Build, partner, or refer? The economics
When a client asks for something you don't staff, you have exactly three options: build the capability, partner for it, or refer the work away. The right answer depends on how often the demand will recur, not on how the current brief feels. A one-off request argues for partnering or referring; a service your clients will keep asking for argues for building or for a standing partnership that behaves like capacity you own.
The cost shapes differ more than the sticker prices do. Hiring converts revenue you hope for into salary you owe regardless: recruitment lead time, ramp-up, management load, and a fixed cost that survives the pipeline dip that eventually follows every rush. Freelancers price flexibly but arrive as individuals: no bench behind them when they're ill, booked or gone, and no team a client can be shown. Referring costs the least and loses the most: the revenue, the expanded relationship, and eventually the client's sense that you are the agency that handles things. White-label partnering sits between: wholesale rates, your price on top, and delivery capacity that scales with the work you actually win.
Four routes for work you can't staff, compared
| Route | Cost shape | Speed to revenue | The risk you carry |
|---|---|---|---|
| Hire in-house | Fixed: salaries survive quiet quarters | Slow: recruit, ramp, then sell the capability | Utilisation. Empty bench time is unrecoverable margin. |
| Freelancers | Variable, per project | Fast when the right person is free | Continuity and cover: individuals, not a bench, and quality that varies per hire. |
| White-label partner | Variable: wholesale rate card, your margin on top | Fast once vetted; near-immediate with paperwork pre-done | Partner quality and client confidentiality: the vetting and contract terms in this guide. |
| Refer it away | Zero cost, zero revenue | N/A | The relationship. Referred clients learn another supplier's name, and scope drifts with them. |
In practice the decision splits by pattern. Overflow on services you already sell is a capacity problem: the arithmetic of pods versus hires is worked through on capacity and delivery. A client asking for a service you've never offered is an expansion decision, whether to say yes before you can staff it, covered from the business side on expanding your service line. Either way, the margin mechanics are the same: the partner's rate card is wholesale, you set the client price, and the difference is yours; how partner pricing is typically structured is published rather than negotiated per brief.
How do you vet a delivery partner?
Vet a partner the way you'd hire a senior employee: on evidence, not on the sales deck. Every white-label provider's website says 'senior team', 'your brand', 'frictionless process'. The deck cannot distinguish a deep bench from a broker who resells your brief to whoever is cheapest this month. The questions below are the ones that expose the difference, because a weak bench cannot fake the answers for long.
“Can we meet the people who'd do the work?”
Not the sales lead. The strategist and project manager who would actually run your account, on a video call, before any brief exists. A partner with a real bench arranges this readily; a broker stalls, because there is nobody stable to introduce. A named pod you can meet is the single strongest signal in the entire vetting process.
“Does the rate card hold?”
Ask for wholesale pricing upfront and whether it is honoured across engagements. Re-quoting every brief from scratch makes your own client pricing impossible to plan, and often signals that the work is being shopped to subcontractors each time, with your quality riding on whoever answered fastest.
“What happens when work comes back wrong?”
Ask who reviews work before you see it, what the revision policy says, and what happens to the relationship when a deliverable misses. A partner with real QA describes a process; a weak bench describes goodwill. You are buying their worst week, not their best one. Price accordingly.
“Could you ever serve our clients directly?”
Ask directly, and ask for the answer in contract form: a zero-competition clause, and a policy for what happens if your client approaches them. Hesitation here is disqualifying: the entire arrangement rests on your relationships staying yours. The full terms belong in chapter three.
“Show us anonymised work, with numbers.”
White-label NDAs mean a credible partner cannot name clients, but they can show sector, engagement shape and real figures, anonymised. Insist on it. Our own published case studies take exactly this form, including a delivery rescue run invisibly under an agency's brand: 85% faster delivery, 0 clients lost. No numbers at all, even anonymised, means no record worth trusting.
