White-label data collection means your client receives the data under your brand, on your terms, with your name on the invoice. They do not contract with the infrastructure underneath, they do not see it, and in most cases they do not want to.
That arrangement is straightforward commercially and it changes your obligations more than most agencies expect. The moment you put your brand on a dataset, you stop being a company that uses a proxy network and become the data provider of record. Reliability, provenance, compliance and continuity all become yours to answer for.
This guide covers what that means in practice and how to structure one account to carry many clients without them ever touching each other.
”One account” means one login, not one pot
Worth settling immediately, because the phrase is ambiguous and the wrong reading causes real damage.
What works: one Shifter account for your agency, one login, one vendor relationship, one place to see everything, with each client separated underneath it into its own workspace and its own plan.
What does not work: every client’s collection running through a single shared plan. That version cannot tell you what each client cost, lets one client’s spike consume another’s allowance, and makes offboarding a manual disentangling exercise.
The mechanics of the first arrangement, workspaces, roles, the sidebar switcher and per-plan usage, are covered in sub-accounts and usage tracking for multiple clients. The short version is one plan per client, always, and a workspace per client wherever billing or client access justifies it.
What you take on when your brand goes on the data
Five things move onto your side of the line.
Reliability. Your client’s report is due whether or not a portal changed its markup last night. You need monitoring, alerting and a fallback plan, because “our supplier had an issue” is not a sentence a white-label provider gets to use.
Provenance. When a client disputes a number, you have to be able to say where and when it was observed. That means every deliverable carries the market, the vantage point, the timestamp and the collection method. The case for treating that as a first-class field is in the vantage-point standard.
Compliance. Your client’s obligations do not stop at your brand. If they are a data controller and you are processing on their behalf, the responsibilities flow to you.
Continuity. If you lose a source, a market or a contract, the client experiences that as your outage.
Support. You are first line. Your client asks you why the number moved, and the answer cannot be a forwarded ticket.
None of this is a reason not to white-label. It is a reason to price it properly, which is the last section.
Compliance does not disappear behind your brand
This is the part most often missed, and the one most likely to surface in a client’s procurement review rather than in your own planning.
If your client is a controller and you are a processor, your data processing agreement will usually require you to disclose subprocessors and to give notice before changing them. Infrastructure you rely on to collect the data can fall within that definition. White-label branding is a commercial arrangement, not a legal shield: it does not remove a disclosure obligation your contract creates.
Two practical consequences. Read your client DPAs before signing them to check what they say about subprocessors, notice periods and audit rights, because those clauses determine how freely you can change your own stack. And keep a current internal record of what your collection depends on, so that when a client asks, the answer takes minutes rather than a week.
Related, and frequently conflated with it: where a request exits is not where your data is processed or stored. A client asking “is our data in the EU” is asking about your pipeline, not about your exit locations. That distinction is set out in data residency is not proxy location, and it is worth understanding before a client asks rather than during the call.
Structuring the account
The operational setup is short, and following it exactly is what makes everything downstream easy.
One plan per client. No exceptions. Usage is reported per plan, so a shared plan is a cost you can never attribute afterwards.
A workspace per client where it earns its place. Clients who want their own billing relationship, want visibility into their own usage, or are large enough that their overage should not draw on a shared wallet. Remember that everyone in a workspace sees everything in it, so a client contact only ever belongs in their own workspace.
Credentials separated per client. Each plan has its own. Store them per client in your secret manager, give each client’s jobs only that client’s credentials, and never borrow one client’s plan to finish another’s job quickly. That single shortcut is how attribution silently breaks.
Session IDs prefixed by client. The panel does not break usage down by session identifier, but your own logs can. sid-acme-listings-04 costs nothing and makes your own debugging readable.
Vantage points recorded per deliverable. Not just for disputes. When two clients in the same category ask why their numbers differ, the vantage point is usually the answer.
With the Shifter gateway, the client prefix and the market both live in the credentials against p.shifter.io:443:
customer-USERNAME-country-de-city-berlin-sid-acme-de-07-ttl-600:PASSWORDOmit sid for independent requests and the gateway rotates per request; include it to hold one exit across a multi-step job, with ttl setting how long. The trade-off is in sticky vs rotating residential proxies, and request pacing in rate limiting and request throttling.
Price the deliverable, not the gigabyte
The most common white-label pricing mistake is reselling bandwidth at cost plus a margin, to clients who are not buying bandwidth. Your client is buying a report, a feed or a dashboard. They have no way to evaluate a per-gigabyte price and no interest in learning.
Price per deliverable, and manage your own margin on the metric that actually drives cost: gigabytes per usable record. A workflow that transfers less but returns blocked or geographically wrong data is more expensive per delivered record than a heavier one that works. The forecasting method is in forecasting residential proxy bandwidth.
Three habits protect the margin. Review consumption per client weekly rather than at invoice time. Treat a scope change as a scope change, since clients routinely add markets or frequency in an email without expecting a price conversation. And decide in advance, per client, whether overage should flow or the job should pause.
Continuity and offboarding
Plan both before you need them.
For continuity: monitor collection success per client per source, so a degrading source is visible before the client notices. Know which of your deliverables depend on a single source, because those are the ones that break hardest.
For offboarding: removing a client should be a checklist, not an investigation. Retire that client’s credentials from every job, close or transfer their plan, remove their people from the workspace, and agree in the contract what happens to historical data. If a client owns their workspace, this is largely their action, which is a further argument for that structure with larger accounts.
FAQ
Can I run several clients on one plan if they are small?
You can, and you will not be able to say what any of them cost or contain a spike. If the volumes are genuinely tiny, the saving is small and the attribution loss is permanent.
Do I have to tell clients which infrastructure I use?
That depends on your contract rather than on your branding. Many client DPAs require subprocessor disclosure and notice of changes. Check the agreements you have signed.
How many clients can one agency account carry?
One login can belong to many workspaces, switching from the sidebar. The seat limit applies per workspace, not to the number of workspaces you work across.
What should appear in a white-label deliverable?
The data, plus enough provenance to defend it: market, vantage point, observation timestamp and method. Your brand goes on the cover. The method still has to exist underneath it.
The bottom line
White-labelling is a promise that the data is yours to stand behind. Structure the account so that promise is cheap to keep: one agency login across many workspaces, one plan per client without exception, credentials and session identifiers separated per client, and provenance attached to every deliverable.
Then price the deliverable rather than the gigabyte, watch usage per client weekly, and read your client DPAs before your stack changes rather than after. The network itself is on the residential proxy network page, with rates on the pricing page.