What does an AI receptionist cost, and when does it pay for itself?

The short answer
Most AI receptionists cost roughly $50 to $150 a month for a small business, with entry plans near $25 to $30 and busy lines running to $300 or more. You pay by monthly plan, per minute, per call or per booking. It pays for itself the moment it saves one new-customer job a month.
What an ai receptionist actually does
An AI receptionist is a voice system that picks up the phone, talks like a person, and handles the routine parts of a call. It answers when your line is busy, after hours, or on the fifth ring when nobody is free. It takes a name and number, answers common questions about hours and pricing, sorts a real job from a sales pitch, books an appointment into your calendar, and texts the caller a confirmation.
The honest framing is narrow. It is very good at the first ninety seconds of a call: greeting, capturing details, sorting the urgent from the routine, and putting something on the books. It is not a substitute for you, your estimator, or a good office manager. Think of it as the thing that stops a ringing phone going to voicemail, not the thing that closes the job.
The four pricing shapes, and what each one hides
AI receptionist pricing comes in four broad shapes, and the one a vendor leads with tells you how they want you to think about cost. None is automatically cheaper. The right one depends on how many calls you get and how long they run.
Read past the headline number to the overage rate, because that is where a quiet month and a busy month diverge. A low base fee with expensive overage minutes can cost more than a higher flat plan once your calls pick up in season.
How you get billed
- Monthly plan: a flat fee with a bundle of calls or minutes included, then an overage rate above the cap. Predictable, and usually the best fit for steady, low-to-moderate volume.
- Per minute: you pay for talk time, often around a quarter to half a dollar a minute. Cheap if callers are brief, punishing if they ramble or you get long booking calls.
- Per call: a fixed price for each answered call regardless of length. Simple, and it favours businesses with longer conversations.
- Per booking or per outcome: you pay only when the system books a job or captures a qualified lead. Attractive on paper, but check exactly what counts as a booking before you sign.
What you will actually pay
Real ranges rather than invented precision. Across current vendors, monthly plans for a small business sit roughly between $25 and $300, with most owners landing in the $50 to $150 band for a full-featured setup. Entry tiers start near $25 to $30 for a small bundle of calls or minutes. Per-minute rates commonly run about $0.25 to $0.50, and per-call pricing tends to fall between roughly $0.75 and $2.40 a call. These figures move constantly, so treat them as a map, not a quote, and confirm today's price before you commit.
The comparison that matters is against a human answering service, not against nothing. Live-answer services typically charge on the order of a dollar or two a minute, and a heavy month can run several hundred to well over a thousand dollars. An AI line handling the same volume usually lands far below that. The trade is coverage and cost against nuance: the human handles the awkward call better, the AI never sleeps and never puts a caller on hold.
Costs that hide outside the monthly fee
- Setup or onboarding fees to script the greeting, questions and call flow.
- Overage rates once you pass the included minutes or calls, which spike in your busy season.
- Calendar, CRM or texting integrations, sometimes gated behind a higher tier.
- Number porting or a new dedicated line, and any SMS charges for confirmations.
The payback maths, worked through
The whole case rests on one question: what is a missed new-customer call worth to you? Not an existing client checking an appointment, but a first-time caller with a job. That number is the entire argument, because a single captured job usually dwarfs a month of subscription.
Work it in four steps. First, your average new-job value: a service call might be a few hundred dollars, a roof replacement several thousand, and a first job often becomes a repeat customer worth far more over the years. Second, how many genuine new-customer calls you miss in a month to voicemail, a busy line, or after hours. Third, the share of those callers who simply ring the next business rather than call back, which your own experience probably confirms is high. Fourth, compare the revenue on those saved jobs against the monthly fee.
The maths is usually lopsided. If your average new job clears even a few hundred dollars in profit and the AI recovers one caller a month who would otherwise have gone elsewhere, a $100 plan has paid for itself with room to spare. For a trades business where a single job runs into the thousands, one saved call can cover a year of the fee. The break-even is not close, which is exactly why the sums are worth doing on your own real numbers rather than a vendor's.