Then run a pilot before you need one. A small, representative, paid brief (internal if you have nothing client-safe) tells you more than every reference call combined: how they scope, how they communicate mid-project, what arrives at handover, and how they take a revision note. Vetting a partner while a live client waits is how agencies end up marrying the first deck they saw.
How do you protect the client relationship?
Protection is contractual or it is imaginary. The client relationship is an agency's entire balance sheet, so the promises that guard it must live in signed documents, not in rapport with a friendly sales lead who may not be there next year. Three terms do the real work, and a serious partner offers them unprompted.
First, an NDA before anything else: before the client's name, the brief or a single credential changes hands, covering your client list and your commercial terms, and surviving the engagement. Second, a zero-competition clause: a contractual commitment that the partner will never approach, pitch or accept your clients directly, paired with a disclosure duty. If your client ever contacts them, they tell you and decline the work. Third, a blackout period extending that protection beyond the engagement itself, so ending the partnership doesn't open a window on your client list. These are the commitments our own white-label partnership publishes: NDA first, a zero-competition guarantee in writing, and a 12-month post-engagement blackout. They are a reasonable baseline to demand from anyone in this market.
Contract terms guard against betrayal; operational discipline guards against the likelier failure: accidental exposure. The partner's name appears nowhere a client looks: reports carry your logo and templates, emails come from your domain, file metadata and meeting invitations are checked before anything ships. Decide the cover story once and hold it: most agencies introduce partner staff as 'our delivery team' on client calls, true in every way the client cares about, and the strongest sign of a practised partner is that they raise the cover protocol before you do. Where the seams show is where white-labelling fails, and it is almost never in the contract.
How to tell real protection from the language of it
A partner who actually protects you
- Raises the NDA before you do, and signs one that covers your client list and your rates, not just their own confidentiality.
- Offers the zero-competition clause unprompted, in the contract, with a disclosure duty attached: if your client contacts them, they tell you and decline.
- Names a blackout period and its length without being asked: ours is 12 months post-engagement.
- Checks file metadata, email domains and calendar invites as routine QA, because that is where cover actually breaks.
- Has a cover protocol already written down, and raises it in the first conversation rather than the first crisis.
What weak protection sounds like
- “We’d never do that” is a values statement where a clause should be. Values do not survive a change of sales lead.
- A non-compete that binds you and not them, or one that expires the day the engagement does.
- Willing to say it on a call, reluctant to put it in the contract. What a partner resists writing down is what it is worth.
- No answer for what happens if your client approaches them directly. That's the exact scenario the clause exists for.
- Cover treated as your job: their logo in the report footer, their domain in the email thread, and an apology afterwards.
None of the five items on the left is expensive for an honest partner to offer, which is precisely why the refusals are informative. A provider who will not put protection in writing is not managing risk. They are keeping an option open.
How does operational integration actually work?
The partner adapts to your operation, never the reverse. That's the test of every integration decision. Work happens in your Slack or Teams, your Asana, Jira or Trello; briefs, statuses and handovers live where your team already looks. An agency forced to check a supplier's portal for its own client work has quietly become the supplier's customer instead of its principal. How a partner runs cadence, tooling and handoffs day to day should be documented and inspectable (ours is published at how we work) because a partner who cannot describe their operating rhythm in writing does not have one.
Cadence is the second pillar: a weekly rhythm with a named project manager, work sized before it is promised, and statuses that arrive before you ask. Velocity that spikes and collapses is worse than velocity that is merely steady, because your client-facing schedule is built on it. The third pillar is QA upstream of the handover: everything reviewed and tested inside the partner's pod before your producers ever see it. If your team is the quality gate, you have bought labour, not delivery; rework arriving at your desk is where white-label margins quietly die.
Integration also differs by discipline, and a partner should show you the shape for each. Reselling search means reporting your clients can present as their own: the deliverable cadence and rank-tracking handoff described on white-label SEO. Reselling creative means working files, not flattened exports: source files, design systems and revision rounds as laid out on white-label design. Whatever the discipline, the finished work must be indistinguishable from work you'd have produced in-house, in your templates and your tone; sampling that regularly is your job, and a good partner makes it easy.
