Answering Service Cost: Per-Minute vs Flat-Rate AI (2026)

Answering Service Cost: Per-Minute vs Flat-Rate AI (2026)

Every owner who has ever shopped for phone coverage has run into the same frustrating gap. The vendor's page shows a clean number — a plan name, a monthly price, a bundle of minutes — and the number looks manageable. Three months later the invoice is materially higher than the number on the page, and nobody at your shop can explain exactly why. Nothing shady happened. The plan simply billed the way it always said it would, in a unit you were not watching.

As of August 2026, the honest answer to "what does an answering service cost" is that it depends almost entirely on which unit you are buying. Buy minutes, and your bill tracks how long your callers talk — which you do not control. Buy calls, and your bill tracks how many people dial you — which you also do not control, and which spikes exactly when your marketing works. Buy a flat plan with metered minutes and a stated overage rate, and your bill is predictable up to a line you can see coming. Those are three genuinely different financial products wearing the same label, and the difference between them is the difference between a budget and a surprise.

This article breaks down each billing model in operator terms: what actually counts as a billable minute, how rounding quietly inflates short calls, which activities get billed that you would not think of as "answering," and what the surcharges look like. Then it puts KeyBot's flat plans next to those models honestly — including our own per-minute overage, because we have one and pretending otherwise would be exactly the sort of thing this article is written against.

One note before the numbers. We are not going to publish a competitor's current price as though it were a fact. Published plans vary by market, change without notice, and are frequently quoted differently on a sales call than on the website. If you want to know what a specific vendor charges today, check that vendor's own pricing page and ask for the rate card in writing. What we can do — and what actually helps — is explain the structures those prices sit inside, so you can read any quote correctly.

Why the quoted monthly price is a floor, not a ceiling

Traditional live answering services are staffed businesses. Real people sit in a call center, and the cost of serving you is the cost of their time. That single fact drives every billing decision downstream: the vendor's cost is measured in minutes, so the vendor's price is measured in minutes too. It is a rational model. It is just not a fixed one.

What that means practically is that the plan tier you pick is a prepaid block, not an all-you-can-eat allowance. You are buying, say, a bundle of receptionist minutes for a monthly rate. Use fewer than the bundle and you generally do not get money back. Use more and you pay an overage rate per minute for everything past the line. The quoted monthly price is what you pay in the quietest month you will ever have.

The failure mode this creates is specific and predictable: your bill rises fastest in your best months. A campaign lands, your Google presence improves, a storm or a heat wave or a lockout-heavy holiday weekend triples inbound volume — and the coverage bill scales right along with it, at exactly the moment your cash is committed to parts, fuel, and overtime. Owners who budget from the quoted tier and get billed from actual usage are not being cheated. They are being billed correctly against a model they did not read closely.

Billing model one: per-minute with a monthly bundle

This is the dominant structure in live answering. You pick a tier that includes N minutes; past N you pay a per-minute overage. Two details decide what you actually pay.

What counts as a billable minute. The intuitive assumption is that you are billed for the time a receptionist spends talking to your customer. In practice, industry billing commonly covers a wider set of activities, and the exact list varies by provider. Time that may be included in your metered total can include:

• Talk time with the caller — the obvious one.
• Hold time. If your caller sits on hold while the receptionist finishes another call, that hold may still be your minutes, depending on the provider.
• Transfer and patch time. When the receptionist reaches you or your on-call tech and stays on the line to hand the call over, that bridging time is frequently billable. A three-minute conversation with your customer plus two minutes spent finding you is a five-minute call on your invoice.
• Message dispatch. Time spent typing, reading back, or relaying the message.
• Outbound calls made on your behalf, if your plan includes callbacks or confirmations.

None of this is hidden — it is generally spelled out in the service agreement. But it means the number you are actually buying is "receptionist labor attributable to your account," not "seconds of customer conversation," and the first is always larger than the second.

The billing increment. This is the detail that moves the most money and gets the least attention. Providers bill in an increment: per second, in six-second blocks, or rounded to the full minute. The increment matters enormously for the calls that dominate a service business's day, because those calls are short.

