I am a...
Learn more
How it worksPricingFAQ
Account
May 19, 2026 · 10 min read · Cadence Editorial

White-label development services: how to sell what you don't build

white label development services — White-label development services: how to sell what you don't build
Photo by [Mikael Blomkvist](https://www.pexels.com/@mikael-blomkvist) on [Pexels](https://www.pexels.com/photo/people-sitting-at-the-table-6476257/)

White-label development services: how to sell what you don't build

White-label development services let an agency sell engineering work, under its own brand, that's actually executed by a partner shop or a booked external team. The typical markup is 50% to 70% over partner cost, with the agency owning the client relationship, the spec, and the delivery experience. Done well, it's the fastest path to scale revenue without scaling headcount.

Done badly, it's a margin trap that quietly destroys client trust the first time the partner ships late.

This post is the operating playbook: how to vet partners, structure margin, manage the comms firewall, write the contract clauses that protect you, and decide between three white-label patterns (subcontracting, reseller, and referral). We'll also cover where Cadence fits as a partner option (10% recurring referral, or booking engineers under your own brand at agency markup).

Why agencies white-label in the first place

Most dev shops hit a ceiling around 12 engineers. Past that, founders are 80% management, hiring takes 8 to 12 weeks per role, and utilization swings 15 to 30 points when a single retainer churns.

White-labeling solves three problems at once:

  1. Pipeline outruns capacity. You won the logo. You don't have the React Native engineer. You can't tell the client "give us 90 days to hire."
  2. Spiky demand. A 6-week scope drop doesn't justify a full-time hire. Booking a partner engineer for the spike preserves margin.
  3. Stack gaps. You're a Ruby shop. The client wants a Swift app. Subcontracting beats turning the work away.

The unspoken fourth reason: hiring is risky and slow, and an agency owner who has fired three bad hires would rather pay a 30% premium for a vetted external engineer than re-run the hiring loop.

The three white-label models (pick one, don't blend them)

There's a real difference between subcontracting, reselling, and referring. Most agencies confuse them and end up with contracts that don't actually protect the relationship they're running.

ModelWho owns the clientWho deliversTypical marginContract complexityBest when
SubcontractYouPartner engineer under NDA40% to 60%Medium (1099 + IP assignment)Stack gap, single project
Reseller / white-labelYouPartner shop, fully cloaked50% to 70%High (MSA + no-poach + SLA)Predictable recurring work
ReferralPartnerPartner10% to 20% recurringLow (1-page referral agreement)Outside your ICP
Cadence partner programYou (booked through Cadence)Cadence engineer at $500 to $2k/wk, billed to your client at your markup50% to 200%+Low (Cadence handles IP + payroll)Spiky work, AI-native stack
Cadence referralCadenceCadence10% recurring on every founder you referZero (link tracking)Founder isn't your ICP

Three rules that follow from the table:

  • Subcontracting is fine for one-offs but doesn't scale. Every project is a new partner search.
  • Reselling is the real white-label pattern. It works when you have a primary partner and predictable volume.
  • Referral is the cleanest economically (no delivery risk) but caps your margin and gives up the client relationship.

The Cadence partner angle is a hybrid: book the engineer, mark it up, ship under your brand, and the partner program also pays 10% recurring if you just want to refer founders who aren't your ICP.

Vetting white-label partners: the 7-point checklist

Most agencies pick partners on price and vibes. Both are wrong. Vet on these seven things in this order:

  1. Reference calls with two of their existing reseller clients. Not testimonials. Live calls. Ask "what's gone wrong, and how fast did they fix it."
  2. A live trial project under $5k. Pay full freight. Watch the comms, the spec interpretation, the merge cadence. This is non-negotiable.
  3. Time-zone overlap with your client base. 4 hours of overlap is the minimum for daily standups. Less and you'll be the comms relay forever.
  4. A demonstrated AI-native workflow. Cursor or Claude Code in daily use. Prompt-as-spec discipline. Without this, your partner is 2x to 3x slower than the market in 2026 and you'll feel it in margin.
  5. Documented engineer churn rate. Partners that lose 40%+ of engineers annually will swap your lead engineer mid-project. Ask. Get the number.
  6. A real escalation path. Named human, not a Slack channel. Someone with authority to refund or swap engineers without a meeting.
  7. Clean IP and contractor classification. Every engineer signs a work-for-hire assignment. Partner provides W-9 or W-8BEN. No exceptions.

