Which Voice Agent Plan Fits a Business Taking 500 Calls a Month?

A 2,000-minute tier, not a 500-minute one. This is the mistake that catches almost everyone: plans are sold in minutes, not calls. At a four-minute average, 500 calls is 2,000 minutes — four times what a 500-minute starter plan covers. On Advantage Labs' published pricing that means the $997 Growth plan, not the $497 Starter.

Why 500 calls is not 500 minutes

Every published tier meters minutes while every business counts calls, and the gap between the two is a multiplier of three to six. A business that reads "up to 500 minutes/month" and thinks "that covers my 500 calls" has under-bought by roughly 75%, and finds out in the second invoice.

This isn't a subtle trap. It's the single most common sizing error in voice AI, and the numbers are unforgiving. Take the plans on Advantage Labs' voice agent pricing page: $497/month for 1 agent and up to 500 minutes, $997/month for up to 3 agents and up to 2,000 minutes, with Enterprise custom.

Now run 500 calls against them. At a 1.5-minute average you need 750 minutes — already 250% over Starter. At 2 minutes it's 1,000, at 3 minutes 1,500, at 4 minutes a flat 2,000. Growth absorbs all of those, though 4-minute calls fill it exactly with no headroom left.

Past that point Growth breaks too. Five-minute calls need 2,500 minutes and six-minute calls need 3,000, both of which push you to Enterprise. Starter fails at every single one of these averages, without exception.

There is no realistic average call length at which 500 calls fits a 500-minute plan. You'd need every call to finish in 60 seconds with zero variance. Even a purely informational agent answering "what are your hours" doesn't hold that line once callers ask a second question.

How do you convert your call volume to minutes?

Multiply your monthly call count by your actual average call duration, then add 20% headroom. Do not use a benchmark for the average — pull it from your own phone system, because this one number determines which tier you buy and it varies more by industry than anything else in the calculation.

The published benchmarks give you a sanity range, not an answer. Call-centre benchmark data puts average handle time around six minutes, with customer-service talk time typically three to six minutes and healthcare running longer. AI-handled calls tend to sit at the shorter end, because the agent doesn't chat and doesn't put anyone on hold — but "tends to" is not your number.

The formula:

1. Pull 90 days of call data from your phone system. Not one month — seasonality will lie to you.

2. Take total inbound minutes ÷ total answered calls. That's your real average.

3. Multiply by projected monthly calls. Use your busiest recent month, not the average month.

4. Add 20%. Covers growth, a bad weather week, and the fact that AI agents get more calls once they stop sending people to voicemail.

5. Check what's billed. Ring time, hold time, and silence often count. Ask.

That last step moves the number more than people expect. If your platform meters from the ring rather than from the answer, and your greeting runs eight seconds, you're paying for time before the conversation starts on every single call.

The number that isn't in your phone report

Your current average is measured on the calls you answered. If you're deploying an agent specifically to catch missed and after-hours calls, those calls are additive and they aren't in your baseline. A practice answering 350 of 500 inbound calls and buying for 350 has under-sized by 30% before accounting for anything else.

Which tier actually fits 500 calls a month?

The 2,000-minute Growth tier at $997/month, for essentially any 500-call business with an average call between two and four minutes. Below two minutes you have real headroom; at five minutes or more you're into Enterprise territory and should get a custom quote rather than hoping.

The effective rate is what makes this clear. Starter works out to $0.99 per minute; Growth is $0.50 per minute at full utilisation. You're not paying twice as much on Growth — you're paying half as much per minute for four times the volume. That's a normal tier structure, and it means the smaller plan is only cheaper if you genuinely stay under 500 minutes.

Growth also carries the features a 500-call business needs rather than wants: CRM integration, lead qualification flows, custom persona and voice, priority support. Starter includes appointment booking, email and SMS confirmations, and standard support. At 500 calls a month you're running a real front-desk function, and no CRM write-back means someone re-types every call into your system by hand.

Where I'd push back on the obvious recommendation: if your 500 calls are mostly 90-second "are you open / where are you / can I get a quote" calls, that's 750 minutes and neither tier fits neatly. Growth gives you 1,250 minutes of unused headroom you're paying for. That's the case to ask about a mid-point or a metered arrangement instead of accepting the published ladder.

What does a 500-minute plan really cover?

A 500-minute plan covers between 100 and 330 calls a month depending on call length — realistically 4 to 11 calls a day. That's an after-hours overflow deployment, not a front desk.

At a 1.5-minute average you get 333 calls a month, about 11 a day. Two-minute calls bring that to 250 a month, roughly 8 a day, and three-minute calls to 167, or about 6. Four-minute calls leave you 125 a month, and at five minutes you're down to 100 — three a day.

Read those numbers before you buy the entry tier. A 500-minute plan is correctly sized for a business catching the calls it currently misses — evenings, lunch, weekends — while staff keep answering during the day. That's a legitimate and often smart first deployment. It is not "our phones now run on AI."

Is a flat monthly plan cheaper than per-minute billing?

At 2,000 minutes, raw per-minute pricing is cheaper on paper and the comparison isn't apples to apples. Platform rates for 2,000 minutes land roughly between $260 and $800; a managed plan at $997 costs more because it includes configuration, integration, and someone to call.

The published component rates are public. Retell — the platform Advantage Labs builds on — lists $0.07 to $0.31 per minute for AI voice agents, plus $2/month per phone number, $8/month per concurrent call beyond 20, and add-ons like PII removal at $0.01/min.

Broader market surveys put managed all-in-one platforms at $0.25–$0.50 per minute and infrastructure providers at $0.05–$0.15 before component costs, with setup fees commonly $500–$2,000 and integration work $1,000–$5,000.