Here is the arithmetic, and it is a labelled hypothetical using round numbers rather than any vendor's real rate. Suppose you take 200 calls in a month and the true average conversation is 70 seconds. At true per-second billing that is 233 minutes. Under minute rounding, every one of those 70-second calls bills as 2 minutes — 400 minutes. Same calls, same customers, same work; 72 percent more billed minutes purely from the rounding rule. Now suppose your plan bundle is 300 minutes: under per-second billing you are comfortably inside your tier, and under minute rounding you are 100 minutes into overage. The plan did not change. The clock did.

The takeaway is not "rounding is a scam" — every metered utility on earth rounds something. The takeaway is that the increment is a price, and you should ask for it explicitly before you compare two quotes, because two vendors with identical per-minute rates and identical bundles can produce bills that differ by half.

Billing model two: per-call pricing

Some providers price by the call rather than the minute — a set number of calls included, then a per-call rate beyond that. Owners often find this more legible, and in one respect it is: you can count calls, and you cannot easily predict how long a distressed caller will talk.

The catch is on the other side. Per-call pricing makes your junk traffic expensive. A robocall that the receptionist answers and disposes of in eleven seconds may bill the same as a genuine four-minute lockout intake. A wrong number bills. A vendor cold-calling you to sell SEO bills. A customer who calls three times in an afternoon chasing a technician bills three times. Under minute-based billing those short nuisance calls cost you almost nothing; under call-based billing they cost you a full unit each.

Whether per-call is cheaper than per-minute for your shop depends entirely on your traffic mix — long real calls and light spam favor per-call; short real calls and heavy spam favor per-minute. If you have never measured how much of your inbound volume is junk, that number is worth pulling before you sign anything, and it is also the number that makes AI screening worth having at all. We covered how much of a typical trade shop's inbound is nuisance traffic in our piece on spam and robocall screening.

The charges that live outside the base plan

Beyond the metered unit, live answering agreements commonly carry a set of line items that never appear in the headline price. These are ordinary industry structures, not gotchas, but you should ask about each one by name:

• Setup or onboarding fees. One-time, sometimes waived on annual commitments.
• Holiday and after-hours premiums. Coverage on major holidays is often billed at a multiple of the standard rate, sometimes with a minimum. This matters disproportionately to emergency trades, because holidays are when your emergency work happens.
• Bilingual or Spanish-language handling, which may sit on a separate rate or a separate team.
• Custom scripting changes beyond a certain number of revisions.
• Per-user or per-extension charges for adding people to the escalation list.
• Integration fees for pushing messages into your CRM or scheduling tool.
• Contract terms. Annual commitments, auto-renewal windows, and early-termination language.
• Minimum monthly spend, which can make the "small" tier effectively unavailable.

You do not need to be adversarial about any of this. You need it in writing, in one document, before you compare. If you are actively evaluating specific vendors in this category, we have written detailed breakdowns of the two most commonly shortlisted ones — see our take on answerforce alternatives and on the posh virtual receptionist alternative for how those models are structured and where the friction usually shows up for trades.

Billing model three: flat monthly AI plans with metered minutes

AI answering changes the cost basis. There is no receptionist whose salaried hour has to be recovered from your account, so the vendor's marginal cost is compute and telephony — which is why AI plans are priced as a flat monthly fee with a generous included allowance and a stated per-minute rate beyond it.

Let us be precise about what that does and does not mean, because this is exactly the claim the industry oversells.

It does not mean unlimited. KeyBot does not sell unlimited calls, does not claim "no per-minute fees," and does have overage. Anyone in this category telling you otherwise is either mispricing themselves or not counting.

What it does mean is that the flat portion of your bill is genuinely flat, the included allowance is sized for real trade volume rather than for a sales demo, and the overage rate is published rather than negotiated. Here are our actual numbers:

• KeyBot Lite — $149/month. A 24/7 bilingual AI message-taking receptionist. 100 calls included; 50¢ per minute after. Messages land in your Telegram with the call recording attached. Live in about ten minutes, no contract, and your number stays. The first 5 answered calls are free on a 7-day trial.
• Core — $500/month. 500 AI minutes included, 45¢/minute overage.
• Pro — $750/month. 1,000 minutes, 40¢/minute.
• Elite — $1,200/month. 2,500 minutes, 35¢/minute.

