A sub-account is a self-contained workspace inside a white-label AI voice platform, one per client, that holds that client's phone number, voice agent configuration, call logs, and calendar connections separately from every other client the agency runs. The agency owns one master account. Each sub-account nests underneath it, branded and billed however the agency decides, while the underlying platform stays invisible to the end client.
Why do agencies need sub-accounts instead of one shared account?
Running every client inside a single account means every client's calls, prompts, and data sit in the same bucket. That's fine for a solo operator testing a demo, but it breaks the moment an agency has more than one paying client, for a few concrete reasons:
- Isolation. A misconfigured prompt or a bad phone number for Client A shouldn't be able to touch Client B's live calls.
- Billing. Each sub-account can carry its own usage, minutes, or seat count, which is what lets an agency actually track margin per client rather than guessing.
- Access control. A client's staff, if given a login at all, should only ever see their own workspace, never a neighbor's call history.
- Reporting. Per-client dashboards and exports need a clean data boundary, otherwise every report needs manual filtering.
This is the same structural pattern most white-label SaaS categories use, and it's worth reading in more depth on what makes a platform white-label in the first place versus just rebranded, covered in what is white-label AI voice software.
What lives inside a single sub-account?
A sub-account typically holds the pieces that make one client's voice agent function independently:
- The client's dedicated phone number and carrier configuration
- Its own voice engine and prompt settings (which LLM and TTS pairing it runs)
- Calendar connections for booking (Google or Microsoft)
- Call recordings, transcripts, and logs scoped only to that client
- Webhook and integration settings specific to that account
Because each sub-account's carrier setup is isolated, one client's number provisioning or call routing issue doesn't cascade into another's. That per-tenant carrier isolation is one of the concrete mechanics that separates a real multi-tenant platform from a thin reseller wrapper sitting on top of someone else's account, a distinction covered in white-label vs reseller.
How does branding work across sub-accounts?
The agency's brand sits at the top: its logo, its domain, its name on every client-facing screen and confirmation text. Sub-accounts inherit that branding automatically, so a client logging into their portal sees the agency's name, not the underlying platform's. The platform vendor's name doesn't need to appear anywhere the client can see it, which is the whole point of buying white-label instead of pointing clients at a generic tool with the agency's logo pasted over the top.
How does pricing work at the sub-account level?
Platforms structure sub-account pricing a few different ways, commonly a flat fee per active sub-account, a usage-based rate on top of a base platform fee, or tiers that unlock more sub-accounts as an agency grows. The right structure depends on how many clients an agency plans to run and how thin the margins are on each one, which is worth modeling against actual client pricing before committing, something we walk through in how to price white-label voice AI to SMB clients without eating margin. Actual per-sub-account and per-minute numbers are on agency pricing rather than repeated here, since they change and shouldn't be quoted secondhand.
Does every white-label AI voice platform structure sub-accounts the same way?
No, and this is worth checking before signing anything. Some platforms genuinely isolate each client's data and carrier setup, which is the model above. Others run a lighter version where "sub-accounts" are really just tagged records inside one shared database and one shared carrier trunk, which can work fine at small scale but creates real risk once an agency is running a dozen clients on the same underlying number pool. Because vendors describe this differently, it's worth reading how competitors define their own account structure directly on their pages rather than taking a reseller's word for it, for example how Vapi or Retell each document their multi-tenant setup.
What's a real edge case agencies run into?
A client asks to see their own dashboard. If sub-accounts are properly isolated, granting that client login access is a scoped permission, they see their number, their calls, their calendar, nothing else. If the platform wasn't built with real sub-account boundaries, giving a client any visibility at all risks exposing another client's data through a shared view or a poorly filtered report. This is one of the first things worth testing in a demo before onboarding a paying client, not after.
What should an agency check before committing to a platform's sub-account model?
Ask directly: is each sub-account's phone number on its own carrier trunk, or shared? Can a client's data be exported or deleted independently of every other client, which matters for churn and offboarding? Does branding actually propagate to every client-facing surface, or only the login screen? These are the same questions worth running through a structured checklist rather than trial and error, which is what the agency readiness scorecard is built to walk through before an agency locks into a vendor.
Twilio's own subaccounts documentation describes this same pattern at the carrier level, letting a platform isolate each customer's resources, usage, and credentials under one parent account, which is the same isolation model a white-label voice platform mirrors for each agency's clients.