The Rounding Trap: How Billing Increments Quietly Change Your AI Answering Bill

Two answering services quote you the same rate, ten cents a minute, and one of them costs you half again as much at the end of the month. The difference is not the rate. It is the billing increment: the unit of time the vendor rounds every call up to.

This is the line item almost nobody asks about on the sales call, and it is one of the biggest quiet drivers of an AI answering bill. Here is how it works and what to check before you sign.

What a billing increment actually is

When a vendor says “per-minute billing,” that does not necessarily mean you pay for the seconds you used. It means you pay per billing increment, and the increment is whatever the vendor defines. The common ones are full-minute rounding (60/60), half-minute increments (30/30), and true per-second metering.

Full-minute rounding is the oldest scheme in telecom, and it is still common with legacy human answering services. A 61-second call bills as two full minutes. A 12-second wrong-number call bills as one full minute. If half your calls are short, which is normal for small businesses, full-minute rounding inflates the bill substantially. One voice-AI pricing analysis published this year estimated that full-minute rounding inflates telephony invoices by 35 to 50 percent in high-volume scenarios where many calls last 15 to 30 seconds.

Half-minute increments cut that waste roughly in half. RingCentral’s AI Receptionist, for example, bills overage at $0.50 per minute in 30-second increments: a 61-second call bills as 90 seconds, not two full minutes. Per-second metering is the cleanest model: a 42-second call bills as exactly 0.70 of a minute. When you compare vendors, this detail matters more than a two-cent difference in the headline rate.

Where the extra seconds hide

Rounding is not the only place time gets added. A few practices are standard in the industry and rarely mentioned on pricing pages:

Message-writing time counts as billable. Several human answering services bill the time the agent spends writing up your message after the caller hangs up. AnswerConnect, for instance, bills in whole-minute increments and counts the wrap-up time after the call toward your minutes. An AI service has the same issue in a different form: the seconds the AI spends generating the summary or dispatching the text can be metered on some platforms.

Short interactions may or may not be waived. AnswerConnect does not charge for the first 30 interactions under 30 seconds in a billing cycle, which is a genuinely buyer-friendly policy for wrong numbers and hang-ups. Other vendors bill every connection from the first second, including the 8-second call where someone dialed the wrong number.

Voicemails and failed calls can be billed. If your AI places an outbound call that reaches voicemail, or a call fails to connect, some platforms still count the attempt. This matters most for outbound-heavy setups like appointment-reminder campaigns, where a meaningful share of calls never reach a person.

Do the math on your own call pattern

Here is a quick way to see whether the increment matters for you. Pull your last month of calls and sort them into buckets: under 30 seconds, 30 to 60 seconds, 60 to 90 seconds, and so on. For each bucket, multiply the call count by the increment the vendor bills in, then compare that to the actual seconds.

A concrete example. Say you get 200 calls a month and half of them are 25-second calls (quick questions, wrong numbers, brief confirmations) and half are 90-second calls (real conversations). Actual usage: 100 x 25 seconds plus 100 x 90 seconds, which is about 192 minutes. Under full-minute rounding: 100 x 1 minute plus 100 x 2 minutes, which is 300 billed minutes, about 56 percent more than you used. Under 30-second increments: 100 x 0.5 plus 100 x 1.5, which is 200 minutes, within a few percent of actual. At $0.50 a minute in overage, that is the difference between $100 and $150 on the same calls.

The pattern to notice: the shorter your average call, the more the increment matters. Businesses with quick transactional calls, like appointment confirmations and hours questions, get hit hardest by full-minute rounding. Businesses with long consultative calls barely notice it.

What to ask before you sign

Four questions, asked on the sales call or read in the terms of service, tell you everything:

First, what is the billing increment? Per-second, 30-second, or full-minute. Get it in writing, not just spoken.

Second, what counts as billable time? Call duration only, or does wrap-up, transcription, and message delivery count too?

Third, are short or failed calls waived? Specifically: calls under 30 seconds, wrong numbers, hang-ups, voicemails, and failed connections.

Fourth, can you see the increment applied? Ask for a sample invoice or a usage report from the trial that shows billed seconds next to actual seconds. If the vendor cannot show you that, take it as a warning sign.

How this changes vendor comparison

When you line up quotes, normalize them. Take the vendor’s published rate and multiply it by your estimated rounding factor: 1.0 for per-second, about 1.05 to 1.15 for 30-second increments on typical small-business call patterns, and 1.25 to 1.5 for full-minute rounding if your calls skew short. Then compare the adjusted rates. A vendor at $0.12 a minute with per-second metering is cheaper than a vendor at $0.09 a minute with full-minute rounding for most small businesses, and it is not close.

This is also why bundled-minute plans can be deceptive. A plan that includes 200 minutes sounds generous until you realize those are 200 billed minutes under full-minute rounding, which might cover only 130 minutes of actual talk time. The bundle is measured in the vendor’s currency, not yours.

The bottom line

The headline rate is the opening bid. The billing increment, what counts as billable, and whether short calls are waived decide the real price. Spend ten minutes on the math before you sign, and you will know which of the two “ten cents a minute” vendors is actually cheaper for your call pattern.

For the full picture on how the models differ, read our guide to per-call, per-minute, and unit billing, our method for estimating your monthly bill before you sign up, and our explainer on overage fees, where rounding does its real damage.