When an agency adds its 11th client to an AI voice platform, the overage usually comes not from the client count itself but from whichever meter your plan caps first: sub-account slots, included minutes, concurrent calls, add-on features, or telephony pass-through. Some platforms charge a flat per-client fee past a threshold, others never notice the 11th client until its call volume tips a different meter. The fix is to know which of the five your platform actually bills, before the client is live.
Which meters can trigger overage on an agency plan?
Pricing pages blur these together, so separate them:
- Client or sub-account slots. The plan includes a set number of client workspaces, and each one past that costs extra or forces a tier upgrade. See what a sub-account is if the term is fuzzy.
- Included minutes or replies. Usage is pooled across all clients, so a busy new client drains the pool for everyone.
- Concurrency. The number of simultaneous calls you can run is often capped separately from minutes. Eleven clients rarely ring at once, but an outbound campaign for one of them can fill every line.
- Feature gates. White label branding, custom domains, API tools or analytics may sit on a higher tier than the one you started on.
- Carrier and model costs. Telephony and AI model charges can be bundled, marked up, or billed straight from the provider.
Only the first meter is about the number 11. The other four scale with what the client does, which is why an agency can add client 11 with no surprise and client 12 with a big one.
What does the 11th client look like in practice?
Take a hypothetical plan that includes ten client accounts. Client 11 might trigger a per-account fee, a mandatory move to the next tier, or nothing at all because the plan counts usage rather than accounts. Each outcome changes your margin differently. A per-account fee is predictable and easy to pass into your client's monthly price. A forced tier jump hits your whole book at once, including the ten clients who did nothing. A usage-only plan feels free until the new client is a high-volume one.
That is why the useful number is not "how many clients can I add" but "what does my cost per client look like at client 11, 15 and 25." If you can't answer that from the pricing page, you are guessing your own margin.
How do you check a competitor's pricing for this?
Read their own public pages, not a comparison table from a third party, including ours. Vapi's pricing page and Retell's pricing page both publish how they charge, and these pages change, so I won't restate their rates here. For each one, look for:
- Whether usage is billed per minute on top of provider costs, or bundled.
- Whether concurrency is a plan limit or a paid add-on.
- Whether white label or multi-client features are described at all, and on which plan.
- What happens when you exceed a limit: a hard stop, an overage rate, or a prompt to upgrade.
If a page doesn't say, that is your answer for now. Ask support in writing and keep the reply. Developer voice APIs in particular are built as parts for engineers to assemble, so agency-specific billing such as per-client prices and margin reporting is often something you build yourself. Our Vapi comparison goes through that gap in more detail.
Why does concurrency deserve its own line item?
Because it fails differently from minutes. Running out of minutes means a bill. Hitting a concurrency cap can mean a caller gets a busy signal or a campaign slows down, during the hours your client cares about most. A dental client's Monday morning rush and a roofing client's post-storm spike are exactly when a shared cap gets tested, and your other clients share the same ceiling. The mechanics are in our post on voice AI concurrency limits and agency pricing. The short version: ask whether the cap applies per account or across the whole agency.
How does NovaReps handle the economics of a growing client list?
NovaReps is a white label platform agencies resell under their own brand, with the pieces that decide margin built in rather than bolted on:
- Client accounts and client plans with your own prices, so what you charge each client is separate from what the platform charges you.
- Agency analytics for usage, revenue and margin per client, so client 11's actual profitability is visible after the first week rather than at quarter end.
- Bring your own keys with no per-minute markup, or managed minutes if you prefer one bill. Our BYOK savings tool shows what the difference looks like for your volume.
- Minute and reply packs with optional auto top up, so a heavy month doesn't silently turn into a surprise invoice.
- Each client on its own carrier account (Twilio or Telnyx), which keeps telephony costs attributable to the client that caused them.
How client accounts are counted for each plan, and what happens beyond the included amount, is on the pricing page, which is the source of truth. I'm deliberately not repeating limits in a blog post that will outlast them.
What should you ask before signing client 11?
Run this checklist against any platform, including ours:
- What exactly is counted as a client: a sub-account, a phone number, an agent, or a seat?
- What is the cost of one more client, and does it change the price for existing ones?
- Is the minute pool shared, and can one client's campaign exhaust it?
- Is concurrency capped per client or across the agency?
- Which features required for resale (branding, domain, API tools, webhooks) sit on which plan?
- Can I see margin per client without exporting to a spreadsheet?
Then price the answers into the client plan before you quote. If the new client's monthly fee doesn't cover the marginal cost at your worst-case usage, raise the price or cap their usage rather than absorbing it. Our guides on pricing white label voice AI to SMB clients and the actual margin on reselling cover how to do that without eating your markup.
Is overage always bad?
No. Overage that scales with real usage, billed at a rate you've already built into client pricing, is a healthy sign: the client is using the product. The problem is overage you can't predict or pass through. An agency that knows its marginal cost per client, per minute and per concurrent call can take on client 11 confidently. One that doesn't is running a margin experiment on a paying customer.