Skip steps 1 and 2 and you'll learn the hard way. The trial project is the highest-signal $5k you'll spend this year.

If you want a faster path on step 4 (AI-native verification), every engineer on Cadence is AI-native by default, vetted on Cursor / Claude / Copilot fluency in a voice interview before they unlock bookings. That removes the audit step entirely for the booking pattern.

The margin math: what white-label actually pays

Let's run real numbers. Assume you're a US agency billing clients $200/hour blended rate.

Subcontract pattern (mid-level partner engineer):

  • Partner cost: $50/hour (this is roughly what $1,000/week works out to at 40 hours)
  • Your bill rate: $200/hour
  • Gross margin: 75%
  • After PM + ops overhead (call it 25 points): 50% net

Reseller pattern (full team from partner shop):

  • Partner cost: $80/hour (they include light PM)
  • Your bill rate: $200/hour
  • Gross margin: 60%
  • After ops: 40% net

Cadence booking pattern (senior engineer at $1,500/week):

  • Cadence cost: $37.50/hour ($1,500 / 40)
  • Your bill rate: $200/hour
  • Gross margin: 81%
  • After ops: 55% net

Pure referral:

  • Cost: $0
  • Revenue: 10% to 20% of partner billings for the life of the engagement
  • Margin: 100% (with zero delivery risk)

The Cadence pattern wins on raw margin because weekly billing collapses the partner overhead. The reseller pattern wins on stability because the partner shop carries the PM layer. Pick based on whether you're optimizing for margin or for not-being-on-pager.

For more on rate setting, dev agency hourly rate benchmarks breaks down what each region charges by tier so you can calibrate the markup honestly.

The contract clauses that actually matter

A standard MSA isn't enough for white-label work. Five clauses are non-negotiable and most templated contracts miss at least two of them.

1. Mutual no-direct-poach (24 to 36 months)

Both directions. The partner cannot solicit your client and you cannot solicit the partner's engineers. Liquidated damages, typically 12 months of expected fees. Without this, your best client meets your partner's BD lead at a conference and you lose both.

2. IP assignment on delivery, not on payment

Work product transfers when it's delivered, not when you pay. This protects you if a payment dispute happens mid-project; the IP doesn't get held hostage. Pair with milestone payments so neither side is holding the work hostage.

3. Cloaking and brand-protection clauses

The partner's engineers operate under your domain, your Slack, your Linear or Jira. No partner branding on commits, on PRs, on emails to the client. If you skip this, the cloak leaks and the client will Google the partner shop within a week.

4. SLAs with real teeth

Response time (4 business hours), bug-fix turnaround (24h for P0, 72h for P1), and engineer-swap timing (5 business days if the relationship breaks). With financial penalties (typically a 10% to 20% credit). An SLA without penalties is a wish list.

5. Termination for convenience with a 30-day notice

Either side, no penalty, no fault required. White-label relationships go bad slowly. You need a clean exit that doesn't require a lawyer to negotiate. Weekly billing arrangements (like Cadence's) collapse this clause to "cancel any week," which is often cleaner.

For a deeper template walkthrough, see dev agency contract templates and gotchas.

The communication firewall (the part everyone gets wrong)

The single biggest white-label failure mode: the partner engineer accidentally Cc's the client on an internal thread, or signs off "Best, Priya from RemoteShop" instead of "Best, Priya from YourAgency." Once the cloak slips, the client renegotiates or leaves.

Build the firewall on day one:

  • Issue your-domain email addresses to every partner engineer. priya@youragency.com, forwarding to their real address. Cost: $6/mo per seat on Google Workspace.
  • Single Slack workspace, yours. Partner engineers join your workspace as multi-channel guests. They don't have a parallel partner Slack with your client in it.
  • All client comms route through one PM on your side. Engineers ship code, the PM ships words. This is non-negotiable for the first 90 days; you can relax it for tenured engineers later.
  • Standardize the spec format. Every ticket gets the same 6-field structure: goal, acceptance criteria, files touched, test plan, dependencies, deploy notes. Vetted partners can ship in 24 to 48 hours from a clean spec; murky specs are how 2-week jobs turn into 6.
  • Daily stand-up async in writing. No video calls with the client unless your PM is on the line. Your client never meets the partner engineer 1:1.

