AI Receptionist for Small Business: An Operator's Guide
Written from building these systems and then running them through live traffic. Which parts of a receptionist's job AI can hold, which parts it should never be given, and how the cost compares with the hire you were actually considering.
Most comparisons of AI receptionists put the product against other products. That is the wrong shelf, and it is why so many of them are unsatisfying to read.
You are not comparing software to software. You are comparing it to the hire. The alternative to an AI receptionist in a small business is almost never a different piece of software. It is a part-time person, or it is nobody and the phone keeps ringing out.
Once you frame it that way the interesting questions change. The monthly fee stops being the point, and what matters is which parts of the job actually get covered.
The statistic everyone quotes, and the one worth using
Almost every page selling this will open by telling you 62% of calls to small businesses go unanswered. It is worth knowing where that number comes from before you make a decision on it.
It traces back to a study by 411 Locals, a lead generation company, which called 85 businesses across 58 industries and found 37.8% were answered by a live person. Eighty-five businesses is a small sample, and the organisation that produced it sells lead generation. That does not make it false. It makes it a marketing survey rather than a measurement, and it has been repeated so widely that it now reads like census data.
The number we would rather work from comes from call analytics rather than a survey. Invoca, which instruments real call traffic for its customers, reported an average missed call rate of around 27% in home services in 2024. Lower than 62%, still high enough to matter, and grounded in calls that actually happened rather than calls placed to test people.
Roughly one call in four goes unanswered at a typical service business, and the rate climbs sharply for solo operators and anyone who spends the day on a job site or in someone's home. That is a real problem. It is not a 62% catastrophe, and you should be suspicious of anyone who needs it to be.
The other figure worth having is timing. A commonly cited breakdown puts around 43% of calls to small businesses outside standard opening hours. That one matches what we see in practice: the calls that go missing are rarely spread evenly through the day. They cluster in the evening, at weekends, and during the two hours a tradesperson spends driving.
Which points at something useful. If your missed calls are concentrated outside hours, you have a coverage problem and almost any form of after-hours answering fixes it. If they are concentrated at 11am on a Tuesday, you have a capacity problem, and that is a different purchase.
Work out which one you have before you buy anything. Most phone systems will export a call log. Missed calls with a timestamp, one month, sorted by hour, answers the question in about ten minutes.
The five jobs a receptionist does, and which ones AI can hold
A front desk in a small service business does roughly five distinct things. They are not equally automatable and the gap between them is where most disappointment comes from.
1. Answer the phone and take the details — AI holds this
The highest-volume, lowest-judgement part of the role. AI does it reliably, at three in the morning, on the fortieth call of the day, without transcription errors. This alone is most of the phone volume in a typical trades or clinic business.
2. Answer routine questions — AI holds this, if grounded
Hours, service area, whether you do that kind of job, roughly what it costs, where to park. The condition is that the answers come from your actual information rather than the model's general knowledge. Grounded, this is solved. Ungrounded, it is the most dangerous thing on this list.
3. Book the appointment — partly
Simple availability rules work. Real scheduling in a service business rarely is simple: it depends on job type, drive time, which technician is certified for what, and whether the part is in the van. We tend to build the booking as a proposal that a human confirms, rather than letting the model commit to something the business cannot deliver.
4. Read the caller — AI does not hold this
Recognising that someone is distressed, or lying, or a supplier pretending to be a customer, or the third call from a person who is about to leave a one-star review. Humans do this instantly. Models do not do it reliably and confident-sounding failure is worse than no attempt.
5. Everything in the room — not applicable
If your front desk also greets people, handles payment, manages the waiting area and signs for deliveries, an AI receptionist covers the phone portion and nothing else. Obvious when stated, routinely forgotten during a sales conversation.
Which decisions should the system make, which should it only draft, and which should it never touch? In a clinic that line sits much further right than in a landscaping business. Getting it wrong in either direction is expensive: too far left and it is useless, too far right and it is dangerous.
What it costs against the hire
The comparison people run in their heads is a part-time receptionist. Here is the honest version of both sides.
| AI receptionist | Part-time hire | |
|---|---|---|
| Entry cost | $49–$95/mo | $17.90/hr median (BLS, May 2024) |
| 20 hrs/week | same | ~$18,600/yr in wages alone |
| Payroll tax & benefits | none | on top of wages |
| Hours covered | 24/7 | the hours you pay for |
| Ramp to productive | days to weeks | 2–3 months, typically |
| Handles an upset caller | poorly | well |
| Knowledge when they leave | stays | leaves with them |
Wage figure: US Bureau of Labor Statistics, Occupational Outlook Handbook, Receptionists (SOC 43-4171), median hourly wage $17.90, median annual $37,230, May 2024. Employer cost is higher than wage once payroll tax, benefits and equipment are included.