Plug in your own figures
- Average profit on a new job, not revenue, and not lifetime value yet.
- New-customer calls missed per month, counting only first-time job enquiries.
- The realistic share of missed callers who never ring back.
- Monthly fee plus expected overage, so you compare like with like.
What it does well, and where it falls over
Be clear-eyed about the edges. An AI receptionist shines on volume, consistency and hours. It answers every call in the same steady voice, it never loses a message, it works at 11pm and on a Sunday, and it turns a busy signal into a booked slot. For qualifying, scheduling and simple frequently-asked questions, it is genuinely reliable.
It falls over on the calls that need a human. Anything complex, unusual, or emotionally charged is where it struggles: an upset customer, a delicate complaint, a technical judgement call, a heavy accent on a noisy line, a caller who goes off-script. Push it to handle those and it will either give a wrong answer with confidence or frustrate someone who was ready to buy. The failure mode is not that it hangs up. It is that it sounds fine while quietly mishandling the one call that mattered.
Good fits and bad fits
- Good: after-hours and overflow cover, appointment booking, lead qualification, routine hours and pricing questions.
- Good: high call volume where the alternative is voicemail or a permanently engaged line.
- Poor: complaints, emotional or sensitive calls, and anything needing genuine judgement.
- Poor: complex custom quotes, edge-case questions, and callers who need reassurance from a person.
The handoff to a human is the whole game
The difference between an AI receptionist that helps and one that costs you customers is the handoff. A good setup recognises when it is out of its depth and routes the caller to a person cleanly, either by warm transfer during working hours or by flagging the call for an urgent callback out of hours. A bad setup traps the caller in a loop, repeats itself, and never admits it cannot help.
Design the escape hatch before you design the script. Decide which calls always go to a human, how quickly a flagged lead reaches someone's phone, and what the AI says when it hands over so the caller does not repeat everything. Test it by calling your own line and deliberately being difficult. The measure of the system is not how it handles the easy call. It is how gracefully it fails the hard one.
A clean handoff needs
- Clear rules for which calls transfer to a person and when.
- A fast, reliable path for urgent leads flagged after hours.
- Context passed on, so the caller is not asked the same questions twice.
- A plain admission when it cannot help, rather than a confident wrong answer.
Common questions
- Is an ai receptionist worth it for a one-van business?
- Often yes, because a solo operator is the person least able to answer while under a sink or on a roof. If you miss even a couple of new-customer calls a week to voicemail, a $50 to $100 plan usually recovers more in booked work than it costs. Run your own numbers on average job value and missed calls before deciding.
- Will callers know they are talking to a machine?
- Many will, especially at the start of the call, though current voices are far better than the old menu systems. What matters more than fooling anyone is that the caller gets their question answered or their appointment booked quickly. A quick, useful AI beats a slow human, and both beat voicemail. Some businesses simply state up front that it is an automated assistant.
- How many calls do i actually miss?
- Almost always more than you think, because you never see the ones that ring out while you are busy or closed. Check your phone logs for unanswered and after-hours calls over a couple of weeks. Estimates of how many callers never ring back vary widely and lean high, so even a conservative figure tends to justify the fee.
- What happens on a complicated or upset call?
- This is the weak spot, and you should plan for it rather than hope. A well-configured system recognises it is out of its depth and either transfers to a person or flags the call for an urgent callback. If a vendor claims their AI handles everything, treat that as a warning sign and test the handoff yourself.
- Can it book straight into my calendar?
- Usually, but check before you buy. Good systems connect to common scheduling and CRM tools and drop the appointment straight in, which is where most of the time-saving comes from. Confirm your specific calendar is supported, and that the integration is included in the tier you are pricing rather than an upgrade.
- Is per-minute or a flat monthly plan cheaper for me?
- It depends on your call length and volume. Per-minute favours short calls and quiet months but stings when calls run long or volume climbs. A flat plan is more predictable and usually better once you get steady traffic. Estimate your monthly minutes from your phone logs, then price both shapes against that number including overage.