So here's the honest comparison at 2,000 minutes a month. Self-assembling at $0.13/min all-in runs about $260, and buys you nothing beyond the minutes — you build, integrate, and maintain it. A managed platform at $0.25–$0.31/min runs $500–$620 and gives you the platform without configuration or strategy. The $997 managed plan adds configuration, CRM integration, persona work, and support.

Whether that spread is worth it depends entirely on whether you have someone who'd otherwise do the integration work. If you have a technical person with spare capacity, self-assembling at $0.13/min is genuinely cheaper and you should say so out loud rather than pretend otherwise. If you don't, the $400–$700 monthly difference buys the thing actually working, and unbilled hours from your operations manager are not free.

What hidden charges change the real bill?

Five, and the first two are close to universal. Silence and hold time billed as talk time, per-call rounding, concurrency caps, overage rates, and integration surcharges. Any one can move a quoted price by 20% or more.

• Silence, ring, and hold time. Most platforms meter the whole interaction. On a two-minute call with thirty seconds of dead air, a quarter of that minute charge bought no conversation.

• Rounding. Per-call rounding up to the next minute costs you nothing on a six-minute call and roughly 40% on a 90-second one. It punishes exactly the short-call businesses that would otherwise fit a small plan.

• Concurrency. Tiers cap simultaneous calls. Volume is bursty — everyone calls at 9am — so you need peak concurrency, not average. Retell's published rate is $8/month per slot beyond 20; other platforms charge $8–$15.

• Overage. Commonly 20–50% above your base rate, and reported at 2–3x on some platforms. This is what turns an under-sized plan into a bad month.

• Setup and integration. Often quoted separately or waived at signing and charged on the second integration.

Advantage Labs' pricing page states "no hidden fees, no long-term lock-in" and doesn't publish an overage rate. That's a good sign and an open question — get the overage number in writing before you sign, because an unpublished rate is not the same as no rate.

What should you ask before signing?

Seven questions, and the first four are about billing mechanics rather than price. A vendor who answers all seven crisply is one who has thought about your bill; hesitation on questions 1 through 3 tells you where the surprise lives.

1. What exactly is a billed minute? From the ring or from the answer? Per-second or rounded up per call?

2. Is silence, hold time, and transfer time billed?

3. What's the overage rate per minute past my included minutes?

4. How many concurrent calls does my tier include, and what's each extra?

5. Are setup, integration, and CRM connection included, or quoted separately?

6. What happens in month one if I've mis-sized — can I move tiers mid-cycle without penalty?

7. Can I see a sample invoice at my projected volume?

Question 7 does the most work. A sample invoice at 2,000 minutes forces every fee into one number and makes vendor comparison trivial. Ask three vendors for one and the cheapest sticker price usually stops being the cheapest bill.

One inconsistency worth raising if you're comparing published sources: the cost analysis on this site states AI receptionist services "typically run between $30 and $300 per month," which sits well below the $497 entry tier on the same site's pricing page. The $30–$300 range describes basic call-answering tools, not a configured agent with CRM integration. Take the cost comparison against front desk staff as the labour-cost argument and the pricing page as the actual price.

Frequently asked questions

How many minutes do 500 calls actually need?

Between 750 and 3,000, depending on average call length. At two minutes it's 1,000; at four minutes it's 2,000; at six minutes it's 3,000. Multiply your real average by your call count, then add 20% headroom. Never assume one call equals one minute — that error under-buys by 200% or more.

What happens if I go over my included minutes?

Depends entirely on the contract, which is why you get the rate in writing first. Overage commonly runs 20–50% above base rate and has been reported at 2–3x on some platforms. Some vendors throttle or stop answering instead of billing, which is worse — your agent goes dark mid-month.

Should I start on the smaller plan and upgrade?

Only if your call volume genuinely fits it, or if you're deliberately deploying after-hours-only first. Starting small to "test" a 500-call operation means overage from week two. A better test is to route only evenings and weekends to the agent for a month, which does fit 500 minutes.

Is 500 calls a month a lot?

It's a real front desk — roughly 17 calls a day, or 23 per working day. It's past the point where an entry tier makes sense and well short of enterprise volume. Businesses at this level are typically multi-provider practices, small field-service companies, or single-location firms with heavy inbound demand.

Do unused minutes roll over?

Usually not. Most plans are use-it-or-lose-it monthly, which is why buying 2,000 minutes for 750 minutes of real usage is a genuine waste rather than a safety margin. Ask specifically about rollover, and if there is none, size to your busy month rather than your worst-case month.

How do I count calls that the agent transfers to a human?

Ask, because vendors differ and it matters at 500 calls. Some bill the full duration including the time after transfer; some stop metering at handoff. If your escalation rate is 20%, this single detail changes your effective minute count meaningfully.

What about multiple locations or phone numbers?

The agent count matters as much as minutes. The $997 tier includes up to three voice agents, which typically maps to three locations or three distinct call flows. A fourth location pushes you toward Enterprise regardless of whether you're using the minutes.

Conclusion

Sizing a voice agent plan comes down to one calculation, not a comparison of headline numbers: your real average call length times your busiest month's call count, plus 20% headroom. For most businesses at 500 calls that lands on a 2,000-minute tier, the effective rate per minute is lower there than on the starter plan anyway. Before you sign with anyone, get the billed-minute definition, the overage rate, and the concurrency allowance in writing, and ask for a sample invoice at your projected volume. Advantage Labs can run those numbers against your own call data and tell you which tier actually fits rather than which one looks cheapest. Schedule a consultation with Advantage Labs once you've pulled your last 90 days of call reports.