What good integration looks like, by discipline
| Discipline | What has to arrive | What to sample, and how often |
|---|---|---|
| Search | Reports in your template and your tone, rank tracking you can log into yourself, and the reasoning behind each recommendation, not only the recommendation. | One client report a month, read as if a client sent it to you. If you would have rewritten it before forwarding, integration is not finished. |
| Design | Working files, not flattened exports: source files, layered components, the design system and its tokens, plus the agreed revision rounds. | Open one delivered file a month and try to make a small change without asking anyone. If you cannot, you bought pictures of design, not design. |
| Development | Code in your repository under your account, a readable commit history, and an environment your own developer could pick up cold. | Every quarter, ask an in-house or trusted developer to review a recent commit range. Handover risk is invisible until the day it is urgent. |
| Dedicated pods | Named people who stay, working in your tools on your cadence, with context that accumulates instead of resetting each brief. | Watch for silent substitutions. A pod whose names change without notice is a contractor pool with better branding. |
The pattern under all four rows is the same: what arrives should be something you could take over tomorrow. Deliverables that only the partner can maintain are not integration. They are dependency, and the bill for it comes due at the worst possible moment.
How do you scale from first brief to a standing bench?
Set the partnership up before you need it. That's the single highest-leverage move in this guide. Vetting, meeting the pod, agreeing the rate card and putting the NDA on file all take days you will not have once a client has said yes. Agencies that do this while things are calm hold a partner 'on the bench': paperwork done, pricing known, pod met, nothing owed. Then the first brief moves at the speed of the work rather than the paperwork: on our bench, scoping starts the day a brief lands, a priced proposal follows within 48 hrs, and the first deliverable typically ships within 14 days of the NDA. The bench model, and what joining it involves, is laid out on the agencies page.
Scale by pattern, not by leap. The first brief should be small and representative. Treat it as the paid extension of vetting. If it lands well, widen gradually: more briefs, then a second discipline, then pre-sales help on pitches you haven't won yet, with the same named pod building context on your clients, your tone and your tools instead of resetting every engagement. When volume steadies into predictable monthly work, per-brief pricing stops being the efficient shape and reserved capacity starts: a dedicated team whose economics reward the commitment you can now forecast honestly.
The end state is a delivery arm that behaves like your own department without the fixed cost of one: sized to this quarter's pipeline, carrying your brand on everything it touches, invisible to every client. That end state is common, not theoretical: 100+ agencies deliver through our bench and the average partnership runs beyond two years. But every one of them started with the same unglamorous steps this guide describes: an economics decision, a vetting call, a contract worth signing and a small first brief. The bench is built before the brief. Start there.
Stage 0: Build the bench before you need it
Vetting, meeting the pod, agreeing the rate card, signing the NDA. Nothing is owed and nothing is spent, but the paperwork that would otherwise cost you a fortnight is done. This is the stage almost every agency skips and every agency wishes it had not.
Stage 1: One small, representative brief
Paid, real, and deliberately modest. Treat it as the last stage of vetting rather than the first stage of delivery: you are buying information about how they scope, communicate and take a revision note, and the fee is cheap for what it tells you.
Stage 2: Widen, one axis at a time
More briefs in the same discipline, then a second discipline, then pre-sales support on pitches you have not won yet. Keep the same named pod so context accumulates. Widening two axes at once means a problem gives you two suspects and no answer.
Stage 3: Reserve capacity when demand is forecastable
Once monthly volume is steady enough to predict honestly, per-brief pricing stops being the efficient shape and a dedicated team starts. Make this move on your forecast, not on a partner's suggestion. The commitment should follow the evidence.
The stages are cumulative and each one is cheap to stop at. That is the point: an agency that has done Stage 0 has bought a genuine option and paid almost nothing for it, while an agency that starts at Stage 1 with a client already waiting has bought whatever it can find that week.