What happens when an answering service gets too many calls?

Two things run out, and they are not the same thing. A staffed answering service runs out of the minutes in your monthly block, which is a billing event you find out about later, and it runs out of receptionists who are free at that second, which is a hold queue your caller hears right now. Published per-minute overage across the ten staffed services in our price index runs from $1.90 to about $4.69 CAD once the US rows are converted, and not one of those ten publishes how many calls it will take at the same time. BotsDontSleep has no block to exceed and no queue — every call is answered at once, 24/7, at $0.60 CAD per answered minute billed by the second. A spike still costs money. It stops costing you a busy signal.

What actually happens when an answering service gets too many calls?

Two limits bite, in a different order, and most shortlists only check one of them.

The first is the monthly minute block. You bought 100 minutes; March brought 220. Nothing stops on the call — the service keeps answering and the extra minutes arrive on the invoice at the plan's overage rate. That is a cost surprise, not a service failure.

The second limit is the one your customer experiences. A staffed service answers with people, and the number of people free in any given sixty seconds is finite. When the fourth caller rings and every receptionist is mid-call, that caller waits. How long depends on staffing you cannot see and the provider does not publish.

The distinction matters because the fixes are opposite. A block that runs out is solved by buying a bigger block. A queue that forms at 8:40 on a Monday is solved by having more capacity at 8:40 on a Monday, and no plan upgrade buys you that directly.

What happens when you run out of plan minutes mid-month?

In most cases the service carries on and bills you the difference, which is the humane version and also the expensive one.

AnswerConnect's own billing help page states it plainly: "Your service will continue as normal if you use all the minutes included in your plan," with extra minutes "charged at your plan's overage rate," and notifications at 80% and 100% of the allowance (read 22 October 2026). That is the common pattern across the staffed services we price-checked.

Two providers in the index behave differently and both are worth knowing about before you sign. Aria, a Canadian AI service, publishes no overage rate at all — its agent pauses when the minutes run out unless you add a minute pack, so an unwatched inbox on a busy week means an unanswered line. Dialbox prices per AI-handled call with no overage billing, and sustained use above the cap moves you up a plan instead. Ruby publishes plans but no overage rate, so the figure only appears on your first big invoice.

Figures below were read from each provider's own pricing page or rate sheet on 15 October 2026; the full twenty-provider set with its sources and method sits on our answering service price index, and any provider's current page is the authority over our reading of it.

What does a 120-minute spike actually cost?

Between about $473 and $1,028 CAD on a 100-minute staffed plan, against $132 for the same 220 minutes at $0.60.

The arithmetic is plan price plus 120 extra minutes at the plan's published overage rate, with US rows converted at the Bank of Canada daily rate of 1.4196 on 15 October 2026. The BotsDontSleep row has no plan, so all 220 minutes are billed.

Two details move these numbers more than the rate does. AnswerConnect rounds any call over 30 seconds up to the whole minute — its own example bills a call of 1 minute 25 seconds as two minutes, which at $2.95 is $5.90 for 85 seconds of talk; how that rounding compares with per-second billing is the single largest line-item difference between the two models on a line full of short calls. And Smith.ai's live service bills per call rather than per minute, at $11.50 USD for each call above a 30-call Starter bundle, so a spike there is counted in conversations, not minutes.

What happens to the caller who rings while every receptionist is busy?

They wait, and the length of the wait is the one number nobody in this market publishes.

We read the pricing page or rate sheet of all twenty providers in the index. Ten of them staff with human receptionists. Not one states how many calls it will answer at the same time, what a caller hears when that number is reached, or how long an average hold runs at peak. Hours are published everywhere. Rates are published almost everywhere. Concurrency is published nowhere.

That silence is not sinister — staffing varies hour by hour and a published figure would be wrong by lunchtime. But it means the capacity question cannot be answered from a website, only from a direct question and a test call, which is what the next section is for.

It also means the spike you are buying cover for is usually the one that breaks the arrangement. A roofing company's phone does not ring evenly; it rings forty times the morning after a windstorm and twice the following Tuesday. A Vancouver plumbing line does the same thing during a hard freeze. The month you most need every call answered is the month your block runs out and the queue is longest, because the same weather is calling everybody.

How do you find out what your service does at peak before you sign?

Five questions, in writing, before the contract. The answers take a working day to collect and they are worth more than any comparison table, including the one above.

1. Ask how many calls you can have in progress at once. Phrase it as a number, not a capability: "if six of my customers ring in the same minute, how many are talking to a person?" A service that will not give a number is telling you something.
2. Ask what the seventh caller hears. Hold music, a queue position, a voicemail prompt, or a busy tone — all four exist in this market, and only one of them keeps the caller.
3. Ask for the overage rate in writing, with the rounding rule. The rate alone is half the answer. A $2.09 rate with minute rounding costs more than a $2.49 rate billed by the second on a line of 40-second calls.
4. Ask what happens at 100% of the plan. Carry on and bill, pause the service, or auto-upgrade the tier. All three are live practices in this index, and the difference only shows up on the week you are busiest.
5. Place three test calls at once, from three phones, at your real peak hour. Not at 2pm on a Wednesday. Whatever you hear on the third line is what your customer hears in March.

Keep the written answers. When the first busy month arrives, they are the difference between a conversation about the invoice and an argument about it.

When does answering every call not help?

Three cases, plainly, and this is the part a comparison table cannot show you.

Volume relief is not demand relief. If forty people ring because a shipment is late or your booking system is down, answering all forty well still leaves forty problems. The AI will take forty accurate messages, and somebody on your team still works through them. In that situation an answering service converts a phone emergency into an inbox backlog, which is better, but it is not solved.

Per-minute billing has no ceiling, and a monthly plan does. That cuts against us. A 100-minute plan caps your exposure at the plan price if you are willing to accept hold times and a block that runs dry — you know the worst case in advance. At $0.60 a minute there is no worst case; a genuinely enormous month costs genuinely more. On the numbers above that trade is overwhelmingly in favour of per-minute pricing, but if a hard monthly ceiling matters more to you than answering everybody, a capped plan is the honest choice and you should buy one.

And if every call in your spike needs a human decision — a price only you can authorize, a dispatch only you can make — then removing the queue buys you a faster route to the same bottleneck. What it still buys is a caller who has not rung your competitor while waiting, which is worth real money. It is less than it sounds like if you expected the surge to be handled rather than held.

Below roughly twenty calls a month that you comfortably answer yourself, buy nothing. A spike you can take on your own phone is not a capacity problem.

What does BotsDontSleep do when the calls arrive at once?

It answers all of them, in the same second, because each call runs as its own conversation rather than waiting for an agent to come free.

There is no plan to exceed, so there is no overage rate to look up and no 80%-of-allowance email. Forty calls on a storm morning are billed as forty calls: a 90-second booking is 90 cents, a 40-second screened robocall is about 40 cents, and the quiet Tuesday after costs nothing. Calls are answered in 90+ languages, including a caller who switches language mid-sentence, and the per-second billing rule is what makes a surge of short calls affordable rather than merely answered.

What it does not do is decide anything your rules do not cover. Each call that needs a person is transferred or escalated with a transcript, and on a forty-call morning that still means a stack of escalations waiting for you. Agents are tuned on your services, hours, prices and escalation rules over about a week before a real customer hears one, and for a line that spikes overnight the after-hours escalation rule deserves more attention than the rate does.

The cheapest way to find out what your own peak costs is to point only your overflow and after-hours calls at it for a month, then read the invoice against your call log. No block to size, so the trial measures your real spike rather than your guess at one.