The short answer.
An AI receptionist has two costs, and most pricing pages only show you one of them. There is the cost of running the line every month, and there is the cost of making it say the right things about your business. The first is usually small. The second is where the real work sits, and it decides whether callers get a useful answer or a polite dead end.
My own published numbers are on the AI receptionist page: setup from $900 to $2,500, and running costs that are typically under $30 a month at small volumes, billed to your own account rather than marked up. The rest of this note explains how the market prices this so you can compare any offer, including mine, on equal terms.
The pricing models you will run into.
Almost every offer you will see is one of a handful of shapes. They are not better or worse in themselves. What matters is how each one behaves in your busiest month, because that is the month the line earns its money.
- A monthly plan with a bucket of minutes. You pay a flat fee and get a set number of call minutes. Simple to budget, but check what happens when a storm week or a heat wave blows through the bucket.
- Pure per minute billing. You pay for exactly the talk time you use. Cheap in a quiet month and honest about volume, but harder to predict.
- Per call billing. Every answered call costs the same regardless of length. Good if your calls are short, expensive if a lot of them are wrong numbers or spam.
- A setup or onboarding fee. A one time charge for writing the script, connecting the number and testing it. Some providers fold this into a higher monthly price instead.
- Human answering services. Staffed call centers usually bill per minute or per call as well, often with a monthly minimum. They are a different product, and I compare them separately in another note.
What the setup money actually buys.
A self serve tool can be live in an afternoon. The question is what it says when a real customer calls with a real problem. Setup is the work of deciding that: what your business does and does not do, which areas you cover, what counts as urgent, what details you need to call back properly, and the list of things the line must refuse to answer.
That refusal list is the part people skip, and it is the part that protects you. A line that quotes a price you never agreed to, promises a time slot you do not have, or guesses at an answer about your service is worse than voicemail. Most of my setup time goes into that list and into testing the awkward calls, not into connecting software.
The other half of setup is deciding where the message lands. A receptionist that takes a perfect message and drops it into an inbox nobody checks has not solved anything. The handoff, a text to your phone, a record in the CRM you already use, an urgent flag for the calls that cannot wait, is decided before anything gets built.
The running costs, honestly.
Once it is built, the monthly cost of an AI line comes from two places: the phone number and the per minute charges for the calls it handles. For a small business that misses a few calls a day, that usually comes to under $30 a month. It goes up with volume, and in a storm or a heat wave it will go up, which is the point: those are the calls you wanted answered.
I put those accounts in your name, on your card, at cost. You can see every charge yourself. If you ever stop working with me, the number and the line are still yours. I would ask the same question of any provider: whose name is the account in, and is the per minute charge marked up?
How to tell whether it pays for itself.
Skip the industry averages. The only numbers that matter are yours: how many calls you miss in a typical week, what an average job is worth to you, and what share of those missed callers simply phone the next company on the list.
Put those three numbers into the missed call calculator and you get a monthly figure you can check with your own arithmetic. If that figure is smaller than what it would cost to fix the problem, do not fix it. Some businesses genuinely do not have a missed call problem, mostly the ones where callers are existing customers who will wait for a callback.
For a trade where an urgent caller who reaches voicemail calls the next name, the arithmetic usually settles the question quickly. One recovered job a month often covers the running cost many times over. That is a calculation for your business, not a promise I can make for it.
Costs that do not show up on the pricing page.
A few costs are real even when nobody bills you for them. Plan for them up front and the line stays useful a year from now instead of slowly drifting out of date.
- Your time in the first two weeks. Someone has to listen to test calls and say what sounds wrong. An hour or two of your attention saves a lot of awkward calls later.
- Keeping the script current. Prices, service areas and hours change. A line that still describes last year's business costs you trust.
- Text messaging registration. If the line sends texts from a regular local number, US carriers expect that business texting to be registered, and unregistered traffic gets filtered.
- The follow up nobody does. Answering the call is half the job. A captured lead that never gets a callback is still a lost job.
Questions to ask before you pay anyone.
If you are choosing between a self serve plan, someone like me, or a human service, the same questions separate a good offer from an expensive one. Ask what happens when the monthly minutes run out. Ask who owns the phone number. Ask to hear a test call about something the line should refuse to answer. Ask where the message goes and how fast. And ask whether the line will say it is an AI if a caller asks. Mine always will.
If the answers are clear, the price is usually fair. If they are vague, the monthly number on the pricing page was never the real cost.