If you're juggling several reseller engagements at once, the comms firewall is the thing that breaks first. Managing multiple client projects as a dev agency covers the allocation and PM-rotation patterns that prevent comms leaks at scale.

What to do this week

If you're at the agency-scaling stage and considering a white-label model:

  1. Pick the model (subcontract, reseller, or referral) based on whether your demand is one-off, recurring, or out-of-ICP.
  2. Run a $5k trial with two partners in parallel. Cheaper than a hiring mistake.
  3. Draft the 5 contract clauses above before signing anything bigger.
  4. Set the comms firewall on day one. Retrofitting it after the cloak slips is impossible.
  5. If you want a partner option with weekly billing and no minimum commit, book a senior engineer through Cadence for the trial. The 48-hour free trial means the partner test costs you nothing if the engineer isn't right.

The agencies that scale past $2M ARR almost always run a mix: a core team they hired, a primary white-label partner for predictable overflow, and a booking channel (like Cadence) for spiky or AI-heavy work. Single-channel agencies cap earlier and feel more fragile.

The Cadence partner program in 30 seconds

Two structural wins for agencies running on Cadence:

  • Book engineers at $500 to $2k/week, bill your client at your markup. Every engineer is AI-native by default, voice-interviewed on Cursor / Claude Code / Copilot fluency. Daily ratings drive auto-replacement, so the swap clause is built into the platform. White-label cloak is yours to set up; we don't email your clients.
  • Earn 10% recurring on referrals. If a founder isn't your ICP, refer them via your partner link and you earn 10% of their weekly spend for as long as they book on Cadence. No cap, no delivery risk.

If you want to test the booking pattern before signing partner contracts, join the Cadence partner program and book one engineer this week at agency markup. It's the lowest-risk way to see whether white-label is actually a fit before you commit to a primary partner.

For agencies still wondering whether the model is for them, how to transition from freelancer to agency owner covers the operating-mode shift; white-label is one of the four scaling levers it discusses.

FAQ

What's a typical white-label development markup?

50% to 70% is normal in 2026. Reseller patterns sit at the lower end (the partner shop is doing more of the PM work); pure subcontract patterns sit higher because you carry more of the delivery overhead yourself. Booking platforms like Cadence can support markups of 100% or more because weekly pricing collapses partner overhead.

Is white-labeling ethical if the client doesn't know?

Most US client contracts allow subcontracting by default and the question rarely comes up. The honest answer: a confidentiality clause in your MSA that says "we may use approved subcontractors who sign equivalent NDAs" is standard. If a client explicitly asks "are you using subcontractors," tell them. Lying about it is the only ethical failure here.

How do I prevent my white-label partner from poaching my client?

Three layers: a mutual no-direct-poach clause in the MSA (24 to 36 months), liquidated damages tied to the value of the engagement, and operational separation (the partner engineers operate under your email, your Slack, your tools, so they have no direct channel to your client's CFO). Cadence engineers are contractually barred from soliciting agency clients booked through the partner program.

How fast can a white-label partner ramp on a new project?

A vetted partner with a clean spec can ship a first commit within 24 to 48 hours. Cadence's median time to first commit across the platform is 27 hours; that's the bar to benchmark against. If your partner needs more than 5 business days to ship anything, your spec is unclear or the partner isn't vetted enough.

What's the difference between white-label and subcontracting?

Subcontracting is the legal mechanism: one party delivers work for another under contract. White-label is the brand pattern: the end client sees only your brand, regardless of who's doing the work. All white-label work involves subcontracting, but not all subcontracting is white-label (some is openly co-branded).

Should I use a white-label partner or just hire?

Hire for work you have 12+ months of committed pipeline for. White-label everything else. Hiring a full-time engineer for a 6-week project costs you $30k to $80k loaded in the first quarter alone; booking the same engineer for 6 weeks costs $6k to $12k. The break-even is roughly 4 to 6 months of continuous utilization.

All posts