering part of that role is not the same as replacing the role, and quotes that compare $99 a month to a $40,000 salary are comparing a feature to a person.
So which one is right?
The useful question is not which is better, it is which problem you have.
Overflow and after-hours. Calls arriving when nobody can answer, mostly routine. This is the strongest case for an answering service and the cheapest thing on this page.
Volume of identical questions. The same six questions all day, eating the time of someone who should be doing skilled work. Automate the questions, keep the person.
Complexity in every call. Each caller needs a real conversation and a judgement. Hire, and automate the paperwork around the calls instead.
Front desk with people physically present. A clinic waiting room needs a human being. The phone can still be automated after hours.
Most small businesses in practice land on a combination: a person during business hours, automatic handling of the after-hours and overflow calls that were previously going to voicemail - which is where the honest version of the missed-call argument lives.
What to check before buying either
Before an answering service:
- Does it hand over cleanly? The moment it cannot help, does the caller reach a person or hit a wall? Test this yourself, twice.
- What does it do with what it collects? Details that land nowhere you look are the same as details nobody took.
- Does it say what it is? Callers who discover mid-conversation that they are talking to software feel misled, and that costs more than the service saves.
- Who updates it when your prices change? An answering service quoting last year's prices is worse than a voicemail.
Before hiring: write down the hours you actually need covered, and check whether the answer is "eight hours a day" or "the four hours around lunch and everything after six". Those are different jobs, and only one of them is a full-time role.
Where the honest line falls
An AI receptionist is a strong buy for a narrow, well-defined slice of the problem: routine questions, out of hours, overflow. It is a weak substitute for a person in a role that involves judgement, presence, or anything a caller would resent being handled by a machine - which we go into more generally in when not to automate a process.
The way to know which you have is to run the four-number calculation above against your own call log. If you would rather have that done properly, along with everything else in your week that repeats, that is what the process audit is for: $299, three business days, and a written map you keep whether or not you hire us.