The line worth staring at is not the cost. It is the last one. In a business running on spreadsheets and memory, the process lives inside the person doing it, and when they leave it goes with them. That is the cost nobody budgets and it is usually larger than the salary.
The setup reality nobody puts on a pricing page
Self-serve providers will tell you it takes an afternoon. Forwarding a number and recording a greeting genuinely does. What takes longer is making the thing correct.
Expect a few weeks after go-live where you are correcting it: a service it described wrongly, a price that moved, an edge case nobody thought to write down. This is not a defect in the product. It is the process of extracting knowledge that has never been written down anywhere, and it is precisely the reason the system becomes valuable. By the end of it you have your business's actual operating knowledge in a form that does not resign.
Two things to insist on regardless of who you buy from:
- You can see every call. Transcript, what it decided, what it told the caller. If you cannot audit it you cannot correct it.
- There is a defined behaviour for "I do not know". Ask directly. Push back on a vague answer. We wrote about why in what happens when AI gets it wrong.
When we would tell you not to buy this
- Your enquiries do not come by phone. Check before you assume.
- You miss fewer than two or three calls a day. The arithmetic does not work and your attention is worth more elsewhere.
- Nobody will own it. An unmaintained AI receptionist becomes wrong slowly and quietly, which is worse than not having one.
- What you actually need is scheduling. If the calls get answered but the diary is chaos, an answering layer will not help.
Receptionist first, then what
Most businesses we work with start here because it is the most visible problem, and the phone is where lost money is easiest to see. It is rarely the only one.
The pattern that repeats: the calls get answered, and then it becomes obvious that answering them was never the bottleneck. The follow-up was. Or the quote that took four days. Or the fact that nobody could say which enquiries turned into work. A receptionist that captures calls into a system that remembers them is worth several times one that texts you a name.
That is the difference between buying an answering service and building an operating system, and it is a real fork rather than a sales upsell. Plenty of businesses should take the first road. We would rather tell you which one you are on.
Common questions
What does an AI receptionist do for a small business?
It answers the phone, identifies the caller and what they need, answers routine questions from your own information, books appointments, and escalates anything it cannot handle to a person. In practice it covers the four or five repetitive jobs a front desk does, and none of the judgement-based ones.
How much does an AI receptionist cost for a small business?
AI-only plans typically start between $49 and $95 a month. A part-time human receptionist in the US costs considerably more: BLS put the median receptionist wage at $17.90 an hour in May 2024, which is roughly $37,230 a year before payroll tax, benefits and equipment.
Can an AI receptionist replace a human receptionist?
It can replace the repetitive portion of the role, which in many small businesses is most of the phone volume. It cannot replace the judgement, the reading of a difficult caller, or the in-person part of a front desk. Businesses that treat it as coverage for the hours and calls a person cannot cover get better results than those treating it as a headcount replacement.
How long does it take to set up an AI receptionist?
A basic setup with a self-serve provider takes an afternoon: forward your number, load your hours and services, record a greeting. Getting it genuinely accurate about your business, pricing, service area and edge cases usually takes a few weeks of correction after go-live, and that part is where the value is.
What happens if the AI receptionist cannot answer a question?
That behaviour is a configuration choice and you should decide it deliberately. The safe default is that it says it does not know, takes a message, and flags it. A system that guesses in order to sound helpful is the most expensive failure mode in this category, because a plausible wrong answer does not look like an error.
Do I need to change my phone number?
No. Conditional call forwarding sends only unanswered calls to the service after a set number of rings. Your number, your caller ID, and anyone who reaches you directly are all unaffected.
Want to know what this would look like in your business?
No demo, no deck. A short call about which calls you are actually losing and whether answering them is worth automating. If it is not, we will say so.
Related: AI answering services explained · What answering services actually cost · AI cost vs hiring, in full
Want this running on your phones?
Tell us how many calls you miss in a week and what an average job is worth. If an answering service will not pay for itself on those numbers, we will say so.