Core, Pro, and Elite are the full platform — live quoting from your own price sheet, real booking, dispatch, and payments on the call, not just message-taking. Every one of those tiers carries a 14-day free trial. All of it is on the pricing page, and we walk through how the tiers actually differ in our pricing explainer.

Two structural differences are worth naming. First, Lite meters calls, not minutes, for its included allowance — 100 calls, with the per-minute rate applying past that. That makes short nuisance calls cheap by construction. Second, there are no per-seat fees. Add every technician, dispatcher, and office person you want; users are unlimited and do not change the price. In a staffed model, more people on the escalation list is frequently more money.

A labelled hypothetical, run three ways

Numbers make this concrete, so here is a worked example. This is a hypothetical, not a customer. Suppose a four-truck shop takes 240 inbound calls in a month. Suppose 60 of them are spam, wrong numbers, or vendor pitches averaging 15 seconds, and the remaining 180 are real customer calls averaging 2 minutes 10 seconds of conversation plus roughly 40 seconds of transfer and message time.

Under a per-minute plan with full-minute rounding, the real calls bill at 3 minutes each — 540 minutes — and the junk calls bill at 1 minute each — 60 minutes. Total metered: 600 minutes. Whatever tier you bought, that is the number your overage is measured against, and it is more than double the 260 minutes of actual customer conversation that happened.

Under a per-call plan, you are billed 240 units, and a quarter of those units bought you nothing at all.

Under KeyBot Lite, the 100 included calls cover the first 100; the remaining 140 calls bill at 50¢ per minute of actual talk time. The junk calls that the AI screens and disposes of are seconds, not units.

We are not going to tell you which of those lands lowest for your shop, because it depends on rates we are not going to invent for someone else. What the exercise shows is which variables decide your bill: the increment, the junk ratio, and whether your allowance is denominated in the thing that spikes. Run your own volume through the missed-call cost calculator to see the other side of the ledger — what the unanswered calls are costing while you optimize the answered ones.

The models side by side

How to price-check any vendor in ten minutes

Whoever you are evaluating — including us — ask these in writing before you sign. The answers, not the headline, are the price:

1. What is the billing increment? Per second, six seconds, or rounded to the minute.
2. Which activities are metered? Talk time only, or also hold, transfer, patch, message dispatch, and outbound callbacks.
3. What is the overage rate, and is it the same at every tier?
4. Do unused minutes or calls roll over?
5. What are the holiday and after-hours rates, and which days count as holidays?
6. Is Spanish handling included at the same rate, or separate?
7. Are there per-user, per-extension, or per-integration charges?
8. What is the contract term, the auto-renewal window, and the cancellation notice?
9. What is the minimum monthly charge regardless of usage?
10. Can I see a sample invoice from a real account of my size, with the line items intact?

That last one is the most revealing question on the list, and the easiest to ask. A vendor confident in their model will show you the shape of a bill.

For sector-specific versions of this comparison, we have run the same exercise for two trades where answering costs bite hardest: plumbing answering service costs and the dedicated plumber answering service page. If you want the head-to-head framing rather than the arithmetic, AI versus a traditional answering service covers it.

For background on how telephone billing and consumer disclosure are regulated generally, the Federal Communications Commission and the Federal Trade Commission both publish guidance on billing transparency and disclosure practices. Neither publishes a benchmark price for answering services, and you should be suspicious of anyone who claims they do.

Who each model fits

Per-minute live answering fits shops with low call volume, long complex calls, and a strong preference for a human voice — and owners who will genuinely read the monthly usage report rather than autopay it.

Per-call live answering fits shops with clean traffic, few nuisance calls, and long conversations, where paying once per real conversation is simpler than watching a clock.

Voicemail fits nobody in an emergency trade, and we have said so at length in voicemail versus an AI receptionist. It is free, and it is the most expensive option on this page.

Flat-rate AI fits shops whose volume is spiky, whose junk-call ratio is meaningful, who need coverage at 2 AM and on Thanksgiving without a premium rate, who want every technician on the alert list without a per-seat charge, and who would rather have a fixed number on the P&L with a visible overage line than a variable one.

The fastest way to judge the product rather than the pricing model is to hear it. Go to the instant demo, enter your business info, and the AI calls your phone in about 30 seconds answering as your own receptionist — no signup, first demo free. If you would rather dial in yourself, the demo line is +1 (716) 350-6391.

The bottom line

Answering service pricing is not expensive or cheap in the abstract — it is denominated, and the denomination is what you are actually buying. Per-minute plans bill you for receptionist labor including hold and transfer time, rounded by an increment that can inflate a short-call business by half. Per-call plans bill you a full unit for every robocall. Both scale up in your busiest months, both commonly carry holiday premiums, per-seat charges, and annual terms, and both quote a monthly figure that is a floor. Flat-rate AI moves the fixed portion of your bill off the receptionist's clock: KeyBot Lite is $149 a month with 100 calls included and 50¢ per minute after, no contract, no per-seat fees, live in about ten minutes, and your number stays; the full platform with live quoting, booking, and dispatch runs $500 to $1,200 a month with 45¢ to 35¢ overage and a 14-day free trial. We have overage and we publish it, because the vendors who do not are the ones whose invoices surprise you. Before you sign anything, get the increment, the metered activities, the overage rate, and a sample invoice in writing — from us and from everyone else.

Frequently asked questions

How much does an answering service cost per minute?

Per-minute rates vary widely by provider, market, and plan tier, and the published rate is only half the answer — the billing increment and the list of metered activities decide what you actually pay. A plan that rounds every call up to a full minute can bill 50 to 70 percent more than true per-second billing on a business whose calls average around a minute. Always ask for the increment, the overage rate, and whether hold and transfer time are metered, in writing, before comparing two quotes.

Why is my answering service bill higher than the plan price I was quoted?

Because the quoted plan price is a bundle floor, not a cap — anything past the included minutes or calls bills at an overage rate, and several activities you may not think of as answering are commonly metered. Hold time, transfer and patch time while the receptionist reaches you, message dispatch, and outbound callbacks can all count against your total depending on the provider. Holiday premiums, per-user charges, and integration fees also sit outside the headline price, so request an itemized sample invoice from an account your size.

What does TheKeyBot cost, and does it have per-minute charges?

KeyBot Lite is $149 per month with 100 calls included and 50¢ per minute after that, with no contract, no per-seat fees, and your first 5 answered calls free on a 7-day trial. The full platform is $500 for Core (500 minutes, 45¢/min overage), $750 for Pro (1,000 minutes, 40¢/min), and $1,200 for Elite (2,500 minutes, 35¢/min), each with a 14-day free trial. Yes, there are per-minute charges past the included allowance — we publish them rather than calling the plans unlimited, and every figure is at https://www.thekeybot.com/pricing.

Is per-call billing cheaper than per-minute billing?

It depends entirely on your traffic mix, and the deciding variable is usually how much junk you get. Per-call billing favors businesses with long real conversations and very little spam, because you pay one unit for a four-minute intake; per-minute billing favors businesses with heavy nuisance traffic, because a fifteen-second robocall costs almost nothing instead of a full billable call. Measure what share of your inbound is spam, wrong numbers, and vendor pitches before you choose a model.

Do answering services charge extra for holidays and after-hours calls?

Holiday and after-hours premium rates are a common structure in staffed live answering, since the provider is paying people premium wages to be there. Rates and which days qualify vary by provider, so confirm the specific holiday list and multiplier on the vendor's own pricing page or rate card. Flat-rate AI plans like KeyBot do not premium-rate nights, weekends, or holidays — the same $149 Lite plan covers 2 AM on Thanksgiving at the same rate as Tuesday afternoon.

How can I try an AI receptionist before committing to a plan?

Go to https://www.thekeybot.com/try, enter your business information, and the AI calls your phone in about 30 seconds answering as your own receptionist — no signup required and the first demo is free. That lets you judge the voice, the intake questions, and the pacing on a real phone call before spending anything. If you want to hear the product without entering anything at all, the demo line is +1 (716) 350-6391, and Lite itself includes the first 5 answered calls free on a 7-day trial.